1 <?xml version="1.0" encoding="iso-8859-1"?>
3 <indexterm><primary>language, GHC</primary></indexterm>
4 <indexterm><primary>extensions, GHC</primary></indexterm>
5 As with all known Haskell systems, GHC implements some extensions to
6 the language. They are all enabled by options; by default GHC
7 understands only plain Haskell 98.
11 Some of the Glasgow extensions serve to give you access to the
12 underlying facilities with which we implement Haskell. Thus, you can
13 get at the Raw Iron, if you are willing to write some non-portable
14 code at a more primitive level. You need not be “stuck”
15 on performance because of the implementation costs of Haskell's
16 “high-level” features—you can always code
17 “under” them. In an extreme case, you can write all your
18 time-critical code in C, and then just glue it together with Haskell!
22 Before you get too carried away working at the lowest level (e.g.,
23 sloshing <literal>MutableByteArray#</literal>s around your
24 program), you may wish to check if there are libraries that provide a
25 “Haskellised veneer” over the features you want. The
26 separate <ulink url="../libraries/index.html">libraries
27 documentation</ulink> describes all the libraries that come with GHC.
30 <!-- LANGUAGE OPTIONS -->
31 <sect1 id="options-language">
32 <title>Language options</title>
34 <indexterm><primary>language</primary><secondary>option</secondary>
36 <indexterm><primary>options</primary><secondary>language</secondary>
38 <indexterm><primary>extensions</primary><secondary>options controlling</secondary>
41 <para>The language option flags control what variation of the language are
42 permitted. Leaving out all of them gives you standard Haskell
45 <para>Language options can be controlled in two ways:
47 <listitem><para>Every language option can switched on by a command-line flag "<option>-X...</option>"
48 (e.g. <option>-XTemplateHaskell</option>), and switched off by the flag "<option>-XNo...</option>";
49 (e.g. <option>-XNoTemplateHaskell</option>).</para></listitem>
51 Language options recognised by Cabal can also be enabled using the <literal>LANGUAGE</literal> pragma,
52 thus <literal>{-# LANGUAGE TemplateHaskell #-}</literal> (see <xref linkend="language-pragma"/>). </para>
54 </itemizedlist></para>
56 <para>The flag <option>-fglasgow-exts</option>
57 <indexterm><primary><option>-fglasgow-exts</option></primary></indexterm>
58 is equivalent to enabling the following extensions:
59 <option>-XPrintExplicitForalls</option>,
60 <option>-XForeignFunctionInterface</option>,
61 <option>-XUnliftedFFITypes</option>,
62 <option>-XGADTs</option>,
63 <option>-XImplicitParams</option>,
64 <option>-XScopedTypeVariables</option>,
65 <option>-XUnboxedTuples</option>,
66 <option>-XTypeSynonymInstances</option>,
67 <option>-XStandaloneDeriving</option>,
68 <option>-XDeriveDataTypeable</option>,
69 <option>-XFlexibleContexts</option>,
70 <option>-XFlexibleInstances</option>,
71 <option>-XConstrainedClassMethods</option>,
72 <option>-XMultiParamTypeClasses</option>,
73 <option>-XFunctionalDependencies</option>,
74 <option>-XMagicHash</option>,
75 <option>-XPolymorphicComponents</option>,
76 <option>-XExistentialQuantification</option>,
77 <option>-XUnicodeSyntax</option>,
78 <option>-XPostfixOperators</option>,
79 <option>-XPatternGuards</option>,
80 <option>-XLiberalTypeSynonyms</option>,
81 <option>-XRankNTypes</option>,
82 <option>-XImpredicativeTypes</option>,
83 <option>-XTypeOperators</option>,
84 <option>-XRecursiveDo</option>,
85 <option>-XParallelListComp</option>,
86 <option>-XEmptyDataDecls</option>,
87 <option>-XKindSignatures</option>,
88 <option>-XGeneralizedNewtypeDeriving</option>,
89 <option>-XTypeFamilies</option>.
90 Enabling these options is the <emphasis>only</emphasis>
91 effect of <option>-fglasgow-exts</option>.
92 We are trying to move away from this portmanteau flag,
93 and towards enabling features individually.</para>
97 <!-- UNBOXED TYPES AND PRIMITIVE OPERATIONS -->
98 <sect1 id="primitives">
99 <title>Unboxed types and primitive operations</title>
101 <para>GHC is built on a raft of primitive data types and operations;
102 "primitive" in the sense that they cannot be defined in Haskell itself.
103 While you really can use this stuff to write fast code,
104 we generally find it a lot less painful, and more satisfying in the
105 long run, to use higher-level language features and libraries. With
106 any luck, the code you write will be optimised to the efficient
107 unboxed version in any case. And if it isn't, we'd like to know
110 <para>All these primitive data types and operations are exported by the
111 library <literal>GHC.Prim</literal>, for which there is
112 <ulink url="../libraries/base/GHC.Prim.html">detailed online documentation</ulink>.
113 (This documentation is generated from the file <filename>compiler/prelude/primops.txt.pp</filename>.)
116 If you want to mention any of the primitive data types or operations in your
117 program, you must first import <literal>GHC.Prim</literal> to bring them
118 into scope. Many of them have names ending in "#", and to mention such
119 names you need the <option>-XMagicHash</option> extension (<xref linkend="magic-hash"/>).
122 <para>The primops make extensive use of <link linkend="glasgow-unboxed">unboxed types</link>
123 and <link linkend="unboxed-tuples">unboxed tuples</link>, which
124 we briefly summarise here. </para>
126 <sect2 id="glasgow-unboxed">
131 <indexterm><primary>Unboxed types (Glasgow extension)</primary></indexterm>
134 <para>Most types in GHC are <firstterm>boxed</firstterm>, which means
135 that values of that type are represented by a pointer to a heap
136 object. The representation of a Haskell <literal>Int</literal>, for
137 example, is a two-word heap object. An <firstterm>unboxed</firstterm>
138 type, however, is represented by the value itself, no pointers or heap
139 allocation are involved.
143 Unboxed types correspond to the “raw machine” types you
144 would use in C: <literal>Int#</literal> (long int),
145 <literal>Double#</literal> (double), <literal>Addr#</literal>
146 (void *), etc. The <emphasis>primitive operations</emphasis>
147 (PrimOps) on these types are what you might expect; e.g.,
148 <literal>(+#)</literal> is addition on
149 <literal>Int#</literal>s, and is the machine-addition that we all
150 know and love—usually one instruction.
154 Primitive (unboxed) types cannot be defined in Haskell, and are
155 therefore built into the language and compiler. Primitive types are
156 always unlifted; that is, a value of a primitive type cannot be
157 bottom. We use the convention (but it is only a convention)
158 that primitive types, values, and
159 operations have a <literal>#</literal> suffix (see <xref linkend="magic-hash"/>).
160 For some primitive types we have special syntax for literals, also
161 described in the <link linkend="magic-hash">same section</link>.
165 Primitive values are often represented by a simple bit-pattern, such
166 as <literal>Int#</literal>, <literal>Float#</literal>,
167 <literal>Double#</literal>. But this is not necessarily the case:
168 a primitive value might be represented by a pointer to a
169 heap-allocated object. Examples include
170 <literal>Array#</literal>, the type of primitive arrays. A
171 primitive array is heap-allocated because it is too big a value to fit
172 in a register, and would be too expensive to copy around; in a sense,
173 it is accidental that it is represented by a pointer. If a pointer
174 represents a primitive value, then it really does point to that value:
175 no unevaluated thunks, no indirections…nothing can be at the
176 other end of the pointer than the primitive value.
177 A numerically-intensive program using unboxed types can
178 go a <emphasis>lot</emphasis> faster than its “standard”
179 counterpart—we saw a threefold speedup on one example.
183 There are some restrictions on the use of primitive types:
185 <listitem><para>The main restriction
186 is that you can't pass a primitive value to a polymorphic
187 function or store one in a polymorphic data type. This rules out
188 things like <literal>[Int#]</literal> (i.e. lists of primitive
189 integers). The reason for this restriction is that polymorphic
190 arguments and constructor fields are assumed to be pointers: if an
191 unboxed integer is stored in one of these, the garbage collector would
192 attempt to follow it, leading to unpredictable space leaks. Or a
193 <function>seq</function> operation on the polymorphic component may
194 attempt to dereference the pointer, with disastrous results. Even
195 worse, the unboxed value might be larger than a pointer
196 (<literal>Double#</literal> for instance).
199 <listitem><para> You cannot define a newtype whose representation type
200 (the argument type of the data constructor) is an unboxed type. Thus,
206 <listitem><para> You cannot bind a variable with an unboxed type
207 in a <emphasis>top-level</emphasis> binding.
209 <listitem><para> You cannot bind a variable with an unboxed type
210 in a <emphasis>recursive</emphasis> binding.
212 <listitem><para> You may bind unboxed variables in a (non-recursive,
213 non-top-level) pattern binding, but any such variable causes the entire
215 to become strict. For example:
217 data Foo = Foo Int Int#
219 f x = let (Foo a b, w) = ..rhs.. in ..body..
221 Since <literal>b</literal> has type <literal>Int#</literal>, the entire pattern
223 is strict, and the program behaves as if you had written
225 data Foo = Foo Int Int#
227 f x = case ..rhs.. of { (Foo a b, w) -> ..body.. }
236 <sect2 id="unboxed-tuples">
237 <title>Unboxed Tuples
241 Unboxed tuples aren't really exported by <literal>GHC.Exts</literal>,
242 they're available by default with <option>-fglasgow-exts</option>. An
243 unboxed tuple looks like this:
255 where <literal>e_1..e_n</literal> are expressions of any
256 type (primitive or non-primitive). The type of an unboxed tuple looks
261 Unboxed tuples are used for functions that need to return multiple
262 values, but they avoid the heap allocation normally associated with
263 using fully-fledged tuples. When an unboxed tuple is returned, the
264 components are put directly into registers or on the stack; the
265 unboxed tuple itself does not have a composite representation. Many
266 of the primitive operations listed in <literal>primops.txt.pp</literal> return unboxed
268 In particular, the <literal>IO</literal> and <literal>ST</literal> monads use unboxed
269 tuples to avoid unnecessary allocation during sequences of operations.
273 There are some pretty stringent restrictions on the use of unboxed tuples:
278 Values of unboxed tuple types are subject to the same restrictions as
279 other unboxed types; i.e. they may not be stored in polymorphic data
280 structures or passed to polymorphic functions.
287 No variable can have an unboxed tuple type, nor may a constructor or function
288 argument have an unboxed tuple type. The following are all illegal:
292 data Foo = Foo (# Int, Int #)
294 f :: (# Int, Int #) -> (# Int, Int #)
297 g :: (# Int, Int #) -> Int
300 h x = let y = (# x,x #) in ...
307 The typical use of unboxed tuples is simply to return multiple values,
308 binding those multiple results with a <literal>case</literal> expression, thus:
310 f x y = (# x+1, y-1 #)
311 g x = case f x x of { (# a, b #) -> a + b }
313 You can have an unboxed tuple in a pattern binding, thus
315 f x = let (# p,q #) = h x in ..body..
317 If the types of <literal>p</literal> and <literal>q</literal> are not unboxed,
318 the resulting binding is lazy like any other Haskell pattern binding. The
319 above example desugars like this:
321 f x = let t = case h x o f{ (# p,q #) -> (p,q)
326 Indeed, the bindings can even be recursive.
333 <!-- ====================== SYNTACTIC EXTENSIONS ======================= -->
335 <sect1 id="syntax-extns">
336 <title>Syntactic extensions</title>
338 <sect2 id="unicode-syntax">
339 <title>Unicode syntax</title>
341 extension <option>-XUnicodeSyntax</option><indexterm><primary><option>-XUnicodeSyntax</option></primary></indexterm>
342 enables Unicode characters to be used to stand for certain ASCII
343 character sequences. The following alternatives are provided:</para>
346 <tgroup cols="2" align="left" colsep="1" rowsep="1">
350 <entry>Unicode alternative</entry>
351 <entry>Code point</entry>
357 <entry><literal>::</literal></entry>
358 <entry>::</entry> <!-- no special char, apparently -->
359 <entry>0x2237</entry>
360 <entry>PROPORTION</entry>
365 <entry><literal>=></literal></entry>
366 <entry>⇒</entry>
367 <entry>0x21D2</entry>
368 <entry>RIGHTWARDS DOUBLE ARROW</entry>
373 <entry><literal>forall</literal></entry>
374 <entry>∀</entry>
375 <entry>0x2200</entry>
376 <entry>FOR ALL</entry>
381 <entry><literal>-></literal></entry>
382 <entry>→</entry>
383 <entry>0x2192</entry>
384 <entry>RIGHTWARDS ARROW</entry>
389 <entry><literal><-</literal></entry>
390 <entry>←</entry>
391 <entry>0x2190</entry>
392 <entry>LEFTWARDS ARROW</entry>
398 <entry>…</entry>
399 <entry>0x22EF</entry>
400 <entry>MIDLINE HORIZONTAL ELLIPSIS</entry>
407 <sect2 id="magic-hash">
408 <title>The magic hash</title>
409 <para>The language extension <option>-XMagicHash</option> allows "#" as a
410 postfix modifier to identifiers. Thus, "x#" is a valid variable, and "T#" is
411 a valid type constructor or data constructor.</para>
413 <para>The hash sign does not change sematics at all. We tend to use variable
414 names ending in "#" for unboxed values or types (e.g. <literal>Int#</literal>),
415 but there is no requirement to do so; they are just plain ordinary variables.
416 Nor does the <option>-XMagicHash</option> extension bring anything into scope.
417 For example, to bring <literal>Int#</literal> into scope you must
418 import <literal>GHC.Prim</literal> (see <xref linkend="primitives"/>);
419 the <option>-XMagicHash</option> extension
420 then allows you to <emphasis>refer</emphasis> to the <literal>Int#</literal>
421 that is now in scope.</para>
422 <para> The <option>-XMagicHash</option> also enables some new forms of literals (see <xref linkend="glasgow-unboxed"/>):
424 <listitem><para> <literal>'x'#</literal> has type <literal>Char#</literal></para> </listitem>
425 <listitem><para> <literal>"foo"#</literal> has type <literal>Addr#</literal></para> </listitem>
426 <listitem><para> <literal>3#</literal> has type <literal>Int#</literal>. In general,
427 any Haskell 98 integer lexeme followed by a <literal>#</literal> is an <literal>Int#</literal> literal, e.g.
428 <literal>-0x3A#</literal> as well as <literal>32#</literal></para>.</listitem>
429 <listitem><para> <literal>3##</literal> has type <literal>Word#</literal>. In general,
430 any non-negative Haskell 98 integer lexeme followed by <literal>##</literal>
431 is a <literal>Word#</literal>. </para> </listitem>
432 <listitem><para> <literal>3.2#</literal> has type <literal>Float#</literal>.</para> </listitem>
433 <listitem><para> <literal>3.2##</literal> has type <literal>Double#</literal></para> </listitem>
438 <sect2 id="new-qualified-operators">
439 <title>New qualified operator syntax</title>
441 <para>A new syntax for referencing qualified operators is
442 planned to be introduced by Haskell', and is enabled in GHC
444 the <option>-XNewQualifiedOperators</option><indexterm><primary><option>-XNewQualifiedOperators</option></primary></indexterm>
445 option. In the new syntax, the prefix form of a qualified
447 written <literal><replaceable>module</replaceable>.(<replaceable>symbol</replaceable>)</literal>
448 (in Haskell 98 this would
449 be <literal>(<replaceable>module</replaceable>.<replaceable>symbol</replaceable>)</literal>),
450 and the infix form is
451 written <literal>`<replaceable>module</replaceable>.(<replaceable>symbol</replaceable>)`</literal>
452 (in Haskell 98 this would
453 be <literal>`<replaceable>module</replaceable>.<replaceable>symbol</replaceable>`</literal>.
456 add x y = Prelude.(+) x y
457 subtract y = (`Prelude.(-)` y)
459 The new form of qualified operators is intended to regularise
460 the syntax by eliminating odd cases
461 like <literal>Prelude..</literal>. For example,
462 when <literal>NewQualifiedOperators</literal> is on, it is possible to
463 write the enumerated sequence <literal>[Monday..]</literal>
464 without spaces, whereas in Haskell 98 this would be a
465 reference to the operator ‘<literal>.</literal>‘
466 from module <literal>Monday</literal>.</para>
468 <para>When <option>-XNewQualifiedOperators</option> is on, the old Haskell
469 98 syntax for qualified operators is not accepted, so this
470 option may cause existing Haskell 98 code to break.</para>
475 <!-- ====================== HIERARCHICAL MODULES ======================= -->
478 <sect2 id="hierarchical-modules">
479 <title>Hierarchical Modules</title>
481 <para>GHC supports a small extension to the syntax of module
482 names: a module name is allowed to contain a dot
483 <literal>‘.’</literal>. This is also known as the
484 “hierarchical module namespace” extension, because
485 it extends the normally flat Haskell module namespace into a
486 more flexible hierarchy of modules.</para>
488 <para>This extension has very little impact on the language
489 itself; modules names are <emphasis>always</emphasis> fully
490 qualified, so you can just think of the fully qualified module
491 name as <quote>the module name</quote>. In particular, this
492 means that the full module name must be given after the
493 <literal>module</literal> keyword at the beginning of the
494 module; for example, the module <literal>A.B.C</literal> must
497 <programlisting>module A.B.C</programlisting>
500 <para>It is a common strategy to use the <literal>as</literal>
501 keyword to save some typing when using qualified names with
502 hierarchical modules. For example:</para>
505 import qualified Control.Monad.ST.Strict as ST
508 <para>For details on how GHC searches for source and interface
509 files in the presence of hierarchical modules, see <xref
510 linkend="search-path"/>.</para>
512 <para>GHC comes with a large collection of libraries arranged
513 hierarchically; see the accompanying <ulink
514 url="../libraries/index.html">library
515 documentation</ulink>. More libraries to install are available
517 url="http://hackage.haskell.org/packages/hackage.html">HackageDB</ulink>.</para>
520 <!-- ====================== PATTERN GUARDS ======================= -->
522 <sect2 id="pattern-guards">
523 <title>Pattern guards</title>
526 <indexterm><primary>Pattern guards (Glasgow extension)</primary></indexterm>
527 The discussion that follows is an abbreviated version of Simon Peyton Jones's original <ulink url="http://research.microsoft.com/~simonpj/Haskell/guards.html">proposal</ulink>. (Note that the proposal was written before pattern guards were implemented, so refers to them as unimplemented.)
531 Suppose we have an abstract data type of finite maps, with a
535 lookup :: FiniteMap -> Int -> Maybe Int
538 The lookup returns <function>Nothing</function> if the supplied key is not in the domain of the mapping, and <function>(Just v)</function> otherwise,
539 where <varname>v</varname> is the value that the key maps to. Now consider the following definition:
543 clunky env var1 var2 | ok1 && ok2 = val1 + val2
544 | otherwise = var1 + var2
555 The auxiliary functions are
559 maybeToBool :: Maybe a -> Bool
560 maybeToBool (Just x) = True
561 maybeToBool Nothing = False
563 expectJust :: Maybe a -> a
564 expectJust (Just x) = x
565 expectJust Nothing = error "Unexpected Nothing"
569 What is <function>clunky</function> doing? The guard <literal>ok1 &&
570 ok2</literal> checks that both lookups succeed, using
571 <function>maybeToBool</function> to convert the <function>Maybe</function>
572 types to booleans. The (lazily evaluated) <function>expectJust</function>
573 calls extract the values from the results of the lookups, and binds the
574 returned values to <varname>val1</varname> and <varname>val2</varname>
575 respectively. If either lookup fails, then clunky takes the
576 <literal>otherwise</literal> case and returns the sum of its arguments.
580 This is certainly legal Haskell, but it is a tremendously verbose and
581 un-obvious way to achieve the desired effect. Arguably, a more direct way
582 to write clunky would be to use case expressions:
586 clunky env var1 var2 = case lookup env var1 of
588 Just val1 -> case lookup env var2 of
590 Just val2 -> val1 + val2
596 This is a bit shorter, but hardly better. Of course, we can rewrite any set
597 of pattern-matching, guarded equations as case expressions; that is
598 precisely what the compiler does when compiling equations! The reason that
599 Haskell provides guarded equations is because they allow us to write down
600 the cases we want to consider, one at a time, independently of each other.
601 This structure is hidden in the case version. Two of the right-hand sides
602 are really the same (<function>fail</function>), and the whole expression
603 tends to become more and more indented.
607 Here is how I would write clunky:
612 | Just val1 <- lookup env var1
613 , Just val2 <- lookup env var2
615 ...other equations for clunky...
619 The semantics should be clear enough. The qualifiers are matched in order.
620 For a <literal><-</literal> qualifier, which I call a pattern guard, the
621 right hand side is evaluated and matched against the pattern on the left.
622 If the match fails then the whole guard fails and the next equation is
623 tried. If it succeeds, then the appropriate binding takes place, and the
624 next qualifier is matched, in the augmented environment. Unlike list
625 comprehensions, however, the type of the expression to the right of the
626 <literal><-</literal> is the same as the type of the pattern to its
627 left. The bindings introduced by pattern guards scope over all the
628 remaining guard qualifiers, and over the right hand side of the equation.
632 Just as with list comprehensions, boolean expressions can be freely mixed
633 with among the pattern guards. For example:
644 Haskell's current guards therefore emerge as a special case, in which the
645 qualifier list has just one element, a boolean expression.
649 <!-- ===================== View patterns =================== -->
651 <sect2 id="view-patterns">
656 View patterns are enabled by the flag <literal>-XViewPatterns</literal>.
657 More information and examples of view patterns can be found on the
658 <ulink url="http://hackage.haskell.org/trac/ghc/wiki/ViewPatterns">Wiki
663 View patterns are somewhat like pattern guards that can be nested inside
664 of other patterns. They are a convenient way of pattern-matching
665 against values of abstract types. For example, in a programming language
666 implementation, we might represent the syntax of the types of the
675 view :: Type -> TypeView
677 -- additional operations for constructing Typ's ...
680 The representation of Typ is held abstract, permitting implementations
681 to use a fancy representation (e.g., hash-consing to manage sharing).
683 Without view patterns, using this signature a little inconvenient:
685 size :: Typ -> Integer
686 size t = case view t of
688 Arrow t1 t2 -> size t1 + size t2
691 It is necessary to iterate the case, rather than using an equational
692 function definition. And the situation is even worse when the matching
693 against <literal>t</literal> is buried deep inside another pattern.
697 View patterns permit calling the view function inside the pattern and
698 matching against the result:
700 size (view -> Unit) = 1
701 size (view -> Arrow t1 t2) = size t1 + size t2
704 That is, we add a new form of pattern, written
705 <replaceable>expression</replaceable> <literal>-></literal>
706 <replaceable>pattern</replaceable> that means "apply the expression to
707 whatever we're trying to match against, and then match the result of
708 that application against the pattern". The expression can be any Haskell
709 expression of function type, and view patterns can be used wherever
714 The semantics of a pattern <literal>(</literal>
715 <replaceable>exp</replaceable> <literal>-></literal>
716 <replaceable>pat</replaceable> <literal>)</literal> are as follows:
722 <para>The variables bound by the view pattern are the variables bound by
723 <replaceable>pat</replaceable>.
727 Any variables in <replaceable>exp</replaceable> are bound occurrences,
728 but variables bound "to the left" in a pattern are in scope. This
729 feature permits, for example, one argument to a function to be used in
730 the view of another argument. For example, the function
731 <literal>clunky</literal> from <xref linkend="pattern-guards" /> can be
732 written using view patterns as follows:
735 clunky env (lookup env -> Just val1) (lookup env -> Just val2) = val1 + val2
736 ...other equations for clunky...
741 More precisely, the scoping rules are:
745 In a single pattern, variables bound by patterns to the left of a view
746 pattern expression are in scope. For example:
748 example :: Maybe ((String -> Integer,Integer), String) -> Bool
749 example Just ((f,_), f -> 4) = True
752 Additionally, in function definitions, variables bound by matching earlier curried
753 arguments may be used in view pattern expressions in later arguments:
755 example :: (String -> Integer) -> String -> Bool
756 example f (f -> 4) = True
758 That is, the scoping is the same as it would be if the curried arguments
759 were collected into a tuple.
765 In mutually recursive bindings, such as <literal>let</literal>,
766 <literal>where</literal>, or the top level, view patterns in one
767 declaration may not mention variables bound by other declarations. That
768 is, each declaration must be self-contained. For example, the following
769 program is not allowed:
776 restriction in the future; the only cost is that type checking patterns
777 would get a little more complicated.)
787 <listitem><para> Typing: If <replaceable>exp</replaceable> has type
788 <replaceable>T1</replaceable> <literal>-></literal>
789 <replaceable>T2</replaceable> and <replaceable>pat</replaceable> matches
790 a <replaceable>T2</replaceable>, then the whole view pattern matches a
791 <replaceable>T1</replaceable>.
794 <listitem><para> Matching: To the equations in Section 3.17.3 of the
795 <ulink url="http://www.haskell.org/onlinereport/">Haskell 98
796 Report</ulink>, add the following:
798 case v of { (e -> p) -> e1 ; _ -> e2 }
800 case (e v) of { p -> e1 ; _ -> e2 }
802 That is, to match a variable <replaceable>v</replaceable> against a pattern
803 <literal>(</literal> <replaceable>exp</replaceable>
804 <literal>-></literal> <replaceable>pat</replaceable>
805 <literal>)</literal>, evaluate <literal>(</literal>
806 <replaceable>exp</replaceable> <replaceable> v</replaceable>
807 <literal>)</literal> and match the result against
808 <replaceable>pat</replaceable>.
811 <listitem><para> Efficiency: When the same view function is applied in
812 multiple branches of a function definition or a case expression (e.g.,
813 in <literal>size</literal> above), GHC makes an attempt to collect these
814 applications into a single nested case expression, so that the view
815 function is only applied once. Pattern compilation in GHC follows the
816 matrix algorithm described in Chapter 4 of <ulink
817 url="http://research.microsoft.com/~simonpj/Papers/slpj-book-1987/">The
818 Implementation of Functional Programming Languages</ulink>. When the
819 top rows of the first column of a matrix are all view patterns with the
820 "same" expression, these patterns are transformed into a single nested
821 case. This includes, for example, adjacent view patterns that line up
824 f ((view -> A, p1), p2) = e1
825 f ((view -> B, p3), p4) = e2
829 <para> The current notion of when two view pattern expressions are "the
830 same" is very restricted: it is not even full syntactic equality.
831 However, it does include variables, literals, applications, and tuples;
832 e.g., two instances of <literal>view ("hi", "there")</literal> will be
833 collected. However, the current implementation does not compare up to
834 alpha-equivalence, so two instances of <literal>(x, view x ->
835 y)</literal> will not be coalesced.
845 <!-- ===================== Recursive do-notation =================== -->
847 <sect2 id="mdo-notation">
848 <title>The recursive do-notation
851 <para> The recursive do-notation (also known as mdo-notation) is implemented as described in
852 <ulink url="http://citeseer.ist.psu.edu/erk02recursive.html">A recursive do for Haskell</ulink>,
853 by Levent Erkok, John Launchbury,
854 Haskell Workshop 2002, pages: 29-37. Pittsburgh, Pennsylvania.
855 This paper is essential reading for anyone making non-trivial use of mdo-notation,
856 and we do not repeat it here.
859 The do-notation of Haskell does not allow <emphasis>recursive bindings</emphasis>,
860 that is, the variables bound in a do-expression are visible only in the textually following
861 code block. Compare this to a let-expression, where bound variables are visible in the entire binding
862 group. It turns out that several applications can benefit from recursive bindings in
863 the do-notation, and this extension provides the necessary syntactic support.
866 Here is a simple (yet contrived) example:
869 import Control.Monad.Fix
871 justOnes = mdo xs <- Just (1:xs)
875 As you can guess <literal>justOnes</literal> will evaluate to <literal>Just [1,1,1,...</literal>.
879 The Control.Monad.Fix library introduces the <literal>MonadFix</literal> class. It's definition is:
882 class Monad m => MonadFix m where
883 mfix :: (a -> m a) -> m a
886 The function <literal>mfix</literal>
887 dictates how the required recursion operation should be performed. For example,
888 <literal>justOnes</literal> desugars as follows:
890 justOnes = mfix (\xs' -> do { xs <- Just (1:xs'); return xs }
892 For full details of the way in which mdo is typechecked and desugared, see
893 the paper <ulink url="http://citeseer.ist.psu.edu/erk02recursive.html">A recursive do for Haskell</ulink>.
894 In particular, GHC implements the segmentation technique described in Section 3.2 of the paper.
897 If recursive bindings are required for a monad,
898 then that monad must be declared an instance of the <literal>MonadFix</literal> class.
899 The following instances of <literal>MonadFix</literal> are automatically provided: List, Maybe, IO.
900 Furthermore, the Control.Monad.ST and Control.Monad.ST.Lazy modules provide the instances of the MonadFix class
901 for Haskell's internal state monad (strict and lazy, respectively).
904 Here are some important points in using the recursive-do notation:
907 The recursive version of the do-notation uses the keyword <literal>mdo</literal> (rather
908 than <literal>do</literal>).
912 It is enabled with the flag <literal>-XRecursiveDo</literal>, which is in turn implied by
913 <literal>-fglasgow-exts</literal>.
917 Unlike ordinary do-notation, but like <literal>let</literal> and <literal>where</literal> bindings,
918 name shadowing is not allowed; that is, all the names bound in a single <literal>mdo</literal> must
919 be distinct (Section 3.3 of the paper).
923 Variables bound by a <literal>let</literal> statement in an <literal>mdo</literal>
924 are monomorphic in the <literal>mdo</literal> (Section 3.1 of the paper). However
925 GHC breaks the <literal>mdo</literal> into segments to enhance polymorphism,
926 and improve termination (Section 3.2 of the paper).
932 Historical note: The old implementation of the mdo-notation (and most
933 of the existing documents) used the name
934 <literal>MonadRec</literal> for the class and the corresponding library.
935 This name is not supported by GHC.
941 <!-- ===================== PARALLEL LIST COMPREHENSIONS =================== -->
943 <sect2 id="parallel-list-comprehensions">
944 <title>Parallel List Comprehensions</title>
945 <indexterm><primary>list comprehensions</primary><secondary>parallel</secondary>
947 <indexterm><primary>parallel list comprehensions</primary>
950 <para>Parallel list comprehensions are a natural extension to list
951 comprehensions. List comprehensions can be thought of as a nice
952 syntax for writing maps and filters. Parallel comprehensions
953 extend this to include the zipWith family.</para>
955 <para>A parallel list comprehension has multiple independent
956 branches of qualifier lists, each separated by a `|' symbol. For
957 example, the following zips together two lists:</para>
960 [ (x, y) | x <- xs | y <- ys ]
963 <para>The behavior of parallel list comprehensions follows that of
964 zip, in that the resulting list will have the same length as the
965 shortest branch.</para>
967 <para>We can define parallel list comprehensions by translation to
968 regular comprehensions. Here's the basic idea:</para>
970 <para>Given a parallel comprehension of the form: </para>
973 [ e | p1 <- e11, p2 <- e12, ...
974 | q1 <- e21, q2 <- e22, ...
979 <para>This will be translated to: </para>
982 [ e | ((p1,p2), (q1,q2), ...) <- zipN [(p1,p2) | p1 <- e11, p2 <- e12, ...]
983 [(q1,q2) | q1 <- e21, q2 <- e22, ...]
988 <para>where `zipN' is the appropriate zip for the given number of
993 <!-- ===================== TRANSFORM LIST COMPREHENSIONS =================== -->
995 <sect2 id="generalised-list-comprehensions">
996 <title>Generalised (SQL-Like) List Comprehensions</title>
997 <indexterm><primary>list comprehensions</primary><secondary>generalised</secondary>
999 <indexterm><primary>extended list comprehensions</primary>
1001 <indexterm><primary>group</primary></indexterm>
1002 <indexterm><primary>sql</primary></indexterm>
1005 <para>Generalised list comprehensions are a further enhancement to the
1006 list comprehension syntatic sugar to allow operations such as sorting
1007 and grouping which are familiar from SQL. They are fully described in the
1008 paper <ulink url="http://research.microsoft.com/~simonpj/papers/list-comp">
1009 Comprehensive comprehensions: comprehensions with "order by" and "group by"</ulink>,
1010 except that the syntax we use differs slightly from the paper.</para>
1011 <para>Here is an example:
1013 employees = [ ("Simon", "MS", 80)
1014 , ("Erik", "MS", 100)
1015 , ("Phil", "Ed", 40)
1016 , ("Gordon", "Ed", 45)
1017 , ("Paul", "Yale", 60)]
1019 output = [ (the dept, sum salary)
1020 | (name, dept, salary) <- employees
1021 , then group by dept
1022 , then sortWith by (sum salary)
1025 In this example, the list <literal>output</literal> would take on
1029 [("Yale", 60), ("Ed", 85), ("MS", 180)]
1032 <para>There are three new keywords: <literal>group</literal>, <literal>by</literal>, and <literal>using</literal>.
1033 (The function <literal>sortWith</literal> is not a keyword; it is an ordinary
1034 function that is exported by <literal>GHC.Exts</literal>.)</para>
1036 <para>There are five new forms of comprehension qualifier,
1037 all introduced by the (existing) keyword <literal>then</literal>:
1045 This statement requires that <literal>f</literal> have the type <literal>
1046 forall a. [a] -> [a]</literal>. You can see an example of it's use in the
1047 motivating example, as this form is used to apply <literal>take 5</literal>.
1058 This form is similar to the previous one, but allows you to create a function
1059 which will be passed as the first argument to f. As a consequence f must have
1060 the type <literal>forall a. (a -> t) -> [a] -> [a]</literal>. As you can see
1061 from the type, this function lets f "project out" some information
1062 from the elements of the list it is transforming.</para>
1064 <para>An example is shown in the opening example, where <literal>sortWith</literal>
1065 is supplied with a function that lets it find out the <literal>sum salary</literal>
1066 for any item in the list comprehension it transforms.</para>
1074 then group by e using f
1077 <para>This is the most general of the grouping-type statements. In this form,
1078 f is required to have type <literal>forall a. (a -> t) -> [a] -> [[a]]</literal>.
1079 As with the <literal>then f by e</literal> case above, the first argument
1080 is a function supplied to f by the compiler which lets it compute e on every
1081 element of the list being transformed. However, unlike the non-grouping case,
1082 f additionally partitions the list into a number of sublists: this means that
1083 at every point after this statement, binders occurring before it in the comprehension
1084 refer to <emphasis>lists</emphasis> of possible values, not single values. To help understand
1085 this, let's look at an example:</para>
1088 -- This works similarly to groupWith in GHC.Exts, but doesn't sort its input first
1089 groupRuns :: Eq b => (a -> b) -> [a] -> [[a]]
1090 groupRuns f = groupBy (\x y -> f x == f y)
1092 output = [ (the x, y)
1093 | x <- ([1..3] ++ [1..2])
1095 , then group by x using groupRuns ]
1098 <para>This results in the variable <literal>output</literal> taking on the value below:</para>
1101 [(1, [4, 5, 6]), (2, [4, 5, 6]), (3, [4, 5, 6]), (1, [4, 5, 6]), (2, [4, 5, 6])]
1104 <para>Note that we have used the <literal>the</literal> function to change the type
1105 of x from a list to its original numeric type. The variable y, in contrast, is left
1106 unchanged from the list form introduced by the grouping.</para>
1116 <para>This form of grouping is essentially the same as the one described above. However,
1117 since no function to use for the grouping has been supplied it will fall back on the
1118 <literal>groupWith</literal> function defined in
1119 <ulink url="../libraries/base/GHC-Exts.html"><literal>GHC.Exts</literal></ulink>. This
1120 is the form of the group statement that we made use of in the opening example.</para>
1131 <para>With this form of the group statement, f is required to simply have the type
1132 <literal>forall a. [a] -> [[a]]</literal>, which will be used to group up the
1133 comprehension so far directly. An example of this form is as follows:</para>
1139 , then group using inits]
1142 <para>This will yield a list containing every prefix of the word "hello" written out 5 times:</para>
1145 ["","h","he","hel","hell","hello","helloh","hellohe","hellohel","hellohell","hellohello","hellohelloh",...]
1153 <!-- ===================== REBINDABLE SYNTAX =================== -->
1155 <sect2 id="rebindable-syntax">
1156 <title>Rebindable syntax and the implicit Prelude import</title>
1158 <para><indexterm><primary>-XNoImplicitPrelude
1159 option</primary></indexterm> GHC normally imports
1160 <filename>Prelude.hi</filename> files for you. If you'd
1161 rather it didn't, then give it a
1162 <option>-XNoImplicitPrelude</option> option. The idea is
1163 that you can then import a Prelude of your own. (But don't
1164 call it <literal>Prelude</literal>; the Haskell module
1165 namespace is flat, and you must not conflict with any
1166 Prelude module.)</para>
1168 <para>Suppose you are importing a Prelude of your own
1169 in order to define your own numeric class
1170 hierarchy. It completely defeats that purpose if the
1171 literal "1" means "<literal>Prelude.fromInteger
1172 1</literal>", which is what the Haskell Report specifies.
1173 So the <option>-XNoImplicitPrelude</option>
1174 flag <emphasis>also</emphasis> causes
1175 the following pieces of built-in syntax to refer to
1176 <emphasis>whatever is in scope</emphasis>, not the Prelude
1180 <para>An integer literal <literal>368</literal> means
1181 "<literal>fromInteger (368::Integer)</literal>", rather than
1182 "<literal>Prelude.fromInteger (368::Integer)</literal>".
1185 <listitem><para>Fractional literals are handed in just the same way,
1186 except that the translation is
1187 <literal>fromRational (3.68::Rational)</literal>.
1190 <listitem><para>The equality test in an overloaded numeric pattern
1191 uses whatever <literal>(==)</literal> is in scope.
1194 <listitem><para>The subtraction operation, and the
1195 greater-than-or-equal test, in <literal>n+k</literal> patterns
1196 use whatever <literal>(-)</literal> and <literal>(>=)</literal> are in scope.
1200 <para>Negation (e.g. "<literal>- (f x)</literal>")
1201 means "<literal>negate (f x)</literal>", both in numeric
1202 patterns, and expressions.
1206 <para>"Do" notation is translated using whatever
1207 functions <literal>(>>=)</literal>,
1208 <literal>(>>)</literal>, and <literal>fail</literal>,
1209 are in scope (not the Prelude
1210 versions). List comprehensions, mdo (<xref linkend="mdo-notation"/>), and parallel array
1211 comprehensions, are unaffected. </para></listitem>
1215 notation (see <xref linkend="arrow-notation"/>)
1216 uses whatever <literal>arr</literal>,
1217 <literal>(>>>)</literal>, <literal>first</literal>,
1218 <literal>app</literal>, <literal>(|||)</literal> and
1219 <literal>loop</literal> functions are in scope. But unlike the
1220 other constructs, the types of these functions must match the
1221 Prelude types very closely. Details are in flux; if you want
1225 In all cases (apart from arrow notation), the static semantics should be that of the desugared form,
1226 even if that is a little unexpected. For example, the
1227 static semantics of the literal <literal>368</literal>
1228 is exactly that of <literal>fromInteger (368::Integer)</literal>; it's fine for
1229 <literal>fromInteger</literal> to have any of the types:
1231 fromInteger :: Integer -> Integer
1232 fromInteger :: forall a. Foo a => Integer -> a
1233 fromInteger :: Num a => a -> Integer
1234 fromInteger :: Integer -> Bool -> Bool
1238 <para>Be warned: this is an experimental facility, with
1239 fewer checks than usual. Use <literal>-dcore-lint</literal>
1240 to typecheck the desugared program. If Core Lint is happy
1241 you should be all right.</para>
1245 <sect2 id="postfix-operators">
1246 <title>Postfix operators</title>
1249 The <option>-XPostfixOperators</option> flag enables a small
1250 extension to the syntax of left operator sections, which allows you to
1251 define postfix operators. The extension is this: the left section
1255 is equivalent (from the point of view of both type checking and execution) to the expression
1259 (for any expression <literal>e</literal> and operator <literal>(!)</literal>.
1260 The strict Haskell 98 interpretation is that the section is equivalent to
1264 That is, the operator must be a function of two arguments. GHC allows it to
1265 take only one argument, and that in turn allows you to write the function
1268 <para>The extension does not extend to the left-hand side of function
1269 definitions; you must define such a function in prefix form.</para>
1273 <sect2 id="disambiguate-fields">
1274 <title>Record field disambiguation</title>
1276 In record construction and record pattern matching
1277 it is entirely unambiguous which field is referred to, even if there are two different
1278 data types in scope with a common field name. For example:
1281 data S = MkS { x :: Int, y :: Bool }
1286 data T = MkT { x :: Int }
1288 ok1 (MkS { x = n }) = n+1 -- Unambiguous
1290 ok2 n = MkT { x = n+1 } -- Unambiguous
1292 bad1 k = k { x = 3 } -- Ambiguous
1293 bad2 k = x k -- Ambiguous
1295 Even though there are two <literal>x</literal>'s in scope,
1296 it is clear that the <literal>x</literal> in the pattern in the
1297 definition of <literal>ok1</literal> can only mean the field
1298 <literal>x</literal> from type <literal>S</literal>. Similarly for
1299 the function <literal>ok2</literal>. However, in the record update
1300 in <literal>bad1</literal> and the record selection in <literal>bad2</literal>
1301 it is not clear which of the two types is intended.
1304 Haskell 98 regards all four as ambiguous, but with the
1305 <option>-XDisambiguateRecordFields</option> flag, GHC will accept
1306 the former two. The rules are precisely the same as those for instance
1307 declarations in Haskell 98, where the method names on the left-hand side
1308 of the method bindings in an instance declaration refer unambiguously
1309 to the method of that class (provided they are in scope at all), even
1310 if there are other variables in scope with the same name.
1311 This reduces the clutter of qualified names when you import two
1312 records from different modules that use the same field name.
1316 <!-- ===================== Record puns =================== -->
1318 <sect2 id="record-puns">
1323 Record puns are enabled by the flag <literal>-XNamedFieldPuns</literal>.
1327 When using records, it is common to write a pattern that binds a
1328 variable with the same name as a record field, such as:
1331 data C = C {a :: Int}
1337 Record punning permits the variable name to be elided, so one can simply
1344 to mean the same pattern as above. That is, in a record pattern, the
1345 pattern <literal>a</literal> expands into the pattern <literal>a =
1346 a</literal> for the same name <literal>a</literal>.
1350 Note that puns and other patterns can be mixed in the same record:
1352 data C = C {a :: Int, b :: Int}
1353 f (C {a, b = 4}) = a
1355 and that puns can be used wherever record patterns occur (e.g. in
1356 <literal>let</literal> bindings or at the top-level).
1360 Record punning can also be used in an expression, writing, for example,
1366 let a = 1 in C {a = a}
1369 Note that this expansion is purely syntactic, so the record pun
1370 expression refers to the nearest enclosing variable that is spelled the
1371 same as the field name.
1376 <!-- ===================== Record wildcards =================== -->
1378 <sect2 id="record-wildcards">
1379 <title>Record wildcards
1383 Record wildcards are enabled by the flag <literal>-XRecordWildCards</literal>.
1387 For records with many fields, it can be tiresome to write out each field
1388 individually in a record pattern, as in
1390 data C = C {a :: Int, b :: Int, c :: Int, d :: Int}
1391 f (C {a = 1, b = b, c = c, d = d}) = b + c + d
1396 Record wildcard syntax permits a (<literal>..</literal>) in a record
1397 pattern, where each elided field <literal>f</literal> is replaced by the
1398 pattern <literal>f = f</literal>. For example, the above pattern can be
1401 f (C {a = 1, ..}) = b + c + d
1406 Note that wildcards can be mixed with other patterns, including puns
1407 (<xref linkend="record-puns"/>); for example, in a pattern <literal>C {a
1408 = 1, b, ..})</literal>. Additionally, record wildcards can be used
1409 wherever record patterns occur, including in <literal>let</literal>
1410 bindings and at the top-level. For example, the top-level binding
1414 defines <literal>b</literal>, <literal>c</literal>, and
1415 <literal>d</literal>.
1419 Record wildcards can also be used in expressions, writing, for example,
1422 let {a = 1; b = 2; c = 3; d = 4} in C {..}
1428 let {a = 1; b = 2; c = 3; d = 4} in C {a=a, b=b, c=c, d=d}
1431 Note that this expansion is purely syntactic, so the record wildcard
1432 expression refers to the nearest enclosing variables that are spelled
1433 the same as the omitted field names.
1438 <!-- ===================== Local fixity declarations =================== -->
1440 <sect2 id="local-fixity-declarations">
1441 <title>Local Fixity Declarations
1444 <para>A careful reading of the Haskell 98 Report reveals that fixity
1445 declarations (<literal>infix</literal>, <literal>infixl</literal>, and
1446 <literal>infixr</literal>) are permitted to appear inside local bindings
1447 such those introduced by <literal>let</literal> and
1448 <literal>where</literal>. However, the Haskell Report does not specify
1449 the semantics of such bindings very precisely.
1452 <para>In GHC, a fixity declaration may accompany a local binding:
1459 and the fixity declaration applies wherever the binding is in scope.
1460 For example, in a <literal>let</literal>, it applies in the right-hand
1461 sides of other <literal>let</literal>-bindings and the body of the
1462 <literal>let</literal>C. Or, in recursive <literal>do</literal>
1463 expressions (<xref linkend="mdo-notation"/>), the local fixity
1464 declarations of a <literal>let</literal> statement scope over other
1465 statements in the group, just as the bound name does.
1469 Moreover, a local fixity declaration *must* accompany a local binding of
1470 that name: it is not possible to revise the fixity of name bound
1473 let infixr 9 $ in ...
1476 Because local fixity declarations are technically Haskell 98, no flag is
1477 necessary to enable them.
1481 <sect2 id="package-imports">
1482 <title>Package-qualified imports</title>
1484 <para>With the <option>-XPackageImports</option> flag, GHC allows
1485 import declarations to be qualified by the package name that the
1486 module is intended to be imported from. For example:</para>
1489 import "network" Network.Socket
1492 <para>would import the module <literal>Network.Socket</literal> from
1493 the package <literal>network</literal> (any version). This may
1494 be used to disambiguate an import when the same module is
1495 available from multiple packages, or is present in both the
1496 current package being built and an external package.</para>
1498 <para>Note: you probably don't need to use this feature, it was
1499 added mainly so that we can build backwards-compatible versions of
1500 packages when APIs change. It can lead to fragile dependencies in
1501 the common case: modules occasionally move from one package to
1502 another, rendering any package-qualified imports broken.</para>
1505 <sect2 id="syntax-stolen">
1506 <title>Summary of stolen syntax</title>
1508 <para>Turning on an option that enables special syntax
1509 <emphasis>might</emphasis> cause working Haskell 98 code to fail
1510 to compile, perhaps because it uses a variable name which has
1511 become a reserved word. This section lists the syntax that is
1512 "stolen" by language extensions.
1514 notation and nonterminal names from the Haskell 98 lexical syntax
1515 (see the Haskell 98 Report).
1516 We only list syntax changes here that might affect
1517 existing working programs (i.e. "stolen" syntax). Many of these
1518 extensions will also enable new context-free syntax, but in all
1519 cases programs written to use the new syntax would not be
1520 compilable without the option enabled.</para>
1522 <para>There are two classes of special
1527 <para>New reserved words and symbols: character sequences
1528 which are no longer available for use as identifiers in the
1532 <para>Other special syntax: sequences of characters that have
1533 a different meaning when this particular option is turned
1538 The following syntax is stolen:
1543 <literal>forall</literal>
1544 <indexterm><primary><literal>forall</literal></primary></indexterm>
1547 Stolen (in types) by: <option>-XScopedTypeVariables</option>,
1548 <option>-XLiberalTypeSynonyms</option>,
1549 <option>-XRank2Types</option>,
1550 <option>-XRankNTypes</option>,
1551 <option>-XPolymorphicComponents</option>,
1552 <option>-XExistentialQuantification</option>
1558 <literal>mdo</literal>
1559 <indexterm><primary><literal>mdo</literal></primary></indexterm>
1562 Stolen by: <option>-XRecursiveDo</option>,
1568 <literal>foreign</literal>
1569 <indexterm><primary><literal>foreign</literal></primary></indexterm>
1572 Stolen by: <option>-XForeignFunctionInterface</option>,
1578 <literal>rec</literal>,
1579 <literal>proc</literal>, <literal>-<</literal>,
1580 <literal>>-</literal>, <literal>-<<</literal>,
1581 <literal>>>-</literal>, and <literal>(|</literal>,
1582 <literal>|)</literal> brackets
1583 <indexterm><primary><literal>proc</literal></primary></indexterm>
1586 Stolen by: <option>-XArrows</option>,
1592 <literal>?<replaceable>varid</replaceable></literal>,
1593 <literal>%<replaceable>varid</replaceable></literal>
1594 <indexterm><primary>implicit parameters</primary></indexterm>
1597 Stolen by: <option>-XImplicitParams</option>,
1603 <literal>[|</literal>,
1604 <literal>[e|</literal>, <literal>[p|</literal>,
1605 <literal>[d|</literal>, <literal>[t|</literal>,
1606 <literal>$(</literal>,
1607 <literal>$<replaceable>varid</replaceable></literal>
1608 <indexterm><primary>Template Haskell</primary></indexterm>
1611 Stolen by: <option>-XTemplateHaskell</option>,
1617 <literal>[:<replaceable>varid</replaceable>|</literal>
1618 <indexterm><primary>quasi-quotation</primary></indexterm>
1621 Stolen by: <option>-XQuasiQuotes</option>,
1627 <replaceable>varid</replaceable>{<literal>#</literal>},
1628 <replaceable>char</replaceable><literal>#</literal>,
1629 <replaceable>string</replaceable><literal>#</literal>,
1630 <replaceable>integer</replaceable><literal>#</literal>,
1631 <replaceable>float</replaceable><literal>#</literal>,
1632 <replaceable>float</replaceable><literal>##</literal>,
1633 <literal>(#</literal>, <literal>#)</literal>,
1636 Stolen by: <option>-XMagicHash</option>,
1645 <!-- TYPE SYSTEM EXTENSIONS -->
1646 <sect1 id="data-type-extensions">
1647 <title>Extensions to data types and type synonyms</title>
1649 <sect2 id="nullary-types">
1650 <title>Data types with no constructors</title>
1652 <para>With the <option>-fglasgow-exts</option> flag, GHC lets you declare
1653 a data type with no constructors. For example:</para>
1657 data T a -- T :: * -> *
1660 <para>Syntactically, the declaration lacks the "= constrs" part. The
1661 type can be parameterised over types of any kind, but if the kind is
1662 not <literal>*</literal> then an explicit kind annotation must be used
1663 (see <xref linkend="kinding"/>).</para>
1665 <para>Such data types have only one value, namely bottom.
1666 Nevertheless, they can be useful when defining "phantom types".</para>
1669 <sect2 id="infix-tycons">
1670 <title>Infix type constructors, classes, and type variables</title>
1673 GHC allows type constructors, classes, and type variables to be operators, and
1674 to be written infix, very much like expressions. More specifically:
1677 A type constructor or class can be an operator, beginning with a colon; e.g. <literal>:*:</literal>.
1678 The lexical syntax is the same as that for data constructors.
1681 Data type and type-synonym declarations can be written infix, parenthesised
1682 if you want further arguments. E.g.
1684 data a :*: b = Foo a b
1685 type a :+: b = Either a b
1686 class a :=: b where ...
1688 data (a :**: b) x = Baz a b x
1689 type (a :++: b) y = Either (a,b) y
1693 Types, and class constraints, can be written infix. For example
1696 f :: (a :=: b) => a -> b
1700 A type variable can be an (unqualified) operator e.g. <literal>+</literal>.
1701 The lexical syntax is the same as that for variable operators, excluding "(.)",
1702 "(!)", and "(*)". In a binding position, the operator must be
1703 parenthesised. For example:
1705 type T (+) = Int + Int
1709 liftA2 :: Arrow (~>)
1710 => (a -> b -> c) -> (e ~> a) -> (e ~> b) -> (e ~> c)
1716 as for expressions, both for type constructors and type variables; e.g. <literal>Int `Either` Bool</literal>, or
1717 <literal>Int `a` Bool</literal>. Similarly, parentheses work the same; e.g. <literal>(:*:) Int Bool</literal>.
1720 Fixities may be declared for type constructors, or classes, just as for data constructors. However,
1721 one cannot distinguish between the two in a fixity declaration; a fixity declaration
1722 sets the fixity for a data constructor and the corresponding type constructor. For example:
1726 sets the fixity for both type constructor <literal>T</literal> and data constructor <literal>T</literal>,
1727 and similarly for <literal>:*:</literal>.
1728 <literal>Int `a` Bool</literal>.
1731 Function arrow is <literal>infixr</literal> with fixity 0. (This might change; I'm not sure what it should be.)
1738 <sect2 id="type-synonyms">
1739 <title>Liberalised type synonyms</title>
1742 Type synonyms are like macros at the type level, but Haskell 98 imposes many rules
1743 on individual synonym declarations.
1744 With the <option>-XLiberalTypeSynonyms</option> extension,
1745 GHC does validity checking on types <emphasis>only after expanding type synonyms</emphasis>.
1746 That means that GHC can be very much more liberal about type synonyms than Haskell 98.
1749 <listitem> <para>You can write a <literal>forall</literal> (including overloading)
1750 in a type synonym, thus:
1752 type Discard a = forall b. Show b => a -> b -> (a, String)
1757 g :: Discard Int -> (Int,String) -- A rank-2 type
1764 If you also use <option>-XUnboxedTuples</option>,
1765 you can write an unboxed tuple in a type synonym:
1767 type Pr = (# Int, Int #)
1775 You can apply a type synonym to a forall type:
1777 type Foo a = a -> a -> Bool
1779 f :: Foo (forall b. b->b)
1781 After expanding the synonym, <literal>f</literal> has the legal (in GHC) type:
1783 f :: (forall b. b->b) -> (forall b. b->b) -> Bool
1788 You can apply a type synonym to a partially applied type synonym:
1790 type Generic i o = forall x. i x -> o x
1793 foo :: Generic Id []
1795 After expanding the synonym, <literal>foo</literal> has the legal (in GHC) type:
1797 foo :: forall x. x -> [x]
1805 GHC currently does kind checking before expanding synonyms (though even that
1809 After expanding type synonyms, GHC does validity checking on types, looking for
1810 the following mal-formedness which isn't detected simply by kind checking:
1813 Type constructor applied to a type involving for-alls.
1816 Unboxed tuple on left of an arrow.
1819 Partially-applied type synonym.
1823 this will be rejected:
1825 type Pr = (# Int, Int #)
1830 because GHC does not allow unboxed tuples on the left of a function arrow.
1835 <sect2 id="existential-quantification">
1836 <title>Existentially quantified data constructors
1840 The idea of using existential quantification in data type declarations
1841 was suggested by Perry, and implemented in Hope+ (Nigel Perry, <emphasis>The Implementation
1842 of Practical Functional Programming Languages</emphasis>, PhD Thesis, University of
1843 London, 1991). It was later formalised by Laufer and Odersky
1844 (<emphasis>Polymorphic type inference and abstract data types</emphasis>,
1845 TOPLAS, 16(5), pp1411-1430, 1994).
1846 It's been in Lennart
1847 Augustsson's <command>hbc</command> Haskell compiler for several years, and
1848 proved very useful. Here's the idea. Consider the declaration:
1854 data Foo = forall a. MkFoo a (a -> Bool)
1861 The data type <literal>Foo</literal> has two constructors with types:
1867 MkFoo :: forall a. a -> (a -> Bool) -> Foo
1874 Notice that the type variable <literal>a</literal> in the type of <function>MkFoo</function>
1875 does not appear in the data type itself, which is plain <literal>Foo</literal>.
1876 For example, the following expression is fine:
1882 [MkFoo 3 even, MkFoo 'c' isUpper] :: [Foo]
1888 Here, <literal>(MkFoo 3 even)</literal> packages an integer with a function
1889 <function>even</function> that maps an integer to <literal>Bool</literal>; and <function>MkFoo 'c'
1890 isUpper</function> packages a character with a compatible function. These
1891 two things are each of type <literal>Foo</literal> and can be put in a list.
1895 What can we do with a value of type <literal>Foo</literal>?. In particular,
1896 what happens when we pattern-match on <function>MkFoo</function>?
1902 f (MkFoo val fn) = ???
1908 Since all we know about <literal>val</literal> and <function>fn</function> is that they
1909 are compatible, the only (useful) thing we can do with them is to
1910 apply <function>fn</function> to <literal>val</literal> to get a boolean. For example:
1917 f (MkFoo val fn) = fn val
1923 What this allows us to do is to package heterogeneous values
1924 together with a bunch of functions that manipulate them, and then treat
1925 that collection of packages in a uniform manner. You can express
1926 quite a bit of object-oriented-like programming this way.
1929 <sect3 id="existential">
1930 <title>Why existential?
1934 What has this to do with <emphasis>existential</emphasis> quantification?
1935 Simply that <function>MkFoo</function> has the (nearly) isomorphic type
1941 MkFoo :: (exists a . (a, a -> Bool)) -> Foo
1947 But Haskell programmers can safely think of the ordinary
1948 <emphasis>universally</emphasis> quantified type given above, thereby avoiding
1949 adding a new existential quantification construct.
1954 <sect3 id="existential-with-context">
1955 <title>Existentials and type classes</title>
1958 An easy extension is to allow
1959 arbitrary contexts before the constructor. For example:
1965 data Baz = forall a. Eq a => Baz1 a a
1966 | forall b. Show b => Baz2 b (b -> b)
1972 The two constructors have the types you'd expect:
1978 Baz1 :: forall a. Eq a => a -> a -> Baz
1979 Baz2 :: forall b. Show b => b -> (b -> b) -> Baz
1985 But when pattern matching on <function>Baz1</function> the matched values can be compared
1986 for equality, and when pattern matching on <function>Baz2</function> the first matched
1987 value can be converted to a string (as well as applying the function to it).
1988 So this program is legal:
1995 f (Baz1 p q) | p == q = "Yes"
1997 f (Baz2 v fn) = show (fn v)
2003 Operationally, in a dictionary-passing implementation, the
2004 constructors <function>Baz1</function> and <function>Baz2</function> must store the
2005 dictionaries for <literal>Eq</literal> and <literal>Show</literal> respectively, and
2006 extract it on pattern matching.
2011 <sect3 id="existential-records">
2012 <title>Record Constructors</title>
2015 GHC allows existentials to be used with records syntax as well. For example:
2018 data Counter a = forall self. NewCounter
2020 , _inc :: self -> self
2021 , _display :: self -> IO ()
2025 Here <literal>tag</literal> is a public field, with a well-typed selector
2026 function <literal>tag :: Counter a -> a</literal>. The <literal>self</literal>
2027 type is hidden from the outside; any attempt to apply <literal>_this</literal>,
2028 <literal>_inc</literal> or <literal>_display</literal> as functions will raise a
2029 compile-time error. In other words, <emphasis>GHC defines a record selector function
2030 only for fields whose type does not mention the existentially-quantified variables</emphasis>.
2031 (This example used an underscore in the fields for which record selectors
2032 will not be defined, but that is only programming style; GHC ignores them.)
2036 To make use of these hidden fields, we need to create some helper functions:
2039 inc :: Counter a -> Counter a
2040 inc (NewCounter x i d t) = NewCounter
2041 { _this = i x, _inc = i, _display = d, tag = t }
2043 display :: Counter a -> IO ()
2044 display NewCounter{ _this = x, _display = d } = d x
2047 Now we can define counters with different underlying implementations:
2050 counterA :: Counter String
2051 counterA = NewCounter
2052 { _this = 0, _inc = (1+), _display = print, tag = "A" }
2054 counterB :: Counter String
2055 counterB = NewCounter
2056 { _this = "", _inc = ('#':), _display = putStrLn, tag = "B" }
2059 display (inc counterA) -- prints "1"
2060 display (inc (inc counterB)) -- prints "##"
2063 Record update syntax is supported for existentials (and GADTs):
2065 setTag :: Counter a -> a -> Counter a
2066 setTag obj t = obj{ tag = t }
2068 The rule for record update is this: <emphasis>
2069 the types of the updated fields may
2070 mention only the universally-quantified type variables
2071 of the data constructor. For GADTs, the field may mention only types
2072 that appear as a simple type-variable argument in the constructor's result
2073 type</emphasis>. For example:
2075 data T a b where { T1 { f1::a, f2::b, f3::(b,c) } :: T a b } -- c is existential
2076 upd1 t x = t { f1=x } -- OK: upd1 :: T a b -> a' -> T a' b
2077 upd2 t x = t { f3=x } -- BAD (f3's type mentions c, which is
2078 -- existentially quantified)
2080 data G a b where { G1 { g1::a, g2::c } :: G a [c] }
2081 upd3 g x = g { g1=x } -- OK: upd3 :: G a b -> c -> G c b
2082 upd4 g x = g { g2=x } -- BAD (f2's type mentions c, which is not a simple
2083 -- type-variable argument in G1's result type)
2091 <title>Restrictions</title>
2094 There are several restrictions on the ways in which existentially-quantified
2095 constructors can be use.
2104 When pattern matching, each pattern match introduces a new,
2105 distinct, type for each existential type variable. These types cannot
2106 be unified with any other type, nor can they escape from the scope of
2107 the pattern match. For example, these fragments are incorrect:
2115 Here, the type bound by <function>MkFoo</function> "escapes", because <literal>a</literal>
2116 is the result of <function>f1</function>. One way to see why this is wrong is to
2117 ask what type <function>f1</function> has:
2121 f1 :: Foo -> a -- Weird!
2125 What is this "<literal>a</literal>" in the result type? Clearly we don't mean
2130 f1 :: forall a. Foo -> a -- Wrong!
2134 The original program is just plain wrong. Here's another sort of error
2138 f2 (Baz1 a b) (Baz1 p q) = a==q
2142 It's ok to say <literal>a==b</literal> or <literal>p==q</literal>, but
2143 <literal>a==q</literal> is wrong because it equates the two distinct types arising
2144 from the two <function>Baz1</function> constructors.
2152 You can't pattern-match on an existentially quantified
2153 constructor in a <literal>let</literal> or <literal>where</literal> group of
2154 bindings. So this is illegal:
2158 f3 x = a==b where { Baz1 a b = x }
2161 Instead, use a <literal>case</literal> expression:
2164 f3 x = case x of Baz1 a b -> a==b
2167 In general, you can only pattern-match
2168 on an existentially-quantified constructor in a <literal>case</literal> expression or
2169 in the patterns of a function definition.
2171 The reason for this restriction is really an implementation one.
2172 Type-checking binding groups is already a nightmare without
2173 existentials complicating the picture. Also an existential pattern
2174 binding at the top level of a module doesn't make sense, because it's
2175 not clear how to prevent the existentially-quantified type "escaping".
2176 So for now, there's a simple-to-state restriction. We'll see how
2184 You can't use existential quantification for <literal>newtype</literal>
2185 declarations. So this is illegal:
2189 newtype T = forall a. Ord a => MkT a
2193 Reason: a value of type <literal>T</literal> must be represented as a
2194 pair of a dictionary for <literal>Ord t</literal> and a value of type
2195 <literal>t</literal>. That contradicts the idea that
2196 <literal>newtype</literal> should have no concrete representation.
2197 You can get just the same efficiency and effect by using
2198 <literal>data</literal> instead of <literal>newtype</literal>. If
2199 there is no overloading involved, then there is more of a case for
2200 allowing an existentially-quantified <literal>newtype</literal>,
2201 because the <literal>data</literal> version does carry an
2202 implementation cost, but single-field existentially quantified
2203 constructors aren't much use. So the simple restriction (no
2204 existential stuff on <literal>newtype</literal>) stands, unless there
2205 are convincing reasons to change it.
2213 You can't use <literal>deriving</literal> to define instances of a
2214 data type with existentially quantified data constructors.
2216 Reason: in most cases it would not make sense. For example:;
2219 data T = forall a. MkT [a] deriving( Eq )
2222 To derive <literal>Eq</literal> in the standard way we would need to have equality
2223 between the single component of two <function>MkT</function> constructors:
2227 (MkT a) == (MkT b) = ???
2230 But <varname>a</varname> and <varname>b</varname> have distinct types, and so can't be compared.
2231 It's just about possible to imagine examples in which the derived instance
2232 would make sense, but it seems altogether simpler simply to prohibit such
2233 declarations. Define your own instances!
2244 <!-- ====================== Generalised algebraic data types ======================= -->
2246 <sect2 id="gadt-style">
2247 <title>Declaring data types with explicit constructor signatures</title>
2249 <para>GHC allows you to declare an algebraic data type by
2250 giving the type signatures of constructors explicitly. For example:
2254 Just :: a -> Maybe a
2256 The form is called a "GADT-style declaration"
2257 because Generalised Algebraic Data Types, described in <xref linkend="gadt"/>,
2258 can only be declared using this form.</para>
2259 <para>Notice that GADT-style syntax generalises existential types (<xref linkend="existential-quantification"/>).
2260 For example, these two declarations are equivalent:
2262 data Foo = forall a. MkFoo a (a -> Bool)
2263 data Foo' where { MKFoo :: a -> (a->Bool) -> Foo' }
2266 <para>Any data type that can be declared in standard Haskell-98 syntax
2267 can also be declared using GADT-style syntax.
2268 The choice is largely stylistic, but GADT-style declarations differ in one important respect:
2269 they treat class constraints on the data constructors differently.
2270 Specifically, if the constructor is given a type-class context, that
2271 context is made available by pattern matching. For example:
2274 MkSet :: Eq a => [a] -> Set a
2276 makeSet :: Eq a => [a] -> Set a
2277 makeSet xs = MkSet (nub xs)
2279 insert :: a -> Set a -> Set a
2280 insert a (MkSet as) | a `elem` as = MkSet as
2281 | otherwise = MkSet (a:as)
2283 A use of <literal>MkSet</literal> as a constructor (e.g. in the definition of <literal>makeSet</literal>)
2284 gives rise to a <literal>(Eq a)</literal>
2285 constraint, as you would expect. The new feature is that pattern-matching on <literal>MkSet</literal>
2286 (as in the definition of <literal>insert</literal>) makes <emphasis>available</emphasis> an <literal>(Eq a)</literal>
2287 context. In implementation terms, the <literal>MkSet</literal> constructor has a hidden field that stores
2288 the <literal>(Eq a)</literal> dictionary that is passed to <literal>MkSet</literal>; so
2289 when pattern-matching that dictionary becomes available for the right-hand side of the match.
2290 In the example, the equality dictionary is used to satisfy the equality constraint
2291 generated by the call to <literal>elem</literal>, so that the type of
2292 <literal>insert</literal> itself has no <literal>Eq</literal> constraint.
2295 For example, one possible application is to reify dictionaries:
2297 data NumInst a where
2298 MkNumInst :: Num a => NumInst a
2300 intInst :: NumInst Int
2303 plus :: NumInst a -> a -> a -> a
2304 plus MkNumInst p q = p + q
2306 Here, a value of type <literal>NumInst a</literal> is equivalent
2307 to an explicit <literal>(Num a)</literal> dictionary.
2310 All this applies to constructors declared using the syntax of <xref linkend="existential-with-context"/>.
2311 For example, the <literal>NumInst</literal> data type above could equivalently be declared
2315 = Num a => MkNumInst (NumInst a)
2317 Notice that, unlike the situation when declaring an existential, there is
2318 no <literal>forall</literal>, because the <literal>Num</literal> constrains the
2319 data type's universally quantified type variable <literal>a</literal>.
2320 A constructor may have both universal and existential type variables: for example,
2321 the following two declarations are equivalent:
2324 = forall b. (Num a, Eq b) => MkT1 a b
2326 MkT2 :: (Num a, Eq b) => a -> b -> T2 a
2329 <para>All this behaviour contrasts with Haskell 98's peculiar treatment of
2330 contexts on a data type declaration (Section 4.2.1 of the Haskell 98 Report).
2331 In Haskell 98 the definition
2333 data Eq a => Set' a = MkSet' [a]
2335 gives <literal>MkSet'</literal> the same type as <literal>MkSet</literal> above. But instead of
2336 <emphasis>making available</emphasis> an <literal>(Eq a)</literal> constraint, pattern-matching
2337 on <literal>MkSet'</literal> <emphasis>requires</emphasis> an <literal>(Eq a)</literal> constraint!
2338 GHC faithfully implements this behaviour, odd though it is. But for GADT-style declarations,
2339 GHC's behaviour is much more useful, as well as much more intuitive.
2343 The rest of this section gives further details about GADT-style data
2348 The result type of each data constructor must begin with the type constructor being defined.
2349 If the result type of all constructors
2350 has the form <literal>T a1 ... an</literal>, where <literal>a1 ... an</literal>
2351 are distinct type variables, then the data type is <emphasis>ordinary</emphasis>;
2352 otherwise is a <emphasis>generalised</emphasis> data type (<xref linkend="gadt"/>).
2356 The type signature of
2357 each constructor is independent, and is implicitly universally quantified as usual.
2358 Different constructors may have different universally-quantified type variables
2359 and different type-class constraints.
2360 For example, this is fine:
2363 T1 :: Eq b => b -> T b
2364 T2 :: (Show c, Ix c) => c -> [c] -> T c
2369 Unlike a Haskell-98-style
2370 data type declaration, the type variable(s) in the "<literal>data Set a where</literal>" header
2371 have no scope. Indeed, one can write a kind signature instead:
2373 data Set :: * -> * where ...
2375 or even a mixture of the two:
2377 data Foo a :: (* -> *) -> * where ...
2379 The type variables (if given) may be explicitly kinded, so we could also write the header for <literal>Foo</literal>
2382 data Foo a (b :: * -> *) where ...
2388 You can use strictness annotations, in the obvious places
2389 in the constructor type:
2392 Lit :: !Int -> Term Int
2393 If :: Term Bool -> !(Term a) -> !(Term a) -> Term a
2394 Pair :: Term a -> Term b -> Term (a,b)
2399 You can use a <literal>deriving</literal> clause on a GADT-style data type
2400 declaration. For example, these two declarations are equivalent
2402 data Maybe1 a where {
2403 Nothing1 :: Maybe1 a ;
2404 Just1 :: a -> Maybe1 a
2405 } deriving( Eq, Ord )
2407 data Maybe2 a = Nothing2 | Just2 a
2413 You can use record syntax on a GADT-style data type declaration:
2417 Adult { name :: String, children :: [Person] } :: Person
2418 Child { name :: String } :: Person
2420 As usual, for every constructor that has a field <literal>f</literal>, the type of
2421 field <literal>f</literal> must be the same (modulo alpha conversion).
2424 At the moment, record updates are not yet possible with GADT-style declarations,
2425 so support is limited to record construction, selection and pattern matching.
2428 aPerson = Adult { name = "Fred", children = [] }
2430 shortName :: Person -> Bool
2431 hasChildren (Adult { children = kids }) = not (null kids)
2432 hasChildren (Child {}) = False
2437 As in the case of existentials declared using the Haskell-98-like record syntax
2438 (<xref linkend="existential-records"/>),
2439 record-selector functions are generated only for those fields that have well-typed
2441 Here is the example of that section, in GADT-style syntax:
2443 data Counter a where
2444 NewCounter { _this :: self
2445 , _inc :: self -> self
2446 , _display :: self -> IO ()
2451 As before, only one selector function is generated here, that for <literal>tag</literal>.
2452 Nevertheless, you can still use all the field names in pattern matching and record construction.
2454 </itemizedlist></para>
2458 <title>Generalised Algebraic Data Types (GADTs)</title>
2460 <para>Generalised Algebraic Data Types generalise ordinary algebraic data types
2461 by allowing constructors to have richer return types. Here is an example:
2464 Lit :: Int -> Term Int
2465 Succ :: Term Int -> Term Int
2466 IsZero :: Term Int -> Term Bool
2467 If :: Term Bool -> Term a -> Term a -> Term a
2468 Pair :: Term a -> Term b -> Term (a,b)
2470 Notice that the return type of the constructors is not always <literal>Term a</literal>, as is the
2471 case with ordinary data types. This generality allows us to
2472 write a well-typed <literal>eval</literal> function
2473 for these <literal>Terms</literal>:
2477 eval (Succ t) = 1 + eval t
2478 eval (IsZero t) = eval t == 0
2479 eval (If b e1 e2) = if eval b then eval e1 else eval e2
2480 eval (Pair e1 e2) = (eval e1, eval e2)
2482 The key point about GADTs is that <emphasis>pattern matching causes type refinement</emphasis>.
2483 For example, in the right hand side of the equation
2488 the type <literal>a</literal> is refined to <literal>Int</literal>. That's the whole point!
2489 A precise specification of the type rules is beyond what this user manual aspires to,
2490 but the design closely follows that described in
2492 url="http://research.microsoft.com/%7Esimonpj/papers/gadt/">Simple
2493 unification-based type inference for GADTs</ulink>,
2495 The general principle is this: <emphasis>type refinement is only carried out
2496 based on user-supplied type annotations</emphasis>.
2497 So if no type signature is supplied for <literal>eval</literal>, no type refinement happens,
2498 and lots of obscure error messages will
2499 occur. However, the refinement is quite general. For example, if we had:
2501 eval :: Term a -> a -> a
2502 eval (Lit i) j = i+j
2504 the pattern match causes the type <literal>a</literal> to be refined to <literal>Int</literal> (because of the type
2505 of the constructor <literal>Lit</literal>), and that refinement also applies to the type of <literal>j</literal>, and
2506 the result type of the <literal>case</literal> expression. Hence the addition <literal>i+j</literal> is legal.
2509 These and many other examples are given in papers by Hongwei Xi, and
2510 Tim Sheard. There is a longer introduction
2511 <ulink url="http://www.haskell.org/haskellwiki/GADT">on the wiki</ulink>,
2513 <ulink url="http://www.informatik.uni-bonn.de/~ralf/publications/With.pdf">Fun with phantom types</ulink> also has a number of examples. Note that papers
2514 may use different notation to that implemented in GHC.
2517 The rest of this section outlines the extensions to GHC that support GADTs. The extension is enabled with
2518 <option>-XGADTs</option>. The <option>-XGADTs</option> flag also sets <option>-XRelaxedPolyRec</option>.
2521 A GADT can only be declared using GADT-style syntax (<xref linkend="gadt-style"/>);
2522 the old Haskell-98 syntax for data declarations always declares an ordinary data type.
2523 The result type of each constructor must begin with the type constructor being defined,
2524 but for a GADT the arguments to the type constructor can be arbitrary monotypes.
2525 For example, in the <literal>Term</literal> data
2526 type above, the type of each constructor must end with <literal>Term ty</literal>, but
2527 the <literal>ty</literal> need not be a type variable (e.g. the <literal>Lit</literal>
2532 It's is permitted to declare an ordinary algebraic data type using GADT-style syntax.
2533 What makes a GADT into a GADT is not the syntax, but rather the presence of data constructors
2534 whose result type is not just <literal>T a b</literal>.
2538 You cannot use a <literal>deriving</literal> clause for a GADT; only for
2539 an ordinary data type.
2543 As mentioned in <xref linkend="gadt-style"/>, record syntax is supported.
2547 Lit { val :: Int } :: Term Int
2548 Succ { num :: Term Int } :: Term Int
2549 Pred { num :: Term Int } :: Term Int
2550 IsZero { arg :: Term Int } :: Term Bool
2551 Pair { arg1 :: Term a
2554 If { cnd :: Term Bool
2559 However, for GADTs there is the following additional constraint:
2560 every constructor that has a field <literal>f</literal> must have
2561 the same result type (modulo alpha conversion)
2562 Hence, in the above example, we cannot merge the <literal>num</literal>
2563 and <literal>arg</literal> fields above into a
2564 single name. Although their field types are both <literal>Term Int</literal>,
2565 their selector functions actually have different types:
2568 num :: Term Int -> Term Int
2569 arg :: Term Bool -> Term Int
2574 When pattern-matching against data constructors drawn from a GADT,
2575 for example in a <literal>case</literal> expression, the following rules apply:
2577 <listitem><para>The type of the scrutinee must be rigid.</para></listitem>
2578 <listitem><para>The type of the entire <literal>case</literal> expression must be rigid.</para></listitem>
2579 <listitem><para>The type of any free variable mentioned in any of
2580 the <literal>case</literal> alternatives must be rigid.</para></listitem>
2582 A type is "rigid" if it is completely known to the compiler at its binding site. The easiest
2583 way to ensure that a variable a rigid type is to give it a type signature.
2584 For more precise details see <ulink url="http://research.microsoft.com/%7Esimonpj/papers/gadt">
2585 Simple unification-based type inference for GADTs
2586 </ulink>. The criteria implemented by GHC are given in the Appendix.
2596 <!-- ====================== End of Generalised algebraic data types ======================= -->
2598 <sect1 id="deriving">
2599 <title>Extensions to the "deriving" mechanism</title>
2601 <sect2 id="deriving-inferred">
2602 <title>Inferred context for deriving clauses</title>
2605 The Haskell Report is vague about exactly when a <literal>deriving</literal> clause is
2608 data T0 f a = MkT0 a deriving( Eq )
2609 data T1 f a = MkT1 (f a) deriving( Eq )
2610 data T2 f a = MkT2 (f (f a)) deriving( Eq )
2612 The natural generated <literal>Eq</literal> code would result in these instance declarations:
2614 instance Eq a => Eq (T0 f a) where ...
2615 instance Eq (f a) => Eq (T1 f a) where ...
2616 instance Eq (f (f a)) => Eq (T2 f a) where ...
2618 The first of these is obviously fine. The second is still fine, although less obviously.
2619 The third is not Haskell 98, and risks losing termination of instances.
2622 GHC takes a conservative position: it accepts the first two, but not the third. The rule is this:
2623 each constraint in the inferred instance context must consist only of type variables,
2624 with no repetitions.
2627 This rule is applied regardless of flags. If you want a more exotic context, you can write
2628 it yourself, using the <link linkend="stand-alone-deriving">standalone deriving mechanism</link>.
2632 <sect2 id="stand-alone-deriving">
2633 <title>Stand-alone deriving declarations</title>
2636 GHC now allows stand-alone <literal>deriving</literal> declarations, enabled by <literal>-XStandaloneDeriving</literal>:
2638 data Foo a = Bar a | Baz String
2640 deriving instance Eq a => Eq (Foo a)
2642 The syntax is identical to that of an ordinary instance declaration apart from (a) the keyword
2643 <literal>deriving</literal>, and (b) the absence of the <literal>where</literal> part.
2644 You must supply a context (in the example the context is <literal>(Eq a)</literal>),
2645 exactly as you would in an ordinary instance declaration.
2646 (In contrast the context is inferred in a <literal>deriving</literal> clause
2647 attached to a data type declaration.)
2649 A <literal>deriving instance</literal> declaration
2650 must obey the same rules concerning form and termination as ordinary instance declarations,
2651 controlled by the same flags; see <xref linkend="instance-decls"/>.
2654 Unlike a <literal>deriving</literal>
2655 declaration attached to a <literal>data</literal> declaration, the instance can be more specific
2656 than the data type (assuming you also use
2657 <literal>-XFlexibleInstances</literal>, <xref linkend="instance-rules"/>). Consider
2660 data Foo a = Bar a | Baz String
2662 deriving instance Eq a => Eq (Foo [a])
2663 deriving instance Eq a => Eq (Foo (Maybe a))
2665 This will generate a derived instance for <literal>(Foo [a])</literal> and <literal>(Foo (Maybe a))</literal>,
2666 but other types such as <literal>(Foo (Int,Bool))</literal> will not be an instance of <literal>Eq</literal>.
2669 <para>The stand-alone syntax is generalised for newtypes in exactly the same
2670 way that ordinary <literal>deriving</literal> clauses are generalised (<xref linkend="newtype-deriving"/>).
2673 newtype Foo a = MkFoo (State Int a)
2675 deriving instance MonadState Int Foo
2677 GHC always treats the <emphasis>last</emphasis> parameter of the instance
2678 (<literal>Foo</literal> in this example) as the type whose instance is being derived.
2684 <sect2 id="deriving-typeable">
2685 <title>Deriving clause for classes <literal>Typeable</literal> and <literal>Data</literal></title>
2688 Haskell 98 allows the programmer to add "<literal>deriving( Eq, Ord )</literal>" to a data type
2689 declaration, to generate a standard instance declaration for classes specified in the <literal>deriving</literal> clause.
2690 In Haskell 98, the only classes that may appear in the <literal>deriving</literal> clause are the standard
2691 classes <literal>Eq</literal>, <literal>Ord</literal>,
2692 <literal>Enum</literal>, <literal>Ix</literal>, <literal>Bounded</literal>, <literal>Read</literal>, and <literal>Show</literal>.
2695 GHC extends this list with two more classes that may be automatically derived
2696 (provided the <option>-XDeriveDataTypeable</option> flag is specified):
2697 <literal>Typeable</literal>, and <literal>Data</literal>. These classes are defined in the library
2698 modules <literal>Data.Typeable</literal> and <literal>Data.Generics</literal> respectively, and the
2699 appropriate class must be in scope before it can be mentioned in the <literal>deriving</literal> clause.
2701 <para>An instance of <literal>Typeable</literal> can only be derived if the
2702 data type has seven or fewer type parameters, all of kind <literal>*</literal>.
2703 The reason for this is that the <literal>Typeable</literal> class is derived using the scheme
2705 <ulink url="http://research.microsoft.com/%7Esimonpj/papers/hmap/gmap2.ps">
2706 Scrap More Boilerplate: Reflection, Zips, and Generalised Casts
2708 (Section 7.4 of the paper describes the multiple <literal>Typeable</literal> classes that
2709 are used, and only <literal>Typeable1</literal> up to
2710 <literal>Typeable7</literal> are provided in the library.)
2711 In other cases, there is nothing to stop the programmer writing a <literal>TypableX</literal>
2712 class, whose kind suits that of the data type constructor, and
2713 then writing the data type instance by hand.
2717 <sect2 id="newtype-deriving">
2718 <title>Generalised derived instances for newtypes</title>
2721 When you define an abstract type using <literal>newtype</literal>, you may want
2722 the new type to inherit some instances from its representation. In
2723 Haskell 98, you can inherit instances of <literal>Eq</literal>, <literal>Ord</literal>,
2724 <literal>Enum</literal> and <literal>Bounded</literal> by deriving them, but for any
2725 other classes you have to write an explicit instance declaration. For
2726 example, if you define
2729 newtype Dollars = Dollars Int
2732 and you want to use arithmetic on <literal>Dollars</literal>, you have to
2733 explicitly define an instance of <literal>Num</literal>:
2736 instance Num Dollars where
2737 Dollars a + Dollars b = Dollars (a+b)
2740 All the instance does is apply and remove the <literal>newtype</literal>
2741 constructor. It is particularly galling that, since the constructor
2742 doesn't appear at run-time, this instance declaration defines a
2743 dictionary which is <emphasis>wholly equivalent</emphasis> to the <literal>Int</literal>
2744 dictionary, only slower!
2748 <sect3> <title> Generalising the deriving clause </title>
2750 GHC now permits such instances to be derived instead,
2751 using the flag <option>-XGeneralizedNewtypeDeriving</option>,
2754 newtype Dollars = Dollars Int deriving (Eq,Show,Num)
2757 and the implementation uses the <emphasis>same</emphasis> <literal>Num</literal> dictionary
2758 for <literal>Dollars</literal> as for <literal>Int</literal>. Notionally, the compiler
2759 derives an instance declaration of the form
2762 instance Num Int => Num Dollars
2765 which just adds or removes the <literal>newtype</literal> constructor according to the type.
2769 We can also derive instances of constructor classes in a similar
2770 way. For example, suppose we have implemented state and failure monad
2771 transformers, such that
2774 instance Monad m => Monad (State s m)
2775 instance Monad m => Monad (Failure m)
2777 In Haskell 98, we can define a parsing monad by
2779 type Parser tok m a = State [tok] (Failure m) a
2782 which is automatically a monad thanks to the instance declarations
2783 above. With the extension, we can make the parser type abstract,
2784 without needing to write an instance of class <literal>Monad</literal>, via
2787 newtype Parser tok m a = Parser (State [tok] (Failure m) a)
2790 In this case the derived instance declaration is of the form
2792 instance Monad (State [tok] (Failure m)) => Monad (Parser tok m)
2795 Notice that, since <literal>Monad</literal> is a constructor class, the
2796 instance is a <emphasis>partial application</emphasis> of the new type, not the
2797 entire left hand side. We can imagine that the type declaration is
2798 "eta-converted" to generate the context of the instance
2803 We can even derive instances of multi-parameter classes, provided the
2804 newtype is the last class parameter. In this case, a ``partial
2805 application'' of the class appears in the <literal>deriving</literal>
2806 clause. For example, given the class
2809 class StateMonad s m | m -> s where ...
2810 instance Monad m => StateMonad s (State s m) where ...
2812 then we can derive an instance of <literal>StateMonad</literal> for <literal>Parser</literal>s by
2814 newtype Parser tok m a = Parser (State [tok] (Failure m) a)
2815 deriving (Monad, StateMonad [tok])
2818 The derived instance is obtained by completing the application of the
2819 class to the new type:
2822 instance StateMonad [tok] (State [tok] (Failure m)) =>
2823 StateMonad [tok] (Parser tok m)
2828 As a result of this extension, all derived instances in newtype
2829 declarations are treated uniformly (and implemented just by reusing
2830 the dictionary for the representation type), <emphasis>except</emphasis>
2831 <literal>Show</literal> and <literal>Read</literal>, which really behave differently for
2832 the newtype and its representation.
2836 <sect3> <title> A more precise specification </title>
2838 Derived instance declarations are constructed as follows. Consider the
2839 declaration (after expansion of any type synonyms)
2842 newtype T v1...vn = T' (t vk+1...vn) deriving (c1...cm)
2848 The <literal>ci</literal> are partial applications of
2849 classes of the form <literal>C t1'...tj'</literal>, where the arity of <literal>C</literal>
2850 is exactly <literal>j+1</literal>. That is, <literal>C</literal> lacks exactly one type argument.
2853 The <literal>k</literal> is chosen so that <literal>ci (T v1...vk)</literal> is well-kinded.
2856 The type <literal>t</literal> is an arbitrary type.
2859 The type variables <literal>vk+1...vn</literal> do not occur in <literal>t</literal>,
2860 nor in the <literal>ci</literal>, and
2863 None of the <literal>ci</literal> is <literal>Read</literal>, <literal>Show</literal>,
2864 <literal>Typeable</literal>, or <literal>Data</literal>. These classes
2865 should not "look through" the type or its constructor. You can still
2866 derive these classes for a newtype, but it happens in the usual way, not
2867 via this new mechanism.
2870 Then, for each <literal>ci</literal>, the derived instance
2873 instance ci t => ci (T v1...vk)
2875 As an example which does <emphasis>not</emphasis> work, consider
2877 newtype NonMonad m s = NonMonad (State s m s) deriving Monad
2879 Here we cannot derive the instance
2881 instance Monad (State s m) => Monad (NonMonad m)
2884 because the type variable <literal>s</literal> occurs in <literal>State s m</literal>,
2885 and so cannot be "eta-converted" away. It is a good thing that this
2886 <literal>deriving</literal> clause is rejected, because <literal>NonMonad m</literal> is
2887 not, in fact, a monad --- for the same reason. Try defining
2888 <literal>>>=</literal> with the correct type: you won't be able to.
2892 Notice also that the <emphasis>order</emphasis> of class parameters becomes
2893 important, since we can only derive instances for the last one. If the
2894 <literal>StateMonad</literal> class above were instead defined as
2897 class StateMonad m s | m -> s where ...
2900 then we would not have been able to derive an instance for the
2901 <literal>Parser</literal> type above. We hypothesise that multi-parameter
2902 classes usually have one "main" parameter for which deriving new
2903 instances is most interesting.
2905 <para>Lastly, all of this applies only for classes other than
2906 <literal>Read</literal>, <literal>Show</literal>, <literal>Typeable</literal>,
2907 and <literal>Data</literal>, for which the built-in derivation applies (section
2908 4.3.3. of the Haskell Report).
2909 (For the standard classes <literal>Eq</literal>, <literal>Ord</literal>,
2910 <literal>Ix</literal>, and <literal>Bounded</literal> it is immaterial whether
2911 the standard method is used or the one described here.)
2918 <!-- TYPE SYSTEM EXTENSIONS -->
2919 <sect1 id="type-class-extensions">
2920 <title>Class and instances declarations</title>
2922 <sect2 id="multi-param-type-classes">
2923 <title>Class declarations</title>
2926 This section, and the next one, documents GHC's type-class extensions.
2927 There's lots of background in the paper <ulink
2928 url="http://research.microsoft.com/~simonpj/Papers/type-class-design-space/">Type
2929 classes: exploring the design space</ulink> (Simon Peyton Jones, Mark
2930 Jones, Erik Meijer).
2933 All the extensions are enabled by the <option>-fglasgow-exts</option> flag.
2937 <title>Multi-parameter type classes</title>
2939 Multi-parameter type classes are permitted. For example:
2943 class Collection c a where
2944 union :: c a -> c a -> c a
2952 <title>The superclasses of a class declaration</title>
2955 There are no restrictions on the context in a class declaration
2956 (which introduces superclasses), except that the class hierarchy must
2957 be acyclic. So these class declarations are OK:
2961 class Functor (m k) => FiniteMap m k where
2964 class (Monad m, Monad (t m)) => Transform t m where
2965 lift :: m a -> (t m) a
2971 As in Haskell 98, The class hierarchy must be acyclic. However, the definition
2972 of "acyclic" involves only the superclass relationships. For example,
2978 op :: D b => a -> b -> b
2981 class C a => D a where { ... }
2985 Here, <literal>C</literal> is a superclass of <literal>D</literal>, but it's OK for a
2986 class operation <literal>op</literal> of <literal>C</literal> to mention <literal>D</literal>. (It
2987 would not be OK for <literal>D</literal> to be a superclass of <literal>C</literal>.)
2994 <sect3 id="class-method-types">
2995 <title>Class method types</title>
2998 Haskell 98 prohibits class method types to mention constraints on the
2999 class type variable, thus:
3002 fromList :: [a] -> s a
3003 elem :: Eq a => a -> s a -> Bool
3005 The type of <literal>elem</literal> is illegal in Haskell 98, because it
3006 contains the constraint <literal>Eq a</literal>, constrains only the
3007 class type variable (in this case <literal>a</literal>).
3008 GHC lifts this restriction (flag <option>-XConstrainedClassMethods</option>).
3015 <sect2 id="functional-dependencies">
3016 <title>Functional dependencies
3019 <para> Functional dependencies are implemented as described by Mark Jones
3020 in “<ulink url="http://citeseer.ist.psu.edu/jones00type.html">Type Classes with Functional Dependencies</ulink>”, Mark P. Jones,
3021 In Proceedings of the 9th European Symposium on Programming,
3022 ESOP 2000, Berlin, Germany, March 2000, Springer-Verlag LNCS 1782,
3026 Functional dependencies are introduced by a vertical bar in the syntax of a
3027 class declaration; e.g.
3029 class (Monad m) => MonadState s m | m -> s where ...
3031 class Foo a b c | a b -> c where ...
3033 There should be more documentation, but there isn't (yet). Yell if you need it.
3036 <sect3><title>Rules for functional dependencies </title>
3038 In a class declaration, all of the class type variables must be reachable (in the sense
3039 mentioned in <xref linkend="type-restrictions"/>)
3040 from the free variables of each method type.
3044 class Coll s a where
3046 insert :: s -> a -> s
3049 is not OK, because the type of <literal>empty</literal> doesn't mention
3050 <literal>a</literal>. Functional dependencies can make the type variable
3053 class Coll s a | s -> a where
3055 insert :: s -> a -> s
3058 Alternatively <literal>Coll</literal> might be rewritten
3061 class Coll s a where
3063 insert :: s a -> a -> s a
3067 which makes the connection between the type of a collection of
3068 <literal>a</literal>'s (namely <literal>(s a)</literal>) and the element type <literal>a</literal>.
3069 Occasionally this really doesn't work, in which case you can split the
3077 class CollE s => Coll s a where
3078 insert :: s -> a -> s
3085 <title>Background on functional dependencies</title>
3087 <para>The following description of the motivation and use of functional dependencies is taken
3088 from the Hugs user manual, reproduced here (with minor changes) by kind
3089 permission of Mark Jones.
3092 Consider the following class, intended as part of a
3093 library for collection types:
3095 class Collects e ce where
3097 insert :: e -> ce -> ce
3098 member :: e -> ce -> Bool
3100 The type variable e used here represents the element type, while ce is the type
3101 of the container itself. Within this framework, we might want to define
3102 instances of this class for lists or characteristic functions (both of which
3103 can be used to represent collections of any equality type), bit sets (which can
3104 be used to represent collections of characters), or hash tables (which can be
3105 used to represent any collection whose elements have a hash function). Omitting
3106 standard implementation details, this would lead to the following declarations:
3108 instance Eq e => Collects e [e] where ...
3109 instance Eq e => Collects e (e -> Bool) where ...
3110 instance Collects Char BitSet where ...
3111 instance (Hashable e, Collects a ce)
3112 => Collects e (Array Int ce) where ...
3114 All this looks quite promising; we have a class and a range of interesting
3115 implementations. Unfortunately, there are some serious problems with the class
3116 declaration. First, the empty function has an ambiguous type:
3118 empty :: Collects e ce => ce
3120 By "ambiguous" we mean that there is a type variable e that appears on the left
3121 of the <literal>=></literal> symbol, but not on the right. The problem with
3122 this is that, according to the theoretical foundations of Haskell overloading,
3123 we cannot guarantee a well-defined semantics for any term with an ambiguous
3127 We can sidestep this specific problem by removing the empty member from the
3128 class declaration. However, although the remaining members, insert and member,
3129 do not have ambiguous types, we still run into problems when we try to use
3130 them. For example, consider the following two functions:
3132 f x y = insert x . insert y
3135 for which GHC infers the following types:
3137 f :: (Collects a c, Collects b c) => a -> b -> c -> c
3138 g :: (Collects Bool c, Collects Char c) => c -> c
3140 Notice that the type for f allows the two parameters x and y to be assigned
3141 different types, even though it attempts to insert each of the two values, one
3142 after the other, into the same collection. If we're trying to model collections
3143 that contain only one type of value, then this is clearly an inaccurate
3144 type. Worse still, the definition for g is accepted, without causing a type
3145 error. As a result, the error in this code will not be flagged at the point
3146 where it appears. Instead, it will show up only when we try to use g, which
3147 might even be in a different module.
3150 <sect4><title>An attempt to use constructor classes</title>
3153 Faced with the problems described above, some Haskell programmers might be
3154 tempted to use something like the following version of the class declaration:
3156 class Collects e c where
3158 insert :: e -> c e -> c e
3159 member :: e -> c e -> Bool
3161 The key difference here is that we abstract over the type constructor c that is
3162 used to form the collection type c e, and not over that collection type itself,
3163 represented by ce in the original class declaration. This avoids the immediate
3164 problems that we mentioned above: empty has type <literal>Collects e c => c
3165 e</literal>, which is not ambiguous.
3168 The function f from the previous section has a more accurate type:
3170 f :: (Collects e c) => e -> e -> c e -> c e
3172 The function g from the previous section is now rejected with a type error as
3173 we would hope because the type of f does not allow the two arguments to have
3175 This, then, is an example of a multiple parameter class that does actually work
3176 quite well in practice, without ambiguity problems.
3177 There is, however, a catch. This version of the Collects class is nowhere near
3178 as general as the original class seemed to be: only one of the four instances
3179 for <literal>Collects</literal>
3180 given above can be used with this version of Collects because only one of
3181 them---the instance for lists---has a collection type that can be written in
3182 the form c e, for some type constructor c, and element type e.
3186 <sect4><title>Adding functional dependencies</title>
3189 To get a more useful version of the Collects class, Hugs provides a mechanism
3190 that allows programmers to specify dependencies between the parameters of a
3191 multiple parameter class (For readers with an interest in theoretical
3192 foundations and previous work: The use of dependency information can be seen
3193 both as a generalization of the proposal for `parametric type classes' that was
3194 put forward by Chen, Hudak, and Odersky, or as a special case of Mark Jones's
3195 later framework for "improvement" of qualified types. The
3196 underlying ideas are also discussed in a more theoretical and abstract setting
3197 in a manuscript [implparam], where they are identified as one point in a
3198 general design space for systems of implicit parameterization.).
3200 To start with an abstract example, consider a declaration such as:
3202 class C a b where ...
3204 which tells us simply that C can be thought of as a binary relation on types
3205 (or type constructors, depending on the kinds of a and b). Extra clauses can be
3206 included in the definition of classes to add information about dependencies
3207 between parameters, as in the following examples:
3209 class D a b | a -> b where ...
3210 class E a b | a -> b, b -> a where ...
3212 The notation <literal>a -> b</literal> used here between the | and where
3213 symbols --- not to be
3214 confused with a function type --- indicates that the a parameter uniquely
3215 determines the b parameter, and might be read as "a determines b." Thus D is
3216 not just a relation, but actually a (partial) function. Similarly, from the two
3217 dependencies that are included in the definition of E, we can see that E
3218 represents a (partial) one-one mapping between types.
3221 More generally, dependencies take the form <literal>x1 ... xn -> y1 ... ym</literal>,
3222 where x1, ..., xn, and y1, ..., yn are type variables with n>0 and
3223 m>=0, meaning that the y parameters are uniquely determined by the x
3224 parameters. Spaces can be used as separators if more than one variable appears
3225 on any single side of a dependency, as in <literal>t -> a b</literal>. Note that a class may be
3226 annotated with multiple dependencies using commas as separators, as in the
3227 definition of E above. Some dependencies that we can write in this notation are
3228 redundant, and will be rejected because they don't serve any useful
3229 purpose, and may instead indicate an error in the program. Examples of
3230 dependencies like this include <literal>a -> a </literal>,
3231 <literal>a -> a a </literal>,
3232 <literal>a -> </literal>, etc. There can also be
3233 some redundancy if multiple dependencies are given, as in
3234 <literal>a->b</literal>,
3235 <literal>b->c </literal>, <literal>a->c </literal>, and
3236 in which some subset implies the remaining dependencies. Examples like this are
3237 not treated as errors. Note that dependencies appear only in class
3238 declarations, and not in any other part of the language. In particular, the
3239 syntax for instance declarations, class constraints, and types is completely
3243 By including dependencies in a class declaration, we provide a mechanism for
3244 the programmer to specify each multiple parameter class more precisely. The
3245 compiler, on the other hand, is responsible for ensuring that the set of
3246 instances that are in scope at any given point in the program is consistent
3247 with any declared dependencies. For example, the following pair of instance
3248 declarations cannot appear together in the same scope because they violate the
3249 dependency for D, even though either one on its own would be acceptable:
3251 instance D Bool Int where ...
3252 instance D Bool Char where ...
3254 Note also that the following declaration is not allowed, even by itself:
3256 instance D [a] b where ...
3258 The problem here is that this instance would allow one particular choice of [a]
3259 to be associated with more than one choice for b, which contradicts the
3260 dependency specified in the definition of D. More generally, this means that,
3261 in any instance of the form:
3263 instance D t s where ...
3265 for some particular types t and s, the only variables that can appear in s are
3266 the ones that appear in t, and hence, if the type t is known, then s will be
3267 uniquely determined.
3270 The benefit of including dependency information is that it allows us to define
3271 more general multiple parameter classes, without ambiguity problems, and with
3272 the benefit of more accurate types. To illustrate this, we return to the
3273 collection class example, and annotate the original definition of <literal>Collects</literal>
3274 with a simple dependency:
3276 class Collects e ce | ce -> e where
3278 insert :: e -> ce -> ce
3279 member :: e -> ce -> Bool
3281 The dependency <literal>ce -> e</literal> here specifies that the type e of elements is uniquely
3282 determined by the type of the collection ce. Note that both parameters of
3283 Collects are of kind *; there are no constructor classes here. Note too that
3284 all of the instances of Collects that we gave earlier can be used
3285 together with this new definition.
3288 What about the ambiguity problems that we encountered with the original
3289 definition? The empty function still has type Collects e ce => ce, but it is no
3290 longer necessary to regard that as an ambiguous type: Although the variable e
3291 does not appear on the right of the => symbol, the dependency for class
3292 Collects tells us that it is uniquely determined by ce, which does appear on
3293 the right of the => symbol. Hence the context in which empty is used can still
3294 give enough information to determine types for both ce and e, without
3295 ambiguity. More generally, we need only regard a type as ambiguous if it
3296 contains a variable on the left of the => that is not uniquely determined
3297 (either directly or indirectly) by the variables on the right.
3300 Dependencies also help to produce more accurate types for user defined
3301 functions, and hence to provide earlier detection of errors, and less cluttered
3302 types for programmers to work with. Recall the previous definition for a
3305 f x y = insert x y = insert x . insert y
3307 for which we originally obtained a type:
3309 f :: (Collects a c, Collects b c) => a -> b -> c -> c
3311 Given the dependency information that we have for Collects, however, we can
3312 deduce that a and b must be equal because they both appear as the second
3313 parameter in a Collects constraint with the same first parameter c. Hence we
3314 can infer a shorter and more accurate type for f:
3316 f :: (Collects a c) => a -> a -> c -> c
3318 In a similar way, the earlier definition of g will now be flagged as a type error.
3321 Although we have given only a few examples here, it should be clear that the
3322 addition of dependency information can help to make multiple parameter classes
3323 more useful in practice, avoiding ambiguity problems, and allowing more general
3324 sets of instance declarations.
3330 <sect2 id="instance-decls">
3331 <title>Instance declarations</title>
3333 <para>An instance declaration has the form
3335 instance ( <replaceable>assertion</replaceable><subscript>1</subscript>, ..., <replaceable>assertion</replaceable><subscript>n</subscript>) => <replaceable>class</replaceable> <replaceable>type</replaceable><subscript>1</subscript> ... <replaceable>type</replaceable><subscript>m</subscript> where ...
3337 The part before the "<literal>=></literal>" is the
3338 <emphasis>context</emphasis>, while the part after the
3339 "<literal>=></literal>" is the <emphasis>head</emphasis> of the instance declaration.
3342 <sect3 id="flexible-instance-head">
3343 <title>Relaxed rules for the instance head</title>
3346 In Haskell 98 the head of an instance declaration
3347 must be of the form <literal>C (T a1 ... an)</literal>, where
3348 <literal>C</literal> is the class, <literal>T</literal> is a data type constructor,
3349 and the <literal>a1 ... an</literal> are distinct type variables.
3350 GHC relaxes these rules in two ways.
3354 The <option>-XFlexibleInstances</option> flag allows the head of the instance
3355 declaration to mention arbitrary nested types.
3356 For example, this becomes a legal instance declaration
3358 instance C (Maybe Int) where ...
3360 See also the <link linkend="instance-overlap">rules on overlap</link>.
3363 With the <option>-XTypeSynonymInstances</option> flag, instance heads may use type
3364 synonyms. As always, using a type synonym is just shorthand for
3365 writing the RHS of the type synonym definition. For example:
3369 type Point = (Int,Int)
3370 instance C Point where ...
3371 instance C [Point] where ...
3375 is legal. However, if you added
3379 instance C (Int,Int) where ...
3383 as well, then the compiler will complain about the overlapping
3384 (actually, identical) instance declarations. As always, type synonyms
3385 must be fully applied. You cannot, for example, write:
3389 instance Monad P where ...
3397 <sect3 id="instance-rules">
3398 <title>Relaxed rules for instance contexts</title>
3400 <para>In Haskell 98, the assertions in the context of the instance declaration
3401 must be of the form <literal>C a</literal> where <literal>a</literal>
3402 is a type variable that occurs in the head.
3406 The <option>-XFlexibleContexts</option> flag relaxes this rule, as well
3407 as the corresponding rule for type signatures (see <xref linkend="flexible-contexts"/>).
3408 With this flag the context of the instance declaration can each consist of arbitrary
3409 (well-kinded) assertions <literal>(C t1 ... tn)</literal> subject only to the
3413 The Paterson Conditions: for each assertion in the context
3415 <listitem><para>No type variable has more occurrences in the assertion than in the head</para></listitem>
3416 <listitem><para>The assertion has fewer constructors and variables (taken together
3417 and counting repetitions) than the head</para></listitem>
3421 <listitem><para>The Coverage Condition. For each functional dependency,
3422 <replaceable>tvs</replaceable><subscript>left</subscript> <literal>-></literal>
3423 <replaceable>tvs</replaceable><subscript>right</subscript>, of the class,
3424 every type variable in
3425 S(<replaceable>tvs</replaceable><subscript>right</subscript>) must appear in
3426 S(<replaceable>tvs</replaceable><subscript>left</subscript>), where S is the
3427 substitution mapping each type variable in the class declaration to the
3428 corresponding type in the instance declaration.
3431 These restrictions ensure that context reduction terminates: each reduction
3432 step makes the problem smaller by at least one
3433 constructor. Both the Paterson Conditions and the Coverage Condition are lifted
3434 if you give the <option>-XUndecidableInstances</option>
3435 flag (<xref linkend="undecidable-instances"/>).
3436 You can find lots of background material about the reason for these
3437 restrictions in the paper <ulink
3438 url="http://research.microsoft.com/%7Esimonpj/papers/fd%2Dchr/">
3439 Understanding functional dependencies via Constraint Handling Rules</ulink>.
3442 For example, these are OK:
3444 instance C Int [a] -- Multiple parameters
3445 instance Eq (S [a]) -- Structured type in head
3447 -- Repeated type variable in head
3448 instance C4 a a => C4 [a] [a]
3449 instance Stateful (ST s) (MutVar s)
3451 -- Head can consist of type variables only
3453 instance (Eq a, Show b) => C2 a b
3455 -- Non-type variables in context
3456 instance Show (s a) => Show (Sized s a)
3457 instance C2 Int a => C3 Bool [a]
3458 instance C2 Int a => C3 [a] b
3462 -- Context assertion no smaller than head
3463 instance C a => C a where ...
3464 -- (C b b) has more more occurrences of b than the head
3465 instance C b b => Foo [b] where ...
3470 The same restrictions apply to instances generated by
3471 <literal>deriving</literal> clauses. Thus the following is accepted:
3473 data MinHeap h a = H a (h a)
3476 because the derived instance
3478 instance (Show a, Show (h a)) => Show (MinHeap h a)
3480 conforms to the above rules.
3484 A useful idiom permitted by the above rules is as follows.
3485 If one allows overlapping instance declarations then it's quite
3486 convenient to have a "default instance" declaration that applies if
3487 something more specific does not:
3495 <sect3 id="undecidable-instances">
3496 <title>Undecidable instances</title>
3499 Sometimes even the rules of <xref linkend="instance-rules"/> are too onerous.
3500 For example, sometimes you might want to use the following to get the
3501 effect of a "class synonym":
3503 class (C1 a, C2 a, C3 a) => C a where { }
3505 instance (C1 a, C2 a, C3 a) => C a where { }
3507 This allows you to write shorter signatures:
3513 f :: (C1 a, C2 a, C3 a) => ...
3515 The restrictions on functional dependencies (<xref
3516 linkend="functional-dependencies"/>) are particularly troublesome.
3517 It is tempting to introduce type variables in the context that do not appear in
3518 the head, something that is excluded by the normal rules. For example:
3520 class HasConverter a b | a -> b where
3523 data Foo a = MkFoo a
3525 instance (HasConverter a b,Show b) => Show (Foo a) where
3526 show (MkFoo value) = show (convert value)
3528 This is dangerous territory, however. Here, for example, is a program that would make the
3533 instance F [a] [[a]]
3534 instance (D c, F a c) => D [a] -- 'c' is not mentioned in the head
3536 Similarly, it can be tempting to lift the coverage condition:
3538 class Mul a b c | a b -> c where
3539 (.*.) :: a -> b -> c
3541 instance Mul Int Int Int where (.*.) = (*)
3542 instance Mul Int Float Float where x .*. y = fromIntegral x * y
3543 instance Mul a b c => Mul a [b] [c] where x .*. v = map (x.*.) v
3545 The third instance declaration does not obey the coverage condition;
3546 and indeed the (somewhat strange) definition:
3548 f = \ b x y -> if b then x .*. [y] else y
3550 makes instance inference go into a loop, because it requires the constraint
3551 <literal>(Mul a [b] b)</literal>.
3554 Nevertheless, GHC allows you to experiment with more liberal rules. If you use
3555 the experimental flag <option>-XUndecidableInstances</option>
3556 <indexterm><primary>-XUndecidableInstances</primary></indexterm>,
3557 both the Paterson Conditions and the Coverage Condition
3558 (described in <xref linkend="instance-rules"/>) are lifted. Termination is ensured by having a
3559 fixed-depth recursion stack. If you exceed the stack depth you get a
3560 sort of backtrace, and the opportunity to increase the stack depth
3561 with <option>-fcontext-stack=</option><emphasis>N</emphasis>.
3567 <sect3 id="instance-overlap">
3568 <title>Overlapping instances</title>
3570 In general, <emphasis>GHC requires that that it be unambiguous which instance
3572 should be used to resolve a type-class constraint</emphasis>. This behaviour
3573 can be modified by two flags: <option>-XOverlappingInstances</option>
3574 <indexterm><primary>-XOverlappingInstances
3575 </primary></indexterm>
3576 and <option>-XIncoherentInstances</option>
3577 <indexterm><primary>-XIncoherentInstances
3578 </primary></indexterm>, as this section discusses. Both these
3579 flags are dynamic flags, and can be set on a per-module basis, using
3580 an <literal>OPTIONS_GHC</literal> pragma if desired (<xref linkend="source-file-options"/>).</para>
3582 When GHC tries to resolve, say, the constraint <literal>C Int Bool</literal>,
3583 it tries to match every instance declaration against the
3585 by instantiating the head of the instance declaration. For example, consider
3588 instance context1 => C Int a where ... -- (A)
3589 instance context2 => C a Bool where ... -- (B)
3590 instance context3 => C Int [a] where ... -- (C)
3591 instance context4 => C Int [Int] where ... -- (D)
3593 The instances (A) and (B) match the constraint <literal>C Int Bool</literal>,
3594 but (C) and (D) do not. When matching, GHC takes
3595 no account of the context of the instance declaration
3596 (<literal>context1</literal> etc).
3597 GHC's default behaviour is that <emphasis>exactly one instance must match the
3598 constraint it is trying to resolve</emphasis>.
3599 It is fine for there to be a <emphasis>potential</emphasis> of overlap (by
3600 including both declarations (A) and (B), say); an error is only reported if a
3601 particular constraint matches more than one.
3605 The <option>-XOverlappingInstances</option> flag instructs GHC to allow
3606 more than one instance to match, provided there is a most specific one. For
3607 example, the constraint <literal>C Int [Int]</literal> matches instances (A),
3608 (C) and (D), but the last is more specific, and hence is chosen. If there is no
3609 most-specific match, the program is rejected.
3612 However, GHC is conservative about committing to an overlapping instance. For example:
3617 Suppose that from the RHS of <literal>f</literal> we get the constraint
3618 <literal>C Int [b]</literal>. But
3619 GHC does not commit to instance (C), because in a particular
3620 call of <literal>f</literal>, <literal>b</literal> might be instantiate
3621 to <literal>Int</literal>, in which case instance (D) would be more specific still.
3622 So GHC rejects the program.
3623 (If you add the flag <option>-XIncoherentInstances</option>,
3624 GHC will instead pick (C), without complaining about
3625 the problem of subsequent instantiations.)
3628 Notice that we gave a type signature to <literal>f</literal>, so GHC had to
3629 <emphasis>check</emphasis> that <literal>f</literal> has the specified type.
3630 Suppose instead we do not give a type signature, asking GHC to <emphasis>infer</emphasis>
3631 it instead. In this case, GHC will refrain from
3632 simplifying the constraint <literal>C Int [b]</literal> (for the same reason
3633 as before) but, rather than rejecting the program, it will infer the type
3635 f :: C Int [b] => [b] -> [b]
3637 That postpones the question of which instance to pick to the
3638 call site for <literal>f</literal>
3639 by which time more is known about the type <literal>b</literal>.
3640 You can write this type signature yourself if you use the
3641 <link linkend="flexible-contexts"><option>-XFlexibleContexts</option></link>
3645 Exactly the same situation can arise in instance declarations themselves. Suppose we have
3649 instance Foo [b] where
3652 and, as before, the constraint <literal>C Int [b]</literal> arises from <literal>f</literal>'s
3653 right hand side. GHC will reject the instance, complaining as before that it does not know how to resolve
3654 the constraint <literal>C Int [b]</literal>, because it matches more than one instance
3655 declaration. The solution is to postpone the choice by adding the constraint to the context
3656 of the instance declaration, thus:
3658 instance C Int [b] => Foo [b] where
3661 (You need <link linkend="instance-rules"><option>-XFlexibleInstances</option></link> to do this.)
3664 The willingness to be overlapped or incoherent is a property of
3665 the <emphasis>instance declaration</emphasis> itself, controlled by the
3666 presence or otherwise of the <option>-XOverlappingInstances</option>
3667 and <option>-XIncoherentInstances</option> flags when that module is
3668 being defined. Neither flag is required in a module that imports and uses the
3669 instance declaration. Specifically, during the lookup process:
3672 An instance declaration is ignored during the lookup process if (a) a more specific
3673 match is found, and (b) the instance declaration was compiled with
3674 <option>-XOverlappingInstances</option>. The flag setting for the
3675 more-specific instance does not matter.
3678 Suppose an instance declaration does not match the constraint being looked up, but
3679 does unify with it, so that it might match when the constraint is further
3680 instantiated. Usually GHC will regard this as a reason for not committing to
3681 some other constraint. But if the instance declaration was compiled with
3682 <option>-XIncoherentInstances</option>, GHC will skip the "does-it-unify?"
3683 check for that declaration.
3686 These rules make it possible for a library author to design a library that relies on
3687 overlapping instances without the library client having to know.
3690 If an instance declaration is compiled without
3691 <option>-XOverlappingInstances</option>,
3692 then that instance can never be overlapped. This could perhaps be
3693 inconvenient. Perhaps the rule should instead say that the
3694 <emphasis>overlapping</emphasis> instance declaration should be compiled in
3695 this way, rather than the <emphasis>overlapped</emphasis> one. Perhaps overlap
3696 at a usage site should be permitted regardless of how the instance declarations
3697 are compiled, if the <option>-XOverlappingInstances</option> flag is
3698 used at the usage site. (Mind you, the exact usage site can occasionally be
3699 hard to pin down.) We are interested to receive feedback on these points.
3701 <para>The <option>-XIncoherentInstances</option> flag implies the
3702 <option>-XOverlappingInstances</option> flag, but not vice versa.
3710 <sect2 id="overloaded-strings">
3711 <title>Overloaded string literals
3715 GHC supports <emphasis>overloaded string literals</emphasis>. Normally a
3716 string literal has type <literal>String</literal>, but with overloaded string
3717 literals enabled (with <literal>-XOverloadedStrings</literal>)
3718 a string literal has type <literal>(IsString a) => a</literal>.
3721 This means that the usual string syntax can be used, e.g., for packed strings
3722 and other variations of string like types. String literals behave very much
3723 like integer literals, i.e., they can be used in both expressions and patterns.
3724 If used in a pattern the literal with be replaced by an equality test, in the same
3725 way as an integer literal is.
3728 The class <literal>IsString</literal> is defined as:
3730 class IsString a where
3731 fromString :: String -> a
3733 The only predefined instance is the obvious one to make strings work as usual:
3735 instance IsString [Char] where
3738 The class <literal>IsString</literal> is not in scope by default. If you want to mention
3739 it explicitly (for example, to give an instance declaration for it), you can import it
3740 from module <literal>GHC.Exts</literal>.
3743 Haskell's defaulting mechanism is extended to cover string literals, when <option>-XOverloadedStrings</option> is specified.
3747 Each type in a default declaration must be an
3748 instance of <literal>Num</literal> <emphasis>or</emphasis> of <literal>IsString</literal>.
3752 The standard defaulting rule (<ulink url="http://www.haskell.org/onlinereport/decls.html#sect4.3.4">Haskell Report, Section 4.3.4</ulink>)
3753 is extended thus: defaulting applies when all the unresolved constraints involve standard classes
3754 <emphasis>or</emphasis> <literal>IsString</literal>; and at least one is a numeric class
3755 <emphasis>or</emphasis> <literal>IsString</literal>.
3764 import GHC.Exts( IsString(..) )
3766 newtype MyString = MyString String deriving (Eq, Show)
3767 instance IsString MyString where
3768 fromString = MyString
3770 greet :: MyString -> MyString
3771 greet "hello" = "world"
3775 print $ greet "hello"
3776 print $ greet "fool"
3780 Note that deriving <literal>Eq</literal> is necessary for the pattern matching
3781 to work since it gets translated into an equality comparison.
3787 <sect1 id="type-families">
3788 <title>Type families</title>
3791 <firstterm>Indexed type families</firstterm> are a new GHC extension to
3792 facilitate type-level
3793 programming. Type families are a generalisation of <firstterm>associated
3794 data types</firstterm>
3795 (“<ulink url="http://www.cse.unsw.edu.au/~chak/papers/CKPM05.html">Associated
3796 Types with Class</ulink>”, M. Chakravarty, G. Keller, S. Peyton Jones,
3797 and S. Marlow. In Proceedings of “The 32nd Annual ACM SIGPLAN-SIGACT
3798 Symposium on Principles of Programming Languages (POPL'05)”, pages
3799 1-13, ACM Press, 2005) and <firstterm>associated type synonyms</firstterm>
3800 (“<ulink url="http://www.cse.unsw.edu.au/~chak/papers/CKP05.html">Type
3801 Associated Type Synonyms</ulink>”. M. Chakravarty, G. Keller, and
3803 In Proceedings of “The Tenth ACM SIGPLAN International Conference on
3804 Functional Programming”, ACM Press, pages 241-253, 2005). Type families
3805 themselves are described in the paper “<ulink
3806 url="http://www.cse.unsw.edu.au/~chak/papers/SPCS08.html">Type
3807 Checking with Open Type Functions</ulink>”, T. Schrijvers,
3809 M. Chakravarty, and M. Sulzmann, in Proceedings of “ICFP 2008: The
3810 13th ACM SIGPLAN International Conference on Functional
3811 Programming”, ACM Press, pages 51-62, 2008. Type families
3812 essentially provide type-indexed data types and named functions on types,
3813 which are useful for generic programming and highly parameterised library
3814 interfaces as well as interfaces with enhanced static information, much like
3815 dependent types. They might also be regarded as an alternative to functional
3816 dependencies, but provide a more functional style of type-level programming
3817 than the relational style of functional dependencies.
3820 Indexed type families, or type families for short, are type constructors that
3821 represent sets of types. Set members are denoted by supplying the type family
3822 constructor with type parameters, which are called <firstterm>type
3823 indices</firstterm>. The
3824 difference between vanilla parametrised type constructors and family
3825 constructors is much like between parametrically polymorphic functions and
3826 (ad-hoc polymorphic) methods of type classes. Parametric polymorphic functions
3827 behave the same at all type instances, whereas class methods can change their
3828 behaviour in dependence on the class type parameters. Similarly, vanilla type
3829 constructors imply the same data representation for all type instances, but
3830 family constructors can have varying representation types for varying type
3834 Indexed type families come in two flavours: <firstterm>data
3835 families</firstterm> and <firstterm>type synonym
3836 families</firstterm>. They are the indexed family variants of algebraic
3837 data types and type synonyms, respectively. The instances of data families
3838 can be data types and newtypes.
3841 Type families are enabled by the flag <option>-XTypeFamilies</option>.
3842 Additional information on the use of type families in GHC is available on
3843 <ulink url="http://www.haskell.org/haskellwiki/GHC/Indexed_types">the
3844 Haskell wiki page on type families</ulink>.
3847 <sect2 id="data-families">
3848 <title>Data families</title>
3851 Data families appear in two flavours: (1) they can be defined on the
3853 or (2) they can appear inside type classes (in which case they are known as
3854 associated types). The former is the more general variant, as it lacks the
3855 requirement for the type-indexes to coincide with the class
3856 parameters. However, the latter can lead to more clearly structured code and
3857 compiler warnings if some type instances were - possibly accidentally -
3858 omitted. In the following, we always discuss the general toplevel form first
3859 and then cover the additional constraints placed on associated types.
3862 <sect3 id="data-family-declarations">
3863 <title>Data family declarations</title>
3866 Indexed data families are introduced by a signature, such as
3868 data family GMap k :: * -> *
3870 The special <literal>family</literal> distinguishes family from standard
3871 data declarations. The result kind annotation is optional and, as
3872 usual, defaults to <literal>*</literal> if omitted. An example is
3876 Named arguments can also be given explicit kind signatures if needed.
3878 [http://www.haskell.org/ghc/docs/latest/html/users_guide/gadt.html GADT
3879 declarations] named arguments are entirely optional, so that we can
3880 declare <literal>Array</literal> alternatively with
3882 data family Array :: * -> *
3886 <sect4 id="assoc-data-family-decl">
3887 <title>Associated data family declarations</title>
3889 When a data family is declared as part of a type class, we drop
3890 the <literal>family</literal> special. The <literal>GMap</literal>
3891 declaration takes the following form
3893 class GMapKey k where
3894 data GMap k :: * -> *
3897 In contrast to toplevel declarations, named arguments must be used for
3898 all type parameters that are to be used as type-indexes. Moreover,
3899 the argument names must be class parameters. Each class parameter may
3900 only be used at most once per associated type, but some may be omitted
3901 and they may be in an order other than in the class head. Hence, the
3902 following contrived example is admissible:
3911 <sect3 id="data-instance-declarations">
3912 <title>Data instance declarations</title>
3915 Instance declarations of data and newtype families are very similar to
3916 standard data and newtype declarations. The only two differences are
3917 that the keyword <literal>data</literal> or <literal>newtype</literal>
3918 is followed by <literal>instance</literal> and that some or all of the
3919 type arguments can be non-variable types, but may not contain forall
3920 types or type synonym families. However, data families are generally
3921 allowed in type parameters, and type synonyms are allowed as long as
3922 they are fully applied and expand to a type that is itself admissible -
3923 exactly as this is required for occurrences of type synonyms in class
3924 instance parameters. For example, the <literal>Either</literal>
3925 instance for <literal>GMap</literal> is
3927 data instance GMap (Either a b) v = GMapEither (GMap a v) (GMap b v)
3929 In this example, the declaration has only one variant. In general, it
3933 Data and newtype instance declarations are only permitted when an
3934 appropriate family declaration is in scope - just as a class instance declaratoin
3935 requires the class declaration to be visible. Moreover, each instance
3936 declaration has to conform to the kind determined by its family
3937 declaration. This implies that the number of parameters of an instance
3938 declaration matches the arity determined by the kind of the family.
3941 A data family instance declaration can use the full exprssiveness of
3942 ordinary <literal>data</literal> or <literal>newtype</literal> declarations:
3944 <listitem><para> Although, a data family is <emphasis>introduced</emphasis> with
3945 the keyword "<literal>data</literal>", a data family <emphasis>instance</emphasis> can
3946 use either <literal>data</literal> or <literal>newtype</literal>. For example:
3949 data instance T Int = T1 Int | T2 Bool
3950 newtype instance T Char = TC Bool
3953 <listitem><para> A <literal>data instance</literal> can use GADT syntax for the data constructors,
3954 and indeed can define a GADT. For example:
3957 data instance G [a] b where
3958 G1 :: c -> G [Int] b
3962 <listitem><para> You can use a <literal>deriving</literal> clause on a
3963 <literal>data instance</literal> or <literal>newtype instance</literal>
3970 Even if type families are defined as toplevel declarations, functions
3971 that perform different computations for different family instances may still
3972 need to be defined as methods of type classes. In particular, the
3973 following is not possible:
3976 data instance T Int = A
3977 data instance T Char = B
3979 foo A = 1 -- WRONG: These two equations together...
3980 foo B = 2 -- ...will produce a type error.
3982 Instead, you would have to write <literal>foo</literal> as a class operation, thus:
3986 instance Foo Int where
3988 instance Foo Char where
3991 (Given the functionality provided by GADTs (Generalised Algebraic Data
3992 Types), it might seem as if a definition, such as the above, should be
3993 feasible. However, type families are - in contrast to GADTs - are
3994 <emphasis>open;</emphasis> i.e., new instances can always be added,
3996 modules. Supporting pattern matching across different data instances
3997 would require a form of extensible case construct.)
4000 <sect4 id="assoc-data-inst">
4001 <title>Associated data instances</title>
4003 When an associated data family instance is declared within a type
4004 class instance, we drop the <literal>instance</literal> keyword in the
4005 family instance. So, the <literal>Either</literal> instance
4006 for <literal>GMap</literal> becomes:
4008 instance (GMapKey a, GMapKey b) => GMapKey (Either a b) where
4009 data GMap (Either a b) v = GMapEither (GMap a v) (GMap b v)
4012 The most important point about associated family instances is that the
4013 type indexes corresponding to class parameters must be identical to
4014 the type given in the instance head; here this is the first argument
4015 of <literal>GMap</literal>, namely <literal>Either a b</literal>,
4016 which coincides with the only class parameter. Any parameters to the
4017 family constructor that do not correspond to class parameters, need to
4018 be variables in every instance; here this is the
4019 variable <literal>v</literal>.
4022 Instances for an associated family can only appear as part of
4023 instances declarations of the class in which the family was declared -
4024 just as with the equations of the methods of a class. Also in
4025 correspondence to how methods are handled, declarations of associated
4026 types can be omitted in class instances. If an associated family
4027 instance is omitted, the corresponding instance type is not inhabited;
4028 i.e., only diverging expressions, such
4029 as <literal>undefined</literal>, can assume the type.
4033 <sect4 id="scoping-class-params">
4034 <title>Scoping of class parameters</title>
4036 In the case of multi-parameter type classes, the visibility of class
4037 parameters in the right-hand side of associated family instances
4038 depends <emphasis>solely</emphasis> on the parameters of the data
4039 family. As an example, consider the simple class declaration
4044 Only one of the two class parameters is a parameter to the data
4045 family. Hence, the following instance declaration is invalid:
4047 instance C [c] d where
4048 data T [c] = MkT (c, d) -- WRONG!! 'd' is not in scope
4050 Here, the right-hand side of the data instance mentions the type
4051 variable <literal>d</literal> that does not occur in its left-hand
4052 side. We cannot admit such data instances as they would compromise
4057 <sect4 id="family-class-inst">
4058 <title>Type class instances of family instances</title>
4060 Type class instances of instances of data families can be defined as
4061 usual, and in particular data instance declarations can
4062 have <literal>deriving</literal> clauses. For example, we can write
4064 data GMap () v = GMapUnit (Maybe v)
4067 which implicitly defines an instance of the form
4069 instance Show v => Show (GMap () v) where ...
4073 Note that class instances are always for
4074 particular <emphasis>instances</emphasis> of a data family and never
4075 for an entire family as a whole. This is for essentially the same
4076 reasons that we cannot define a toplevel function that performs
4077 pattern matching on the data constructors
4078 of <emphasis>different</emphasis> instances of a single type family.
4079 It would require a form of extensible case construct.
4083 <sect4 id="data-family-overlap">
4084 <title>Overlap of data instances</title>
4086 The instance declarations of a data family used in a single program
4087 may not overlap at all, independent of whether they are associated or
4088 not. In contrast to type class instances, this is not only a matter
4089 of consistency, but one of type safety.
4095 <sect3 id="data-family-import-export">
4096 <title>Import and export</title>
4099 The association of data constructors with type families is more dynamic
4100 than that is the case with standard data and newtype declarations. In
4101 the standard case, the notation <literal>T(..)</literal> in an import or
4102 export list denotes the type constructor and all the data constructors
4103 introduced in its declaration. However, a family declaration never
4104 introduces any data constructors; instead, data constructors are
4105 introduced by family instances. As a result, which data constructors
4106 are associated with a type family depends on the currently visible
4107 instance declarations for that family. Consequently, an import or
4108 export item of the form <literal>T(..)</literal> denotes the family
4109 constructor and all currently visible data constructors - in the case of
4110 an export item, these may be either imported or defined in the current
4111 module. The treatment of import and export items that explicitly list
4112 data constructors, such as <literal>GMap(GMapEither)</literal>, is
4116 <sect4 id="data-family-impexp-assoc">
4117 <title>Associated families</title>
4119 As expected, an import or export item of the
4120 form <literal>C(..)</literal> denotes all of the class' methods and
4121 associated types. However, when associated types are explicitly
4122 listed as subitems of a class, we need some new syntax, as uppercase
4123 identifiers as subitems are usually data constructors, not type
4124 constructors. To clarify that we denote types here, each associated
4125 type name needs to be prefixed by the keyword <literal>type</literal>.
4126 So for example, when explicitly listing the components of
4127 the <literal>GMapKey</literal> class, we write <literal>GMapKey(type
4128 GMap, empty, lookup, insert)</literal>.
4132 <sect4 id="data-family-impexp-examples">
4133 <title>Examples</title>
4135 Assuming our running <literal>GMapKey</literal> class example, let us
4136 look at some export lists and their meaning:
4139 <para><literal>module GMap (GMapKey) where...</literal>: Exports
4140 just the class name.</para>
4143 <para><literal>module GMap (GMapKey(..)) where...</literal>:
4144 Exports the class, the associated type <literal>GMap</literal>
4146 functions <literal>empty</literal>, <literal>lookup</literal>,
4147 and <literal>insert</literal>. None of the data constructors is
4151 <para><literal>module GMap (GMapKey(..), GMap(..))
4152 where...</literal>: As before, but also exports all the data
4153 constructors <literal>GMapInt</literal>,
4154 <literal>GMapChar</literal>,
4155 <literal>GMapUnit</literal>, <literal>GMapPair</literal>,
4156 and <literal>GMapUnit</literal>.</para>
4159 <para><literal>module GMap (GMapKey(empty, lookup, insert),
4160 GMap(..)) where...</literal>: As before.</para>
4163 <para><literal>module GMap (GMapKey, empty, lookup, insert, GMap(..))
4164 where...</literal>: As before.</para>
4169 Finally, you can write <literal>GMapKey(type GMap)</literal> to denote
4170 both the class <literal>GMapKey</literal> as well as its associated
4171 type <literal>GMap</literal>. However, you cannot
4172 write <literal>GMapKey(type GMap(..))</literal> — i.e.,
4173 sub-component specifications cannot be nested. To
4174 specify <literal>GMap</literal>'s data constructors, you have to list
4179 <sect4 id="data-family-impexp-instances">
4180 <title>Instances</title>
4182 Family instances are implicitly exported, just like class instances.
4183 However, this applies only to the heads of instances, not to the data
4184 constructors an instance defines.
4192 <sect2 id="synonym-families">
4193 <title>Synonym families</title>
4196 Type families appear in two flavours: (1) they can be defined on the
4197 toplevel or (2) they can appear inside type classes (in which case they
4198 are known as associated type synonyms). The former is the more general
4199 variant, as it lacks the requirement for the type-indexes to coincide with
4200 the class parameters. However, the latter can lead to more clearly
4201 structured code and compiler warnings if some type instances were -
4202 possibly accidentally - omitted. In the following, we always discuss the
4203 general toplevel form first and then cover the additional constraints
4204 placed on associated types.
4207 <sect3 id="type-family-declarations">
4208 <title>Type family declarations</title>
4211 Indexed type families are introduced by a signature, such as
4213 type family Elem c :: *
4215 The special <literal>family</literal> distinguishes family from standard
4216 type declarations. The result kind annotation is optional and, as
4217 usual, defaults to <literal>*</literal> if omitted. An example is
4221 Parameters can also be given explicit kind signatures if needed. We
4222 call the number of parameters in a type family declaration, the family's
4223 arity, and all applications of a type family must be fully saturated
4224 w.r.t. to that arity. This requirement is unlike ordinary type synonyms
4225 and it implies that the kind of a type family is not sufficient to
4226 determine a family's arity, and hence in general, also insufficient to
4227 determine whether a type family application is well formed. As an
4228 example, consider the following declaration:
4230 type family F a b :: * -> * -- F's arity is 2,
4231 -- although it's overall kind is * -> * -> * -> *
4233 Given this declaration the following are examples of well-formed and
4236 F Char [Int] -- OK! Kind: * -> *
4237 F Char [Int] Bool -- OK! Kind: *
4238 F IO Bool -- WRONG: kind mismatch in the first argument
4239 F Bool -- WRONG: unsaturated application
4243 <sect4 id="assoc-type-family-decl">
4244 <title>Associated type family declarations</title>
4246 When a type family is declared as part of a type class, we drop
4247 the <literal>family</literal> special. The <literal>Elem</literal>
4248 declaration takes the following form
4250 class Collects ce where
4254 The argument names of the type family must be class parameters. Each
4255 class parameter may only be used at most once per associated type, but
4256 some may be omitted and they may be in an order other than in the
4257 class head. Hence, the following contrived example is admissible:
4262 These rules are exactly as for associated data families.
4267 <sect3 id="type-instance-declarations">
4268 <title>Type instance declarations</title>
4270 Instance declarations of type families are very similar to standard type
4271 synonym declarations. The only two differences are that the
4272 keyword <literal>type</literal> is followed
4273 by <literal>instance</literal> and that some or all of the type
4274 arguments can be non-variable types, but may not contain forall types or
4275 type synonym families. However, data families are generally allowed, and
4276 type synonyms are allowed as long as they are fully applied and expand
4277 to a type that is admissible - these are the exact same requirements as
4278 for data instances. For example, the <literal>[e]</literal> instance
4279 for <literal>Elem</literal> is
4281 type instance Elem [e] = e
4285 Type family instance declarations are only legitimate when an
4286 appropriate family declaration is in scope - just like class instances
4287 require the class declaration to be visible. Moreover, each instance
4288 declaration has to conform to the kind determined by its family
4289 declaration, and the number of type parameters in an instance
4290 declaration must match the number of type parameters in the family
4291 declaration. Finally, the right-hand side of a type instance must be a
4292 monotype (i.e., it may not include foralls) and after the expansion of
4293 all saturated vanilla type synonyms, no synonyms, except family synonyms
4294 may remain. Here are some examples of admissible and illegal type
4297 type family F a :: *
4298 type instance F [Int] = Int -- OK!
4299 type instance F String = Char -- OK!
4300 type instance F (F a) = a -- WRONG: type parameter mentions a type family
4301 type instance F (forall a. (a, b)) = b -- WRONG: a forall type appears in a type parameter
4302 type instance F Float = forall a.a -- WRONG: right-hand side may not be a forall type
4304 type family G a b :: * -> *
4305 type instance G Int = (,) -- WRONG: must be two type parameters
4306 type instance G Int Char Float = Double -- WRONG: must be two type parameters
4310 <sect4 id="assoc-type-instance">
4311 <title>Associated type instance declarations</title>
4313 When an associated family instance is declared within a type class
4314 instance, we drop the <literal>instance</literal> keyword in the family
4315 instance. So, the <literal>[e]</literal> instance
4316 for <literal>Elem</literal> becomes:
4318 instance (Eq (Elem [e])) => Collects ([e]) where
4322 The most important point about associated family instances is that the
4323 type indexes corresponding to class parameters must be identical to the
4324 type given in the instance head; here this is <literal>[e]</literal>,
4325 which coincides with the only class parameter.
4328 Instances for an associated family can only appear as part of instances
4329 declarations of the class in which the family was declared - just as
4330 with the equations of the methods of a class. Also in correspondence to
4331 how methods are handled, declarations of associated types can be omitted
4332 in class instances. If an associated family instance is omitted, the
4333 corresponding instance type is not inhabited; i.e., only diverging
4334 expressions, such as <literal>undefined</literal>, can assume the type.
4338 <sect4 id="type-family-overlap">
4339 <title>Overlap of type synonym instances</title>
4341 The instance declarations of a type family used in a single program
4342 may only overlap if the right-hand sides of the overlapping instances
4343 coincide for the overlapping types. More formally, two instance
4344 declarations overlap if there is a substitution that makes the
4345 left-hand sides of the instances syntactically the same. Whenever
4346 that is the case, the right-hand sides of the instances must also be
4347 syntactically equal under the same substitution. This condition is
4348 independent of whether the type family is associated or not, and it is
4349 not only a matter of consistency, but one of type safety.
4352 Here are two example to illustrate the condition under which overlap
4355 type instance F (a, Int) = [a]
4356 type instance F (Int, b) = [b] -- overlap permitted
4358 type instance G (a, Int) = [a]
4359 type instance G (Char, a) = [a] -- ILLEGAL overlap, as [Char] /= [Int]
4364 <sect4 id="type-family-decidability">
4365 <title>Decidability of type synonym instances</title>
4367 In order to guarantee that type inference in the presence of type
4368 families decidable, we need to place a number of additional
4369 restrictions on the formation of type instance declarations (c.f.,
4370 Definition 5 (Relaxed Conditions) of “<ulink
4371 url="http://www.cse.unsw.edu.au/~chak/papers/SPCS08.html">Type
4372 Checking with Open Type Functions</ulink>”). Instance
4373 declarations have the general form
4375 type instance F t1 .. tn = t
4377 where we require that for every type family application <literal>(G s1
4378 .. sm)</literal> in <literal>t</literal>,
4381 <para><literal>s1 .. sm</literal> do not contain any type family
4382 constructors,</para>
4385 <para>the total number of symbols (data type constructors and type
4386 variables) in <literal>s1 .. sm</literal> is strictly smaller than
4387 in <literal>t1 .. tn</literal>, and</para>
4390 <para>for every type
4391 variable <literal>a</literal>, <literal>a</literal> occurs
4392 in <literal>s1 .. sm</literal> at most as often as in <literal>t1
4393 .. tn</literal>.</para>
4396 These restrictions are easily verified and ensure termination of type
4397 inference. However, they are not sufficient to guarantee completeness
4398 of type inference in the presence of, so called, ''loopy equalities'',
4399 such as <literal>a ~ [F a]</literal>, where a recursive occurrence of
4400 a type variable is underneath a family application and data
4401 constructor application - see the above mentioned paper for details.
4404 If the option <option>-XUndecidableInstances</option> is passed to the
4405 compiler, the above restrictions are not enforced and it is on the
4406 programmer to ensure termination of the normalisation of type families
4407 during type inference.
4412 <sect3 id-="equality-constraints">
4413 <title>Equality constraints</title>
4415 Type context can include equality constraints of the form <literal>t1 ~
4416 t2</literal>, which denote that the types <literal>t1</literal>
4417 and <literal>t2</literal> need to be the same. In the presence of type
4418 families, whether two types are equal cannot generally be decided
4419 locally. Hence, the contexts of function signatures may include
4420 equality constraints, as in the following example:
4422 sumCollects :: (Collects c1, Collects c2, Elem c1 ~ Elem c2) => c1 -> c2 -> c2
4424 where we require that the element type of <literal>c1</literal>
4425 and <literal>c2</literal> are the same. In general, the
4426 types <literal>t1</literal> and <literal>t2</literal> of an equality
4427 constraint may be arbitrary monotypes; i.e., they may not contain any
4428 quantifiers, independent of whether higher-rank types are otherwise
4432 Equality constraints can also appear in class and instance contexts.
4433 The former enable a simple translation of programs using functional
4434 dependencies into programs using family synonyms instead. The general
4435 idea is to rewrite a class declaration of the form
4437 class C a b | a -> b
4441 class (F a ~ b) => C a b where
4444 That is, we represent every functional dependency (FD) <literal>a1 .. an
4445 -> b</literal> by an FD type family <literal>F a1 .. an</literal> and a
4446 superclass context equality <literal>F a1 .. an ~ b</literal>,
4447 essentially giving a name to the functional dependency. In class
4448 instances, we define the type instances of FD families in accordance
4449 with the class head. Method signatures are not affected by that
4453 NB: Equalities in superclass contexts are not fully implemented in
4458 <sect3 id-="ty-fams-in-instances">
4459 <title>Type families and instance declarations</title>
4460 <para>Type families require us to extend the rules for
4461 the form of instance heads, which are given
4462 in <xref linkend="flexible-instance-head"/>.
4465 <listitem><para>Data type families may appear in an instance head</para></listitem>
4466 <listitem><para>Type synonym families may not appear (at all) in an instance head</para></listitem>
4468 The reason for the latter restriction is that there is no way to check for. Consider
4471 type instance F Bool = Int
4478 Now a constraint <literal>(C (F Bool))</literal> would match both instances.
4479 The situation is especially bad because the type instance for <literal>F Bool</literal>
4480 might be in another module, or even in a module that is not yet written.
4487 <sect1 id="other-type-extensions">
4488 <title>Other type system extensions</title>
4490 <sect2 id="type-restrictions">
4491 <title>Type signatures</title>
4493 <sect3 id="flexible-contexts"><title>The context of a type signature</title>
4495 The <option>-XFlexibleContexts</option> flag lifts the Haskell 98 restriction
4496 that the type-class constraints in a type signature must have the
4497 form <emphasis>(class type-variable)</emphasis> or
4498 <emphasis>(class (type-variable type-variable ...))</emphasis>.
4499 With <option>-XFlexibleContexts</option>
4500 these type signatures are perfectly OK
4503 g :: Ord (T a ()) => ...
4507 GHC imposes the following restrictions on the constraints in a type signature.
4511 forall tv1..tvn (c1, ...,cn) => type
4514 (Here, we write the "foralls" explicitly, although the Haskell source
4515 language omits them; in Haskell 98, all the free type variables of an
4516 explicit source-language type signature are universally quantified,
4517 except for the class type variables in a class declaration. However,
4518 in GHC, you can give the foralls if you want. See <xref linkend="universal-quantification"/>).
4527 <emphasis>Each universally quantified type variable
4528 <literal>tvi</literal> must be reachable from <literal>type</literal></emphasis>.
4530 A type variable <literal>a</literal> is "reachable" if it appears
4531 in the same constraint as either a type variable free in
4532 <literal>type</literal>, or another reachable type variable.
4533 A value with a type that does not obey
4534 this reachability restriction cannot be used without introducing
4535 ambiguity; that is why the type is rejected.
4536 Here, for example, is an illegal type:
4540 forall a. Eq a => Int
4544 When a value with this type was used, the constraint <literal>Eq tv</literal>
4545 would be introduced where <literal>tv</literal> is a fresh type variable, and
4546 (in the dictionary-translation implementation) the value would be
4547 applied to a dictionary for <literal>Eq tv</literal>. The difficulty is that we
4548 can never know which instance of <literal>Eq</literal> to use because we never
4549 get any more information about <literal>tv</literal>.
4553 that the reachability condition is weaker than saying that <literal>a</literal> is
4554 functionally dependent on a type variable free in
4555 <literal>type</literal> (see <xref
4556 linkend="functional-dependencies"/>). The reason for this is there
4557 might be a "hidden" dependency, in a superclass perhaps. So
4558 "reachable" is a conservative approximation to "functionally dependent".
4559 For example, consider:
4561 class C a b | a -> b where ...
4562 class C a b => D a b where ...
4563 f :: forall a b. D a b => a -> a
4565 This is fine, because in fact <literal>a</literal> does functionally determine <literal>b</literal>
4566 but that is not immediately apparent from <literal>f</literal>'s type.
4572 <emphasis>Every constraint <literal>ci</literal> must mention at least one of the
4573 universally quantified type variables <literal>tvi</literal></emphasis>.
4575 For example, this type is OK because <literal>C a b</literal> mentions the
4576 universally quantified type variable <literal>b</literal>:
4580 forall a. C a b => burble
4584 The next type is illegal because the constraint <literal>Eq b</literal> does not
4585 mention <literal>a</literal>:
4589 forall a. Eq b => burble
4593 The reason for this restriction is milder than the other one. The
4594 excluded types are never useful or necessary (because the offending
4595 context doesn't need to be witnessed at this point; it can be floated
4596 out). Furthermore, floating them out increases sharing. Lastly,
4597 excluding them is a conservative choice; it leaves a patch of
4598 territory free in case we need it later.
4612 <sect2 id="implicit-parameters">
4613 <title>Implicit parameters</title>
4615 <para> Implicit parameters are implemented as described in
4616 "Implicit parameters: dynamic scoping with static types",
4617 J Lewis, MB Shields, E Meijer, J Launchbury,
4618 27th ACM Symposium on Principles of Programming Languages (POPL'00),
4622 <para>(Most of the following, still rather incomplete, documentation is
4623 due to Jeff Lewis.)</para>
4625 <para>Implicit parameter support is enabled with the option
4626 <option>-XImplicitParams</option>.</para>
4629 A variable is called <emphasis>dynamically bound</emphasis> when it is bound by the calling
4630 context of a function and <emphasis>statically bound</emphasis> when bound by the callee's
4631 context. In Haskell, all variables are statically bound. Dynamic
4632 binding of variables is a notion that goes back to Lisp, but was later
4633 discarded in more modern incarnations, such as Scheme. Dynamic binding
4634 can be very confusing in an untyped language, and unfortunately, typed
4635 languages, in particular Hindley-Milner typed languages like Haskell,
4636 only support static scoping of variables.
4639 However, by a simple extension to the type class system of Haskell, we
4640 can support dynamic binding. Basically, we express the use of a
4641 dynamically bound variable as a constraint on the type. These
4642 constraints lead to types of the form <literal>(?x::t') => t</literal>, which says "this
4643 function uses a dynamically-bound variable <literal>?x</literal>
4644 of type <literal>t'</literal>". For
4645 example, the following expresses the type of a sort function,
4646 implicitly parameterized by a comparison function named <literal>cmp</literal>.
4648 sort :: (?cmp :: a -> a -> Bool) => [a] -> [a]
4650 The dynamic binding constraints are just a new form of predicate in the type class system.
4653 An implicit parameter occurs in an expression using the special form <literal>?x</literal>,
4654 where <literal>x</literal> is
4655 any valid identifier (e.g. <literal>ord ?x</literal> is a valid expression).
4656 Use of this construct also introduces a new
4657 dynamic-binding constraint in the type of the expression.
4658 For example, the following definition
4659 shows how we can define an implicitly parameterized sort function in
4660 terms of an explicitly parameterized <literal>sortBy</literal> function:
4662 sortBy :: (a -> a -> Bool) -> [a] -> [a]
4664 sort :: (?cmp :: a -> a -> Bool) => [a] -> [a]
4670 <title>Implicit-parameter type constraints</title>
4672 Dynamic binding constraints behave just like other type class
4673 constraints in that they are automatically propagated. Thus, when a
4674 function is used, its implicit parameters are inherited by the
4675 function that called it. For example, our <literal>sort</literal> function might be used
4676 to pick out the least value in a list:
4678 least :: (?cmp :: a -> a -> Bool) => [a] -> a
4679 least xs = head (sort xs)
4681 Without lifting a finger, the <literal>?cmp</literal> parameter is
4682 propagated to become a parameter of <literal>least</literal> as well. With explicit
4683 parameters, the default is that parameters must always be explicit
4684 propagated. With implicit parameters, the default is to always
4688 An implicit-parameter type constraint differs from other type class constraints in the
4689 following way: All uses of a particular implicit parameter must have
4690 the same type. This means that the type of <literal>(?x, ?x)</literal>
4691 is <literal>(?x::a) => (a,a)</literal>, and not
4692 <literal>(?x::a, ?x::b) => (a, b)</literal>, as would be the case for type
4696 <para> You can't have an implicit parameter in the context of a class or instance
4697 declaration. For example, both these declarations are illegal:
4699 class (?x::Int) => C a where ...
4700 instance (?x::a) => Foo [a] where ...
4702 Reason: exactly which implicit parameter you pick up depends on exactly where
4703 you invoke a function. But the ``invocation'' of instance declarations is done
4704 behind the scenes by the compiler, so it's hard to figure out exactly where it is done.
4705 Easiest thing is to outlaw the offending types.</para>
4707 Implicit-parameter constraints do not cause ambiguity. For example, consider:
4709 f :: (?x :: [a]) => Int -> Int
4712 g :: (Read a, Show a) => String -> String
4715 Here, <literal>g</literal> has an ambiguous type, and is rejected, but <literal>f</literal>
4716 is fine. The binding for <literal>?x</literal> at <literal>f</literal>'s call site is
4717 quite unambiguous, and fixes the type <literal>a</literal>.
4722 <title>Implicit-parameter bindings</title>
4725 An implicit parameter is <emphasis>bound</emphasis> using the standard
4726 <literal>let</literal> or <literal>where</literal> binding forms.
4727 For example, we define the <literal>min</literal> function by binding
4728 <literal>cmp</literal>.
4731 min = let ?cmp = (<=) in least
4735 A group of implicit-parameter bindings may occur anywhere a normal group of Haskell
4736 bindings can occur, except at top level. That is, they can occur in a <literal>let</literal>
4737 (including in a list comprehension, or do-notation, or pattern guards),
4738 or a <literal>where</literal> clause.
4739 Note the following points:
4742 An implicit-parameter binding group must be a
4743 collection of simple bindings to implicit-style variables (no
4744 function-style bindings, and no type signatures); these bindings are
4745 neither polymorphic or recursive.
4748 You may not mix implicit-parameter bindings with ordinary bindings in a
4749 single <literal>let</literal>
4750 expression; use two nested <literal>let</literal>s instead.
4751 (In the case of <literal>where</literal> you are stuck, since you can't nest <literal>where</literal> clauses.)
4755 You may put multiple implicit-parameter bindings in a
4756 single binding group; but they are <emphasis>not</emphasis> treated
4757 as a mutually recursive group (as ordinary <literal>let</literal> bindings are).
4758 Instead they are treated as a non-recursive group, simultaneously binding all the implicit
4759 parameter. The bindings are not nested, and may be re-ordered without changing
4760 the meaning of the program.
4761 For example, consider:
4763 f t = let { ?x = t; ?y = ?x+(1::Int) } in ?x + ?y
4765 The use of <literal>?x</literal> in the binding for <literal>?y</literal> does not "see"
4766 the binding for <literal>?x</literal>, so the type of <literal>f</literal> is
4768 f :: (?x::Int) => Int -> Int
4776 <sect3><title>Implicit parameters and polymorphic recursion</title>
4779 Consider these two definitions:
4782 len1 xs = let ?acc = 0 in len_acc1 xs
4785 len_acc1 (x:xs) = let ?acc = ?acc + (1::Int) in len_acc1 xs
4790 len2 xs = let ?acc = 0 in len_acc2 xs
4792 len_acc2 :: (?acc :: Int) => [a] -> Int
4794 len_acc2 (x:xs) = let ?acc = ?acc + (1::Int) in len_acc2 xs
4796 The only difference between the two groups is that in the second group
4797 <literal>len_acc</literal> is given a type signature.
4798 In the former case, <literal>len_acc1</literal> is monomorphic in its own
4799 right-hand side, so the implicit parameter <literal>?acc</literal> is not
4800 passed to the recursive call. In the latter case, because <literal>len_acc2</literal>
4801 has a type signature, the recursive call is made to the
4802 <emphasis>polymorphic</emphasis> version, which takes <literal>?acc</literal>
4803 as an implicit parameter. So we get the following results in GHCi:
4810 Adding a type signature dramatically changes the result! This is a rather
4811 counter-intuitive phenomenon, worth watching out for.
4815 <sect3><title>Implicit parameters and monomorphism</title>
4817 <para>GHC applies the dreaded Monomorphism Restriction (section 4.5.5 of the
4818 Haskell Report) to implicit parameters. For example, consider:
4826 Since the binding for <literal>y</literal> falls under the Monomorphism
4827 Restriction it is not generalised, so the type of <literal>y</literal> is
4828 simply <literal>Int</literal>, not <literal>(?x::Int) => Int</literal>.
4829 Hence, <literal>(f 9)</literal> returns result <literal>9</literal>.
4830 If you add a type signature for <literal>y</literal>, then <literal>y</literal>
4831 will get type <literal>(?x::Int) => Int</literal>, so the occurrence of
4832 <literal>y</literal> in the body of the <literal>let</literal> will see the
4833 inner binding of <literal>?x</literal>, so <literal>(f 9)</literal> will return
4834 <literal>14</literal>.
4839 <!-- ======================= COMMENTED OUT ========================
4841 We intend to remove linear implicit parameters, so I'm at least removing
4842 them from the 6.6 user manual
4844 <sect2 id="linear-implicit-parameters">
4845 <title>Linear implicit parameters</title>
4847 Linear implicit parameters are an idea developed by Koen Claessen,
4848 Mark Shields, and Simon PJ. They address the long-standing
4849 problem that monads seem over-kill for certain sorts of problem, notably:
4852 <listitem> <para> distributing a supply of unique names </para> </listitem>
4853 <listitem> <para> distributing a supply of random numbers </para> </listitem>
4854 <listitem> <para> distributing an oracle (as in QuickCheck) </para> </listitem>
4858 Linear implicit parameters are just like ordinary implicit parameters,
4859 except that they are "linear"; that is, they cannot be copied, and
4860 must be explicitly "split" instead. Linear implicit parameters are
4861 written '<literal>%x</literal>' instead of '<literal>?x</literal>'.
4862 (The '/' in the '%' suggests the split!)
4867 import GHC.Exts( Splittable )
4869 data NameSupply = ...
4871 splitNS :: NameSupply -> (NameSupply, NameSupply)
4872 newName :: NameSupply -> Name
4874 instance Splittable NameSupply where
4878 f :: (%ns :: NameSupply) => Env -> Expr -> Expr
4879 f env (Lam x e) = Lam x' (f env e)
4882 env' = extend env x x'
4883 ...more equations for f...
4885 Notice that the implicit parameter %ns is consumed
4887 <listitem> <para> once by the call to <literal>newName</literal> </para> </listitem>
4888 <listitem> <para> once by the recursive call to <literal>f</literal> </para></listitem>
4892 So the translation done by the type checker makes
4893 the parameter explicit:
4895 f :: NameSupply -> Env -> Expr -> Expr
4896 f ns env (Lam x e) = Lam x' (f ns1 env e)
4898 (ns1,ns2) = splitNS ns
4900 env = extend env x x'
4902 Notice the call to 'split' introduced by the type checker.
4903 How did it know to use 'splitNS'? Because what it really did
4904 was to introduce a call to the overloaded function 'split',
4905 defined by the class <literal>Splittable</literal>:
4907 class Splittable a where
4910 The instance for <literal>Splittable NameSupply</literal> tells GHC how to implement
4911 split for name supplies. But we can simply write
4917 g :: (Splittable a, %ns :: a) => b -> (b,a,a)
4919 The <literal>Splittable</literal> class is built into GHC. It's exported by module
4920 <literal>GHC.Exts</literal>.
4925 <listitem> <para> '<literal>?x</literal>' and '<literal>%x</literal>'
4926 are entirely distinct implicit parameters: you
4927 can use them together and they won't interfere with each other. </para>
4930 <listitem> <para> You can bind linear implicit parameters in 'with' clauses. </para> </listitem>
4932 <listitem> <para>You cannot have implicit parameters (whether linear or not)
4933 in the context of a class or instance declaration. </para></listitem>
4937 <sect3><title>Warnings</title>
4940 The monomorphism restriction is even more important than usual.
4941 Consider the example above:
4943 f :: (%ns :: NameSupply) => Env -> Expr -> Expr
4944 f env (Lam x e) = Lam x' (f env e)
4947 env' = extend env x x'
4949 If we replaced the two occurrences of x' by (newName %ns), which is
4950 usually a harmless thing to do, we get:
4952 f :: (%ns :: NameSupply) => Env -> Expr -> Expr
4953 f env (Lam x e) = Lam (newName %ns) (f env e)
4955 env' = extend env x (newName %ns)
4957 But now the name supply is consumed in <emphasis>three</emphasis> places
4958 (the two calls to newName,and the recursive call to f), so
4959 the result is utterly different. Urk! We don't even have
4963 Well, this is an experimental change. With implicit
4964 parameters we have already lost beta reduction anyway, and
4965 (as John Launchbury puts it) we can't sensibly reason about
4966 Haskell programs without knowing their typing.
4971 <sect3><title>Recursive functions</title>
4972 <para>Linear implicit parameters can be particularly tricky when you have a recursive function
4975 foo :: %x::T => Int -> [Int]
4977 foo n = %x : foo (n-1)
4979 where T is some type in class Splittable.</para>
4981 Do you get a list of all the same T's or all different T's
4982 (assuming that split gives two distinct T's back)?
4984 If you supply the type signature, taking advantage of polymorphic
4985 recursion, you get what you'd probably expect. Here's the
4986 translated term, where the implicit param is made explicit:
4989 foo x n = let (x1,x2) = split x
4990 in x1 : foo x2 (n-1)
4992 But if you don't supply a type signature, GHC uses the Hindley
4993 Milner trick of using a single monomorphic instance of the function
4994 for the recursive calls. That is what makes Hindley Milner type inference
4995 work. So the translation becomes
4999 foom n = x : foom (n-1)
5003 Result: 'x' is not split, and you get a list of identical T's. So the
5004 semantics of the program depends on whether or not foo has a type signature.
5007 You may say that this is a good reason to dislike linear implicit parameters
5008 and you'd be right. That is why they are an experimental feature.
5014 ================ END OF Linear Implicit Parameters commented out -->
5016 <sect2 id="kinding">
5017 <title>Explicitly-kinded quantification</title>
5020 Haskell infers the kind of each type variable. Sometimes it is nice to be able
5021 to give the kind explicitly as (machine-checked) documentation,
5022 just as it is nice to give a type signature for a function. On some occasions,
5023 it is essential to do so. For example, in his paper "Restricted Data Types in Haskell" (Haskell Workshop 1999)
5024 John Hughes had to define the data type:
5026 data Set cxt a = Set [a]
5027 | Unused (cxt a -> ())
5029 The only use for the <literal>Unused</literal> constructor was to force the correct
5030 kind for the type variable <literal>cxt</literal>.
5033 GHC now instead allows you to specify the kind of a type variable directly, wherever
5034 a type variable is explicitly bound, with the flag <option>-XKindSignatures</option>.
5037 This flag enables kind signatures in the following places:
5039 <listitem><para><literal>data</literal> declarations:
5041 data Set (cxt :: * -> *) a = Set [a]
5042 </screen></para></listitem>
5043 <listitem><para><literal>type</literal> declarations:
5045 type T (f :: * -> *) = f Int
5046 </screen></para></listitem>
5047 <listitem><para><literal>class</literal> declarations:
5049 class (Eq a) => C (f :: * -> *) a where ...
5050 </screen></para></listitem>
5051 <listitem><para><literal>forall</literal>'s in type signatures:
5053 f :: forall (cxt :: * -> *). Set cxt Int
5054 </screen></para></listitem>
5059 The parentheses are required. Some of the spaces are required too, to
5060 separate the lexemes. If you write <literal>(f::*->*)</literal> you
5061 will get a parse error, because "<literal>::*->*</literal>" is a
5062 single lexeme in Haskell.
5066 As part of the same extension, you can put kind annotations in types
5069 f :: (Int :: *) -> Int
5070 g :: forall a. a -> (a :: *)
5074 atype ::= '(' ctype '::' kind ')
5076 The parentheses are required.
5081 <sect2 id="universal-quantification">
5082 <title>Arbitrary-rank polymorphism
5086 Haskell type signatures are implicitly quantified. The new keyword <literal>forall</literal>
5087 allows us to say exactly what this means. For example:
5095 g :: forall b. (b -> b)
5097 The two are treated identically.
5101 However, GHC's type system supports <emphasis>arbitrary-rank</emphasis>
5102 explicit universal quantification in
5104 For example, all the following types are legal:
5106 f1 :: forall a b. a -> b -> a
5107 g1 :: forall a b. (Ord a, Eq b) => a -> b -> a
5109 f2 :: (forall a. a->a) -> Int -> Int
5110 g2 :: (forall a. Eq a => [a] -> a -> Bool) -> Int -> Int
5112 f3 :: ((forall a. a->a) -> Int) -> Bool -> Bool
5114 f4 :: Int -> (forall a. a -> a)
5116 Here, <literal>f1</literal> and <literal>g1</literal> are rank-1 types, and
5117 can be written in standard Haskell (e.g. <literal>f1 :: a->b->a</literal>).
5118 The <literal>forall</literal> makes explicit the universal quantification that
5119 is implicitly added by Haskell.
5122 The functions <literal>f2</literal> and <literal>g2</literal> have rank-2 types;
5123 the <literal>forall</literal> is on the left of a function arrow. As <literal>g2</literal>
5124 shows, the polymorphic type on the left of the function arrow can be overloaded.
5127 The function <literal>f3</literal> has a rank-3 type;
5128 it has rank-2 types on the left of a function arrow.
5131 GHC has three flags to control higher-rank types:
5134 <option>-XPolymorphicComponents</option>: data constructors (only) can have polymorphic argument types.
5137 <option>-XRank2Types</option>: any function (including data constructors) can have a rank-2 type.
5140 <option>-XRankNTypes</option>: any function (including data constructors) can have an arbitrary-rank type.
5141 That is, you can nest <literal>forall</literal>s
5142 arbitrarily deep in function arrows.
5143 In particular, a forall-type (also called a "type scheme"),
5144 including an operational type class context, is legal:
5146 <listitem> <para> On the left or right (see <literal>f4</literal>, for example)
5147 of a function arrow </para> </listitem>
5148 <listitem> <para> As the argument of a constructor, or type of a field, in a data type declaration. For
5149 example, any of the <literal>f1,f2,f3,g1,g2</literal> above would be valid
5150 field type signatures.</para> </listitem>
5151 <listitem> <para> As the type of an implicit parameter </para> </listitem>
5152 <listitem> <para> In a pattern type signature (see <xref linkend="scoped-type-variables"/>) </para> </listitem>
5156 Of course <literal>forall</literal> becomes a keyword; you can't use <literal>forall</literal> as
5157 a type variable any more!
5166 In a <literal>data</literal> or <literal>newtype</literal> declaration one can quantify
5167 the types of the constructor arguments. Here are several examples:
5173 data T a = T1 (forall b. b -> b -> b) a
5175 data MonadT m = MkMonad { return :: forall a. a -> m a,
5176 bind :: forall a b. m a -> (a -> m b) -> m b
5179 newtype Swizzle = MkSwizzle (Ord a => [a] -> [a])
5185 The constructors have rank-2 types:
5191 T1 :: forall a. (forall b. b -> b -> b) -> a -> T a
5192 MkMonad :: forall m. (forall a. a -> m a)
5193 -> (forall a b. m a -> (a -> m b) -> m b)
5195 MkSwizzle :: (Ord a => [a] -> [a]) -> Swizzle
5201 Notice that you don't need to use a <literal>forall</literal> if there's an
5202 explicit context. For example in the first argument of the
5203 constructor <function>MkSwizzle</function>, an implicit "<literal>forall a.</literal>" is
5204 prefixed to the argument type. The implicit <literal>forall</literal>
5205 quantifies all type variables that are not already in scope, and are
5206 mentioned in the type quantified over.
5210 As for type signatures, implicit quantification happens for non-overloaded
5211 types too. So if you write this:
5214 data T a = MkT (Either a b) (b -> b)
5217 it's just as if you had written this:
5220 data T a = MkT (forall b. Either a b) (forall b. b -> b)
5223 That is, since the type variable <literal>b</literal> isn't in scope, it's
5224 implicitly universally quantified. (Arguably, it would be better
5225 to <emphasis>require</emphasis> explicit quantification on constructor arguments
5226 where that is what is wanted. Feedback welcomed.)
5230 You construct values of types <literal>T1, MonadT, Swizzle</literal> by applying
5231 the constructor to suitable values, just as usual. For example,
5242 a3 = MkSwizzle reverse
5245 a4 = let r x = Just x
5252 mkTs :: (forall b. b -> b -> b) -> a -> [T a]
5253 mkTs f x y = [T1 f x, T1 f y]
5259 The type of the argument can, as usual, be more general than the type
5260 required, as <literal>(MkSwizzle reverse)</literal> shows. (<function>reverse</function>
5261 does not need the <literal>Ord</literal> constraint.)
5265 When you use pattern matching, the bound variables may now have
5266 polymorphic types. For example:
5272 f :: T a -> a -> (a, Char)
5273 f (T1 w k) x = (w k x, w 'c' 'd')
5275 g :: (Ord a, Ord b) => Swizzle -> [a] -> (a -> b) -> [b]
5276 g (MkSwizzle s) xs f = s (map f (s xs))
5278 h :: MonadT m -> [m a] -> m [a]
5279 h m [] = return m []
5280 h m (x:xs) = bind m x $ \y ->
5281 bind m (h m xs) $ \ys ->
5288 In the function <function>h</function> we use the record selectors <literal>return</literal>
5289 and <literal>bind</literal> to extract the polymorphic bind and return functions
5290 from the <literal>MonadT</literal> data structure, rather than using pattern
5296 <title>Type inference</title>
5299 In general, type inference for arbitrary-rank types is undecidable.
5300 GHC uses an algorithm proposed by Odersky and Laufer ("Putting type annotations to work", POPL'96)
5301 to get a decidable algorithm by requiring some help from the programmer.
5302 We do not yet have a formal specification of "some help" but the rule is this:
5305 <emphasis>For a lambda-bound or case-bound variable, x, either the programmer
5306 provides an explicit polymorphic type for x, or GHC's type inference will assume
5307 that x's type has no foralls in it</emphasis>.
5310 What does it mean to "provide" an explicit type for x? You can do that by
5311 giving a type signature for x directly, using a pattern type signature
5312 (<xref linkend="scoped-type-variables"/>), thus:
5314 \ f :: (forall a. a->a) -> (f True, f 'c')
5316 Alternatively, you can give a type signature to the enclosing
5317 context, which GHC can "push down" to find the type for the variable:
5319 (\ f -> (f True, f 'c')) :: (forall a. a->a) -> (Bool,Char)
5321 Here the type signature on the expression can be pushed inwards
5322 to give a type signature for f. Similarly, and more commonly,
5323 one can give a type signature for the function itself:
5325 h :: (forall a. a->a) -> (Bool,Char)
5326 h f = (f True, f 'c')
5328 You don't need to give a type signature if the lambda bound variable
5329 is a constructor argument. Here is an example we saw earlier:
5331 f :: T a -> a -> (a, Char)
5332 f (T1 w k) x = (w k x, w 'c' 'd')
5334 Here we do not need to give a type signature to <literal>w</literal>, because
5335 it is an argument of constructor <literal>T1</literal> and that tells GHC all
5342 <sect3 id="implicit-quant">
5343 <title>Implicit quantification</title>
5346 GHC performs implicit quantification as follows. <emphasis>At the top level (only) of
5347 user-written types, if and only if there is no explicit <literal>forall</literal>,
5348 GHC finds all the type variables mentioned in the type that are not already
5349 in scope, and universally quantifies them.</emphasis> For example, the following pairs are
5353 f :: forall a. a -> a
5360 h :: forall b. a -> b -> b
5366 Notice that GHC does <emphasis>not</emphasis> find the innermost possible quantification
5369 f :: (a -> a) -> Int
5371 f :: forall a. (a -> a) -> Int
5373 f :: (forall a. a -> a) -> Int
5376 g :: (Ord a => a -> a) -> Int
5377 -- MEANS the illegal type
5378 g :: forall a. (Ord a => a -> a) -> Int
5380 g :: (forall a. Ord a => a -> a) -> Int
5382 The latter produces an illegal type, which you might think is silly,
5383 but at least the rule is simple. If you want the latter type, you
5384 can write your for-alls explicitly. Indeed, doing so is strongly advised
5391 <sect2 id="impredicative-polymorphism">
5392 <title>Impredicative polymorphism
5394 <para>GHC supports <emphasis>impredicative polymorphism</emphasis>,
5395 enabled with <option>-XImpredicativeTypes</option>.
5397 that you can call a polymorphic function at a polymorphic type, and
5398 parameterise data structures over polymorphic types. For example:
5400 f :: Maybe (forall a. [a] -> [a]) -> Maybe ([Int], [Char])
5401 f (Just g) = Just (g [3], g "hello")
5404 Notice here that the <literal>Maybe</literal> type is parameterised by the
5405 <emphasis>polymorphic</emphasis> type <literal>(forall a. [a] ->
5408 <para>The technical details of this extension are described in the paper
5409 <ulink url="http://research.microsoft.com/%7Esimonpj/papers/boxy/">Boxy types:
5410 type inference for higher-rank types and impredicativity</ulink>,
5411 which appeared at ICFP 2006.
5415 <sect2 id="scoped-type-variables">
5416 <title>Lexically scoped type variables
5420 GHC supports <emphasis>lexically scoped type variables</emphasis>, without
5421 which some type signatures are simply impossible to write. For example:
5423 f :: forall a. [a] -> [a]
5429 The type signature for <literal>f</literal> brings the type variable <literal>a</literal> into scope,
5430 because of the explicit <literal>forall</literal> (<xref linkend="decl-type-sigs"/>).
5431 The type variables bound by a <literal>forall</literal> scope over
5432 the entire definition of the accompanying value declaration.
5433 In this example, the type variable <literal>a</literal> scopes over the whole
5434 definition of <literal>f</literal>, including over
5435 the type signature for <varname>ys</varname>.
5436 In Haskell 98 it is not possible to declare
5437 a type for <varname>ys</varname>; a major benefit of scoped type variables is that
5438 it becomes possible to do so.
5440 <para>Lexically-scoped type variables are enabled by
5441 <option>-XScopedTypeVariables</option>. This flag implies <option>-XRelaxedPolyRec</option>.
5443 <para>Note: GHC 6.6 contains substantial changes to the way that scoped type
5444 variables work, compared to earlier releases. Read this section
5448 <title>Overview</title>
5450 <para>The design follows the following principles
5452 <listitem><para>A scoped type variable stands for a type <emphasis>variable</emphasis>, and not for
5453 a <emphasis>type</emphasis>. (This is a change from GHC's earlier
5454 design.)</para></listitem>
5455 <listitem><para>Furthermore, distinct lexical type variables stand for distinct
5456 type variables. This means that every programmer-written type signature
5457 (including one that contains free scoped type variables) denotes a
5458 <emphasis>rigid</emphasis> type; that is, the type is fully known to the type
5459 checker, and no inference is involved.</para></listitem>
5460 <listitem><para>Lexical type variables may be alpha-renamed freely, without
5461 changing the program.</para></listitem>
5465 A <emphasis>lexically scoped type variable</emphasis> can be bound by:
5467 <listitem><para>A declaration type signature (<xref linkend="decl-type-sigs"/>)</para></listitem>
5468 <listitem><para>An expression type signature (<xref linkend="exp-type-sigs"/>)</para></listitem>
5469 <listitem><para>A pattern type signature (<xref linkend="pattern-type-sigs"/>)</para></listitem>
5470 <listitem><para>Class and instance declarations (<xref linkend="cls-inst-scoped-tyvars"/>)</para></listitem>
5474 In Haskell, a programmer-written type signature is implicitly quantified over
5475 its free type variables (<ulink
5476 url="http://www.haskell.org/onlinereport/decls.html#sect4.1.2">Section
5478 of the Haskell Report).
5479 Lexically scoped type variables affect this implicit quantification rules
5480 as follows: any type variable that is in scope is <emphasis>not</emphasis> universally
5481 quantified. For example, if type variable <literal>a</literal> is in scope,
5484 (e :: a -> a) means (e :: a -> a)
5485 (e :: b -> b) means (e :: forall b. b->b)
5486 (e :: a -> b) means (e :: forall b. a->b)
5494 <sect3 id="decl-type-sigs">
5495 <title>Declaration type signatures</title>
5496 <para>A declaration type signature that has <emphasis>explicit</emphasis>
5497 quantification (using <literal>forall</literal>) brings into scope the
5498 explicitly-quantified
5499 type variables, in the definition of the named function. For example:
5501 f :: forall a. [a] -> [a]
5502 f (x:xs) = xs ++ [ x :: a ]
5504 The "<literal>forall a</literal>" brings "<literal>a</literal>" into scope in
5505 the definition of "<literal>f</literal>".
5507 <para>This only happens if:
5509 <listitem><para> The quantification in <literal>f</literal>'s type
5510 signature is explicit. For example:
5513 g (x:xs) = xs ++ [ x :: a ]
5515 This program will be rejected, because "<literal>a</literal>" does not scope
5516 over the definition of "<literal>f</literal>", so "<literal>x::a</literal>"
5517 means "<literal>x::forall a. a</literal>" by Haskell's usual implicit
5518 quantification rules.
5520 <listitem><para> The signature gives a type for a function binding or a bare variable binding,
5521 not a pattern binding.
5524 f1 :: forall a. [a] -> [a]
5525 f1 (x:xs) = xs ++ [ x :: a ] -- OK
5527 f2 :: forall a. [a] -> [a]
5528 f2 = \(x:xs) -> xs ++ [ x :: a ] -- OK
5530 f3 :: forall a. [a] -> [a]
5531 Just f3 = Just (\(x:xs) -> xs ++ [ x :: a ]) -- Not OK!
5533 The binding for <literal>f3</literal> is a pattern binding, and so its type signature
5534 does not bring <literal>a</literal> into scope. However <literal>f1</literal> is a
5535 function binding, and <literal>f2</literal> binds a bare variable; in both cases
5536 the type signature brings <literal>a</literal> into scope.
5542 <sect3 id="exp-type-sigs">
5543 <title>Expression type signatures</title>
5545 <para>An expression type signature that has <emphasis>explicit</emphasis>
5546 quantification (using <literal>forall</literal>) brings into scope the
5547 explicitly-quantified
5548 type variables, in the annotated expression. For example:
5550 f = runST ( (op >>= \(x :: STRef s Int) -> g x) :: forall s. ST s Bool )
5552 Here, the type signature <literal>forall a. ST s Bool</literal> brings the
5553 type variable <literal>s</literal> into scope, in the annotated expression
5554 <literal>(op >>= \(x :: STRef s Int) -> g x)</literal>.
5559 <sect3 id="pattern-type-sigs">
5560 <title>Pattern type signatures</title>
5562 A type signature may occur in any pattern; this is a <emphasis>pattern type
5563 signature</emphasis>.
5566 -- f and g assume that 'a' is already in scope
5567 f = \(x::Int, y::a) -> x
5569 h ((x,y) :: (Int,Bool)) = (y,x)
5571 In the case where all the type variables in the pattern type signature are
5572 already in scope (i.e. bound by the enclosing context), matters are simple: the
5573 signature simply constrains the type of the pattern in the obvious way.
5576 Unlike expression and declaration type signatures, pattern type signatures are not implicitly generalised.
5577 The pattern in a <emphasis>pattern binding</emphasis> may only mention type variables
5578 that are already in scope. For example:
5580 f :: forall a. [a] -> (Int, [a])
5583 (ys::[a], n) = (reverse xs, length xs) -- OK
5584 zs::[a] = xs ++ ys -- OK
5586 Just (v::b) = ... -- Not OK; b is not in scope
5588 Here, the pattern signatures for <literal>ys</literal> and <literal>zs</literal>
5589 are fine, but the one for <literal>v</literal> is not because <literal>b</literal> is
5593 However, in all patterns <emphasis>other</emphasis> than pattern bindings, a pattern
5594 type signature may mention a type variable that is not in scope; in this case,
5595 <emphasis>the signature brings that type variable into scope</emphasis>.
5596 This is particularly important for existential data constructors. For example:
5598 data T = forall a. MkT [a]
5601 k (MkT [t::a]) = MkT t3
5605 Here, the pattern type signature <literal>(t::a)</literal> mentions a lexical type
5606 variable that is not already in scope. Indeed, it <emphasis>cannot</emphasis> already be in scope,
5607 because it is bound by the pattern match. GHC's rule is that in this situation
5608 (and only then), a pattern type signature can mention a type variable that is
5609 not already in scope; the effect is to bring it into scope, standing for the
5610 existentially-bound type variable.
5613 When a pattern type signature binds a type variable in this way, GHC insists that the
5614 type variable is bound to a <emphasis>rigid</emphasis>, or fully-known, type variable.
5615 This means that any user-written type signature always stands for a completely known type.
5618 If all this seems a little odd, we think so too. But we must have
5619 <emphasis>some</emphasis> way to bring such type variables into scope, else we
5620 could not name existentially-bound type variables in subsequent type signatures.
5623 This is (now) the <emphasis>only</emphasis> situation in which a pattern type
5624 signature is allowed to mention a lexical variable that is not already in
5626 For example, both <literal>f</literal> and <literal>g</literal> would be
5627 illegal if <literal>a</literal> was not already in scope.
5633 <!-- ==================== Commented out part about result type signatures
5635 <sect3 id="result-type-sigs">
5636 <title>Result type signatures</title>
5639 The result type of a function, lambda, or case expression alternative can be given a signature, thus:
5642 {- f assumes that 'a' is already in scope -}
5643 f x y :: [a] = [x,y,x]
5645 g = \ x :: [Int] -> [3,4]
5647 h :: forall a. [a] -> a
5651 The final <literal>:: [a]</literal> after the patterns of <literal>f</literal> gives the type of
5652 the result of the function. Similarly, the body of the lambda in the RHS of
5653 <literal>g</literal> is <literal>[Int]</literal>, and the RHS of the case
5654 alternative in <literal>h</literal> is <literal>a</literal>.
5656 <para> A result type signature never brings new type variables into scope.</para>
5658 There are a couple of syntactic wrinkles. First, notice that all three
5659 examples would parse quite differently with parentheses:
5661 {- f assumes that 'a' is already in scope -}
5662 f x (y :: [a]) = [x,y,x]
5664 g = \ (x :: [Int]) -> [3,4]
5666 h :: forall a. [a] -> a
5670 Now the signature is on the <emphasis>pattern</emphasis>; and
5671 <literal>h</literal> would certainly be ill-typed (since the pattern
5672 <literal>(y:ys)</literal> cannot have the type <literal>a</literal>.
5674 Second, to avoid ambiguity, the type after the “<literal>::</literal>” in a result
5675 pattern signature on a lambda or <literal>case</literal> must be atomic (i.e. a single
5676 token or a parenthesised type of some sort). To see why,
5677 consider how one would parse this:
5686 <sect3 id="cls-inst-scoped-tyvars">
5687 <title>Class and instance declarations</title>
5690 The type variables in the head of a <literal>class</literal> or <literal>instance</literal> declaration
5691 scope over the methods defined in the <literal>where</literal> part. For example:
5709 <sect2 id="typing-binds">
5710 <title>Generalised typing of mutually recursive bindings</title>
5713 The Haskell Report specifies that a group of bindings (at top level, or in a
5714 <literal>let</literal> or <literal>where</literal>) should be sorted into
5715 strongly-connected components, and then type-checked in dependency order
5716 (<ulink url="http://www.haskell.org/onlinereport/decls.html#sect4.5.1">Haskell
5717 Report, Section 4.5.1</ulink>).
5718 As each group is type-checked, any binders of the group that
5720 an explicit type signature are put in the type environment with the specified
5722 and all others are monomorphic until the group is generalised
5723 (<ulink url="http://www.haskell.org/onlinereport/decls.html#sect4.5.2">Haskell Report, Section 4.5.2</ulink>).
5726 <para>Following a suggestion of Mark Jones, in his paper
5727 <ulink url="http://citeseer.ist.psu.edu/424440.html">Typing Haskell in
5729 GHC implements a more general scheme. If <option>-XRelaxedPolyRec</option> is
5731 <emphasis>the dependency analysis ignores references to variables that have an explicit
5732 type signature</emphasis>.
5733 As a result of this refined dependency analysis, the dependency groups are smaller, and more bindings will
5734 typecheck. For example, consider:
5736 f :: Eq a => a -> Bool
5737 f x = (x == x) || g True || g "Yes"
5739 g y = (y <= y) || f True
5741 This is rejected by Haskell 98, but under Jones's scheme the definition for
5742 <literal>g</literal> is typechecked first, separately from that for
5743 <literal>f</literal>,
5744 because the reference to <literal>f</literal> in <literal>g</literal>'s right
5745 hand side is ignored by the dependency analysis. Then <literal>g</literal>'s
5746 type is generalised, to get
5748 g :: Ord a => a -> Bool
5750 Now, the definition for <literal>f</literal> is typechecked, with this type for
5751 <literal>g</literal> in the type environment.
5755 The same refined dependency analysis also allows the type signatures of
5756 mutually-recursive functions to have different contexts, something that is illegal in
5757 Haskell 98 (Section 4.5.2, last sentence). With
5758 <option>-XRelaxedPolyRec</option>
5759 GHC only insists that the type signatures of a <emphasis>refined</emphasis> group have identical
5760 type signatures; in practice this means that only variables bound by the same
5761 pattern binding must have the same context. For example, this is fine:
5763 f :: Eq a => a -> Bool
5764 f x = (x == x) || g True
5766 g :: Ord a => a -> Bool
5767 g y = (y <= y) || f True
5773 <!-- ==================== End of type system extensions ================= -->
5775 <!-- ====================== TEMPLATE HASKELL ======================= -->
5777 <sect1 id="template-haskell">
5778 <title>Template Haskell</title>
5780 <para>Template Haskell allows you to do compile-time meta-programming in
5783 the main technical innovations is discussed in "<ulink
5784 url="http://research.microsoft.com/~simonpj/papers/meta-haskell/">
5785 Template Meta-programming for Haskell</ulink>" (Proc Haskell Workshop 2002).
5788 There is a Wiki page about
5789 Template Haskell at <ulink url="http://www.haskell.org/haskellwiki/Template_Haskell">
5790 http://www.haskell.org/haskellwiki/Template_Haskell</ulink>, and that is the best place to look for
5794 url="http://www.haskell.org/ghc/docs/latest/html/libraries/index.html">online
5795 Haskell library reference material</ulink>
5796 (look for module <literal>Language.Haskell.TH</literal>).
5797 Many changes to the original design are described in
5798 <ulink url="http://research.microsoft.com/~simonpj/papers/meta-haskell/notes2.ps">
5799 Notes on Template Haskell version 2</ulink>.
5800 Not all of these changes are in GHC, however.
5803 <para> The first example from that paper is set out below (<xref linkend="th-example"/>)
5804 as a worked example to help get you started.
5808 The documentation here describes the realisation of Template Haskell in GHC. It is not detailed enough to
5809 understand Template Haskell; see the <ulink url="http://haskell.org/haskellwiki/Template_Haskell">
5814 <title>Syntax</title>
5816 <para> Template Haskell has the following new syntactic
5817 constructions. You need to use the flag
5818 <option>-XTemplateHaskell</option>
5819 <indexterm><primary><option>-XTemplateHaskell</option></primary>
5820 </indexterm>to switch these syntactic extensions on
5821 (<option>-XTemplateHaskell</option> is no longer implied by
5822 <option>-fglasgow-exts</option>).</para>
5826 A splice is written <literal>$x</literal>, where <literal>x</literal> is an
5827 identifier, or <literal>$(...)</literal>, where the "..." is an arbitrary expression.
5828 There must be no space between the "$" and the identifier or parenthesis. This use
5829 of "$" overrides its meaning as an infix operator, just as "M.x" overrides the meaning
5830 of "." as an infix operator. If you want the infix operator, put spaces around it.
5832 <para> A splice can occur in place of
5834 <listitem><para> an expression; the spliced expression must
5835 have type <literal>Q Exp</literal></para></listitem>
5836 <listitem><para> a list of top-level declarations; the spliced expression must have type <literal>Q [Dec]</literal></para></listitem>
5839 Inside a splice you can can only call functions defined in imported modules,
5840 not functions defined elsewhere in the same module.</listitem>
5844 A expression quotation is written in Oxford brackets, thus:
5846 <listitem><para> <literal>[| ... |]</literal>, where the "..." is an expression;
5847 the quotation has type <literal>Q Exp</literal>.</para></listitem>
5848 <listitem><para> <literal>[d| ... |]</literal>, where the "..." is a list of top-level declarations;
5849 the quotation has type <literal>Q [Dec]</literal>.</para></listitem>
5850 <listitem><para> <literal>[t| ... |]</literal>, where the "..." is a type;
5851 the quotation has type <literal>Q Typ</literal>.</para></listitem>
5852 </itemizedlist></para></listitem>
5855 A quasi-quotation can appear in either a pattern context or an
5856 expression context and is also written in Oxford brackets:
5858 <listitem><para> <literal>[:<replaceable>varid</replaceable>| ... |]</literal>,
5859 where the "..." is an arbitrary string; a full description of the
5860 quasi-quotation facility is given in <xref linkend="th-quasiquotation"/>.</para></listitem>
5861 </itemizedlist></para></listitem>
5864 A name can be quoted with either one or two prefix single quotes:
5866 <listitem><para> <literal>'f</literal> has type <literal>Name</literal>, and names the function <literal>f</literal>.
5867 Similarly <literal>'C</literal> has type <literal>Name</literal> and names the data constructor <literal>C</literal>.
5868 In general <literal>'</literal><replaceable>thing</replaceable> interprets <replaceable>thing</replaceable> in an expression context.
5870 <listitem><para> <literal>''T</literal> has type <literal>Name</literal>, and names the type constructor <literal>T</literal>.
5871 That is, <literal>''</literal><replaceable>thing</replaceable> interprets <replaceable>thing</replaceable> in a type context.
5874 These <literal>Names</literal> can be used to construct Template Haskell expressions, patterns, declarations etc. They
5875 may also be given as an argument to the <literal>reify</literal> function.
5881 (Compared to the original paper, there are many differences of detail.
5882 The syntax for a declaration splice uses "<literal>$</literal>" not "<literal>splice</literal>".
5883 The type of the enclosed expression must be <literal>Q [Dec]</literal>, not <literal>[Q Dec]</literal>.
5884 Type splices are not implemented, and neither are pattern splices or quotations.
5888 <sect2> <title> Using Template Haskell </title>
5892 The data types and monadic constructor functions for Template Haskell are in the library
5893 <literal>Language.Haskell.THSyntax</literal>.
5897 You can only run a function at compile time if it is imported from another module. That is,
5898 you can't define a function in a module, and call it from within a splice in the same module.
5899 (It would make sense to do so, but it's hard to implement.)
5903 You can only run a function at compile time if it is imported
5904 from another module <emphasis>that is not part of a mutually-recursive group of modules
5905 that includes the module currently being compiled</emphasis>. Furthermore, all of the modules of
5906 the mutually-recursive group must be reachable by non-SOURCE imports from the module where the
5907 splice is to be run.</para>
5909 For example, when compiling module A,
5910 you can only run Template Haskell functions imported from B if B does not import A (directly or indirectly).
5911 The reason should be clear: to run B we must compile and run A, but we are currently type-checking A.
5915 The flag <literal>-ddump-splices</literal> shows the expansion of all top-level splices as they happen.
5918 If you are building GHC from source, you need at least a stage-2 bootstrap compiler to
5919 run Template Haskell. A stage-1 compiler will reject the TH constructs. Reason: TH
5920 compiles and runs a program, and then looks at the result. So it's important that
5921 the program it compiles produces results whose representations are identical to
5922 those of the compiler itself.
5926 <para> Template Haskell works in any mode (<literal>--make</literal>, <literal>--interactive</literal>,
5927 or file-at-a-time). There used to be a restriction to the former two, but that restriction
5932 <sect2 id="th-example"> <title> A Template Haskell Worked Example </title>
5933 <para>To help you get over the confidence barrier, try out this skeletal worked example.
5934 First cut and paste the two modules below into "Main.hs" and "Printf.hs":</para>
5941 -- Import our template "pr"
5942 import Printf ( pr )
5944 -- The splice operator $ takes the Haskell source code
5945 -- generated at compile time by "pr" and splices it into
5946 -- the argument of "putStrLn".
5947 main = putStrLn ( $(pr "Hello") )
5953 -- Skeletal printf from the paper.
5954 -- It needs to be in a separate module to the one where
5955 -- you intend to use it.
5957 -- Import some Template Haskell syntax
5958 import Language.Haskell.TH
5960 -- Describe a format string
5961 data Format = D | S | L String
5963 -- Parse a format string. This is left largely to you
5964 -- as we are here interested in building our first ever
5965 -- Template Haskell program and not in building printf.
5966 parse :: String -> [Format]
5969 -- Generate Haskell source code from a parsed representation
5970 -- of the format string. This code will be spliced into
5971 -- the module which calls "pr", at compile time.
5972 gen :: [Format] -> Q Exp
5973 gen [D] = [| \n -> show n |]
5974 gen [S] = [| \s -> s |]
5975 gen [L s] = stringE s
5977 -- Here we generate the Haskell code for the splice
5978 -- from an input format string.
5979 pr :: String -> Q Exp
5980 pr s = gen (parse s)
5983 <para>Now run the compiler (here we are a Cygwin prompt on Windows):
5986 $ ghc --make -XTemplateHaskell main.hs -o main.exe
5989 <para>Run "main.exe" and here is your output:</para>
5999 <title>Using Template Haskell with Profiling</title>
6000 <indexterm><primary>profiling</primary><secondary>with Template Haskell</secondary></indexterm>
6002 <para>Template Haskell relies on GHC's built-in bytecode compiler and
6003 interpreter to run the splice expressions. The bytecode interpreter
6004 runs the compiled expression on top of the same runtime on which GHC
6005 itself is running; this means that the compiled code referred to by
6006 the interpreted expression must be compatible with this runtime, and
6007 in particular this means that object code that is compiled for
6008 profiling <emphasis>cannot</emphasis> be loaded and used by a splice
6009 expression, because profiled object code is only compatible with the
6010 profiling version of the runtime.</para>
6012 <para>This causes difficulties if you have a multi-module program
6013 containing Template Haskell code and you need to compile it for
6014 profiling, because GHC cannot load the profiled object code and use it
6015 when executing the splices. Fortunately GHC provides a workaround.
6016 The basic idea is to compile the program twice:</para>
6020 <para>Compile the program or library first the normal way, without
6021 <option>-prof</option><indexterm><primary><option>-prof</option></primary></indexterm>.</para>
6024 <para>Then compile it again with <option>-prof</option>, and
6025 additionally use <option>-osuf
6026 p_o</option><indexterm><primary><option>-osuf</option></primary></indexterm>
6027 to name the object files differently (you can choose any suffix
6028 that isn't the normal object suffix here). GHC will automatically
6029 load the object files built in the first step when executing splice
6030 expressions. If you omit the <option>-osuf</option> flag when
6031 building with <option>-prof</option> and Template Haskell is used,
6032 GHC will emit an error message. </para>
6037 <sect2 id="th-quasiquotation"> <title> Template Haskell Quasi-quotation </title>
6038 <para>Quasi-quotation allows patterns and expressions to be written using
6039 programmer-defined concrete syntax; the motivation behind the extension and
6040 several examples are documented in
6041 "<ulink url="http://www.eecs.harvard.edu/~mainland/ghc-quasiquoting/">Why It's
6042 Nice to be Quoted: Quasiquoting for Haskell</ulink>" (Proc Haskell Workshop
6043 2007). The example below shows how to write a quasiquoter for a simple
6044 expression language.</para>
6047 In the example, the quasiquoter <literal>expr</literal> is bound to a value of
6048 type <literal>Language.Haskell.TH.Quote.QuasiQuoter</literal> which contains two
6049 functions for quoting expressions and patterns, respectively. The first argument
6050 to each quoter is the (arbitrary) string enclosed in the Oxford brackets. The
6051 context of the quasi-quotation statement determines which of the two parsers is
6052 called: if the quasi-quotation occurs in an expression context, the expression
6053 parser is called, and if it occurs in a pattern context, the pattern parser is
6057 Note that in the example we make use of an antiquoted
6058 variable <literal>n</literal>, indicated by the syntax <literal>'int:n</literal>
6059 (this syntax for anti-quotation was defined by the parser's
6060 author, <emphasis>not</emphasis> by GHC). This binds <literal>n</literal> to the
6061 integer value argument of the constructor <literal>IntExpr</literal> when
6062 pattern matching. Please see the referenced paper for further details regarding
6063 anti-quotation as well as the description of a technique that uses SYB to
6064 leverage a single parser of type <literal>String -> a</literal> to generate both
6065 an expression parser that returns a value of type <literal>Q Exp</literal> and a
6066 pattern parser that returns a value of type <literal>Q Pat</literal>.
6069 <para>In general, a quasi-quote has the form
6070 <literal>[$<replaceable>quoter</replaceable>| <replaceable>string</replaceable> |]</literal>.
6071 The <replaceable>quoter</replaceable> must be the name of an imported quoter; it
6072 cannot be an arbitrary expression. The quoted <replaceable>string</replaceable>
6073 can be arbitrary, and may contain newlines.
6076 Quasiquoters must obey the same stage restrictions as Template Haskell, e.g., in
6077 the example, <literal>expr</literal> cannot be defined
6078 in <literal>Main.hs</literal> where it is used, but must be imported.
6089 main = do { print $ eval [$expr|1 + 2|]
6091 { [$expr|'int:n|] -> print n
6100 import qualified Language.Haskell.TH as TH
6101 import Language.Haskell.TH.Quote
6103 data Expr = IntExpr Integer
6104 | AntiIntExpr String
6105 | BinopExpr BinOp Expr Expr
6107 deriving(Show, Typeable, Data)
6113 deriving(Show, Typeable, Data)
6115 eval :: Expr -> Integer
6116 eval (IntExpr n) = n
6117 eval (BinopExpr op x y) = (opToFun op) (eval x) (eval y)
6124 expr = QuasiQuoter parseExprExp parseExprPat
6126 -- Parse an Expr, returning its representation as
6127 -- either a Q Exp or a Q Pat. See the referenced paper
6128 -- for how to use SYB to do this by writing a single
6129 -- parser of type String -> Expr instead of two
6130 -- separate parsers.
6132 parseExprExp :: String -> Q Exp
6135 parseExprPat :: String -> Q Pat
6139 <para>Now run the compiler:
6142 $ ghc --make -XQuasiQuotes Main.hs -o main
6145 <para>Run "main" and here is your output:</para>
6157 <!-- ===================== Arrow notation =================== -->
6159 <sect1 id="arrow-notation">
6160 <title>Arrow notation
6163 <para>Arrows are a generalization of monads introduced by John Hughes.
6164 For more details, see
6169 “Generalising Monads to Arrows”,
6170 John Hughes, in <citetitle>Science of Computer Programming</citetitle> 37,
6171 pp67–111, May 2000.
6172 The paper that introduced arrows: a friendly introduction, motivated with
6173 programming examples.
6179 “<ulink url="http://www.soi.city.ac.uk/~ross/papers/notation.html">A New Notation for Arrows</ulink>”,
6180 Ross Paterson, in <citetitle>ICFP</citetitle>, Sep 2001.
6181 Introduced the notation described here.
6187 “<ulink url="http://www.soi.city.ac.uk/~ross/papers/fop.html">Arrows and Computation</ulink>”,
6188 Ross Paterson, in <citetitle>The Fun of Programming</citetitle>,
6195 “<ulink url="http://www.cs.chalmers.se/~rjmh/afp-arrows.pdf">Programming with Arrows</ulink>”,
6196 John Hughes, in <citetitle>5th International Summer School on
6197 Advanced Functional Programming</citetitle>,
6198 <citetitle>Lecture Notes in Computer Science</citetitle> vol. 3622,
6200 This paper includes another introduction to the notation,
6201 with practical examples.
6207 “<ulink url="http://www.haskell.org/ghc/docs/papers/arrow-rules.pdf">Type and Translation Rules for Arrow Notation in GHC</ulink>”,
6208 Ross Paterson and Simon Peyton Jones, September 16, 2004.
6209 A terse enumeration of the formal rules used
6210 (extracted from comments in the source code).
6216 The arrows web page at
6217 <ulink url="http://www.haskell.org/arrows/"><literal>http://www.haskell.org/arrows/</literal></ulink>.
6222 With the <option>-XArrows</option> flag, GHC supports the arrow
6223 notation described in the second of these papers,
6224 translating it using combinators from the
6225 <ulink url="../libraries/base/Control-Arrow.html"><literal>Control.Arrow</literal></ulink>
6227 What follows is a brief introduction to the notation;
6228 it won't make much sense unless you've read Hughes's paper.
6231 <para>The extension adds a new kind of expression for defining arrows:
6233 <replaceable>exp</replaceable><superscript>10</superscript> ::= ...
6234 | proc <replaceable>apat</replaceable> -> <replaceable>cmd</replaceable>
6236 where <literal>proc</literal> is a new keyword.
6237 The variables of the pattern are bound in the body of the
6238 <literal>proc</literal>-expression,
6239 which is a new sort of thing called a <firstterm>command</firstterm>.
6240 The syntax of commands is as follows:
6242 <replaceable>cmd</replaceable> ::= <replaceable>exp</replaceable><superscript>10</superscript> -< <replaceable>exp</replaceable>
6243 | <replaceable>exp</replaceable><superscript>10</superscript> -<< <replaceable>exp</replaceable>
6244 | <replaceable>cmd</replaceable><superscript>0</superscript>
6246 with <replaceable>cmd</replaceable><superscript>0</superscript> up to
6247 <replaceable>cmd</replaceable><superscript>9</superscript> defined using
6248 infix operators as for expressions, and
6250 <replaceable>cmd</replaceable><superscript>10</superscript> ::= \ <replaceable>apat</replaceable> ... <replaceable>apat</replaceable> -> <replaceable>cmd</replaceable>
6251 | let <replaceable>decls</replaceable> in <replaceable>cmd</replaceable>
6252 | if <replaceable>exp</replaceable> then <replaceable>cmd</replaceable> else <replaceable>cmd</replaceable>
6253 | case <replaceable>exp</replaceable> of { <replaceable>calts</replaceable> }
6254 | do { <replaceable>cstmt</replaceable> ; ... <replaceable>cstmt</replaceable> ; <replaceable>cmd</replaceable> }
6255 | <replaceable>fcmd</replaceable>
6257 <replaceable>fcmd</replaceable> ::= <replaceable>fcmd</replaceable> <replaceable>aexp</replaceable>
6258 | ( <replaceable>cmd</replaceable> )
6259 | (| <replaceable>aexp</replaceable> <replaceable>cmd</replaceable> ... <replaceable>cmd</replaceable> |)
6261 <replaceable>cstmt</replaceable> ::= let <replaceable>decls</replaceable>
6262 | <replaceable>pat</replaceable> <- <replaceable>cmd</replaceable>
6263 | rec { <replaceable>cstmt</replaceable> ; ... <replaceable>cstmt</replaceable> [;] }
6264 | <replaceable>cmd</replaceable>
6266 where <replaceable>calts</replaceable> are like <replaceable>alts</replaceable>
6267 except that the bodies are commands instead of expressions.
6271 Commands produce values, but (like monadic computations)
6272 may yield more than one value,
6273 or none, and may do other things as well.
6274 For the most part, familiarity with monadic notation is a good guide to
6276 However the values of expressions, even monadic ones,
6277 are determined by the values of the variables they contain;
6278 this is not necessarily the case for commands.
6282 A simple example of the new notation is the expression
6284 proc x -> f -< x+1
6286 We call this a <firstterm>procedure</firstterm> or
6287 <firstterm>arrow abstraction</firstterm>.
6288 As with a lambda expression, the variable <literal>x</literal>
6289 is a new variable bound within the <literal>proc</literal>-expression.
6290 It refers to the input to the arrow.
6291 In the above example, <literal>-<</literal> is not an identifier but an
6292 new reserved symbol used for building commands from an expression of arrow
6293 type and an expression to be fed as input to that arrow.
6294 (The weird look will make more sense later.)
6295 It may be read as analogue of application for arrows.
6296 The above example is equivalent to the Haskell expression
6298 arr (\ x -> x+1) >>> f
6300 That would make no sense if the expression to the left of
6301 <literal>-<</literal> involves the bound variable <literal>x</literal>.
6302 More generally, the expression to the left of <literal>-<</literal>
6303 may not involve any <firstterm>local variable</firstterm>,
6304 i.e. a variable bound in the current arrow abstraction.
6305 For such a situation there is a variant <literal>-<<</literal>, as in
6307 proc x -> f x -<< x+1
6309 which is equivalent to
6311 arr (\ x -> (f x, x+1)) >>> app
6313 so in this case the arrow must belong to the <literal>ArrowApply</literal>
6315 Such an arrow is equivalent to a monad, so if you're using this form
6316 you may find a monadic formulation more convenient.
6320 <title>do-notation for commands</title>
6323 Another form of command is a form of <literal>do</literal>-notation.
6324 For example, you can write
6333 You can read this much like ordinary <literal>do</literal>-notation,
6334 but with commands in place of monadic expressions.
6335 The first line sends the value of <literal>x+1</literal> as an input to
6336 the arrow <literal>f</literal>, and matches its output against
6337 <literal>y</literal>.
6338 In the next line, the output is discarded.
6339 The arrow <function>returnA</function> is defined in the
6340 <ulink url="../libraries/base/Control-Arrow.html"><literal>Control.Arrow</literal></ulink>
6341 module as <literal>arr id</literal>.
6342 The above example is treated as an abbreviation for
6344 arr (\ x -> (x, x)) >>>
6345 first (arr (\ x -> x+1) >>> f) >>>
6346 arr (\ (y, x) -> (y, (x, y))) >>>
6347 first (arr (\ y -> 2*y) >>> g) >>>
6349 arr (\ (x, y) -> let z = x+y in ((x, z), z)) >>>
6350 first (arr (\ (x, z) -> x*z) >>> h) >>>
6351 arr (\ (t, z) -> t+z) >>>
6354 Note that variables not used later in the composition are projected out.
6355 After simplification using rewrite rules (see <xref linkend="rewrite-rules"/>)
6357 <ulink url="../libraries/base/Control-Arrow.html"><literal>Control.Arrow</literal></ulink>
6358 module, this reduces to
6360 arr (\ x -> (x+1, x)) >>>
6362 arr (\ (y, x) -> (2*y, (x, y))) >>>
6364 arr (\ (_, (x, y)) -> let z = x+y in (x*z, z)) >>>
6366 arr (\ (t, z) -> t+z)
6368 which is what you might have written by hand.
6369 With arrow notation, GHC keeps track of all those tuples of variables for you.
6373 Note that although the above translation suggests that
6374 <literal>let</literal>-bound variables like <literal>z</literal> must be
6375 monomorphic, the actual translation produces Core,
6376 so polymorphic variables are allowed.
6380 It's also possible to have mutually recursive bindings,
6381 using the new <literal>rec</literal> keyword, as in the following example:
6383 counter :: ArrowCircuit a => a Bool Int
6384 counter = proc reset -> do
6385 rec output <- returnA -< if reset then 0 else next
6386 next <- delay 0 -< output+1
6387 returnA -< output
6389 The translation of such forms uses the <function>loop</function> combinator,
6390 so the arrow concerned must belong to the <literal>ArrowLoop</literal> class.
6396 <title>Conditional commands</title>
6399 In the previous example, we used a conditional expression to construct the
6401 Sometimes we want to conditionally execute different commands, as in
6408 which is translated to
6410 arr (\ (x,y) -> if f x y then Left x else Right y) >>>
6411 (arr (\x -> x+1) >>> f) ||| (arr (\y -> y+2) >>> g)
6413 Since the translation uses <function>|||</function>,
6414 the arrow concerned must belong to the <literal>ArrowChoice</literal> class.
6418 There are also <literal>case</literal> commands, like
6424 y <- h -< (x1, x2)
6428 The syntax is the same as for <literal>case</literal> expressions,
6429 except that the bodies of the alternatives are commands rather than expressions.
6430 The translation is similar to that of <literal>if</literal> commands.
6436 <title>Defining your own control structures</title>
6439 As we're seen, arrow notation provides constructs,
6440 modelled on those for expressions,
6441 for sequencing, value recursion and conditionals.
6442 But suitable combinators,
6443 which you can define in ordinary Haskell,
6444 may also be used to build new commands out of existing ones.
6445 The basic idea is that a command defines an arrow from environments to values.
6446 These environments assign values to the free local variables of the command.
6447 Thus combinators that produce arrows from arrows
6448 may also be used to build commands from commands.
6449 For example, the <literal>ArrowChoice</literal> class includes a combinator
6451 ArrowChoice a => (<+>) :: a e c -> a e c -> a e c
6453 so we can use it to build commands:
6455 expr' = proc x -> do
6458 symbol Plus -< ()
6459 y <- term -< ()
6462 symbol Minus -< ()
6463 y <- term -< ()
6466 (The <literal>do</literal> on the first line is needed to prevent the first
6467 <literal><+> ...</literal> from being interpreted as part of the
6468 expression on the previous line.)
6469 This is equivalent to
6471 expr' = (proc x -> returnA -< x)
6472 <+> (proc x -> do
6473 symbol Plus -< ()
6474 y <- term -< ()
6476 <+> (proc x -> do
6477 symbol Minus -< ()
6478 y <- term -< ()
6481 It is essential that this operator be polymorphic in <literal>e</literal>
6482 (representing the environment input to the command
6483 and thence to its subcommands)
6484 and satisfy the corresponding naturality property
6486 arr k >>> (f <+> g) = (arr k >>> f) <+> (arr k >>> g)
6488 at least for strict <literal>k</literal>.
6489 (This should be automatic if you're not using <function>seq</function>.)
6490 This ensures that environments seen by the subcommands are environments
6491 of the whole command,
6492 and also allows the translation to safely trim these environments.
6493 The operator must also not use any variable defined within the current
6498 We could define our own operator
6500 untilA :: ArrowChoice a => a e () -> a e Bool -> a e ()
6501 untilA body cond = proc x ->
6502 b <- cond -< x
6503 if b then returnA -< ()
6506 untilA body cond -< x
6508 and use it in the same way.
6509 Of course this infix syntax only makes sense for binary operators;
6510 there is also a more general syntax involving special brackets:
6514 (|untilA (increment -< x+y) (within 0.5 -< x)|)
6521 <title>Primitive constructs</title>
6524 Some operators will need to pass additional inputs to their subcommands.
6525 For example, in an arrow type supporting exceptions,
6526 the operator that attaches an exception handler will wish to pass the
6527 exception that occurred to the handler.
6528 Such an operator might have a type
6530 handleA :: ... => a e c -> a (e,Ex) c -> a e c
6532 where <literal>Ex</literal> is the type of exceptions handled.
6533 You could then use this with arrow notation by writing a command
6535 body `handleA` \ ex -> handler
6537 so that if an exception is raised in the command <literal>body</literal>,
6538 the variable <literal>ex</literal> is bound to the value of the exception
6539 and the command <literal>handler</literal>,
6540 which typically refers to <literal>ex</literal>, is entered.
6541 Though the syntax here looks like a functional lambda,
6542 we are talking about commands, and something different is going on.
6543 The input to the arrow represented by a command consists of values for
6544 the free local variables in the command, plus a stack of anonymous values.
6545 In all the prior examples, this stack was empty.
6546 In the second argument to <function>handleA</function>,
6547 this stack consists of one value, the value of the exception.
6548 The command form of lambda merely gives this value a name.
6553 the values on the stack are paired to the right of the environment.
6554 So operators like <function>handleA</function> that pass
6555 extra inputs to their subcommands can be designed for use with the notation
6556 by pairing the values with the environment in this way.
6557 More precisely, the type of each argument of the operator (and its result)
6558 should have the form
6560 a (...(e,t1), ... tn) t
6562 where <replaceable>e</replaceable> is a polymorphic variable
6563 (representing the environment)
6564 and <replaceable>ti</replaceable> are the types of the values on the stack,
6565 with <replaceable>t1</replaceable> being the <quote>top</quote>.
6566 The polymorphic variable <replaceable>e</replaceable> must not occur in
6567 <replaceable>a</replaceable>, <replaceable>ti</replaceable> or
6568 <replaceable>t</replaceable>.
6569 However the arrows involved need not be the same.
6570 Here are some more examples of suitable operators:
6572 bracketA :: ... => a e b -> a (e,b) c -> a (e,c) d -> a e d
6573 runReader :: ... => a e c -> a' (e,State) c
6574 runState :: ... => a e c -> a' (e,State) (c,State)
6576 We can supply the extra input required by commands built with the last two
6577 by applying them to ordinary expressions, as in
6581 (|runReader (do { ... })|) s
6583 which adds <literal>s</literal> to the stack of inputs to the command
6584 built using <function>runReader</function>.
6588 The command versions of lambda abstraction and application are analogous to
6589 the expression versions.
6590 In particular, the beta and eta rules describe equivalences of commands.
6591 These three features (operators, lambda abstraction and application)
6592 are the core of the notation; everything else can be built using them,
6593 though the results would be somewhat clumsy.
6594 For example, we could simulate <literal>do</literal>-notation by defining
6596 bind :: Arrow a => a e b -> a (e,b) c -> a e c
6597 u `bind` f = returnA &&& u >>> f
6599 bind_ :: Arrow a => a e b -> a e c -> a e c
6600 u `bind_` f = u `bind` (arr fst >>> f)
6602 We could simulate <literal>if</literal> by defining
6604 cond :: ArrowChoice a => a e b -> a e b -> a (e,Bool) b
6605 cond f g = arr (\ (e,b) -> if b then Left e else Right e) >>> f ||| g
6612 <title>Differences with the paper</title>
6617 <para>Instead of a single form of arrow application (arrow tail) with two
6618 translations, the implementation provides two forms
6619 <quote><literal>-<</literal></quote> (first-order)
6620 and <quote><literal>-<<</literal></quote> (higher-order).
6625 <para>User-defined operators are flagged with banana brackets instead of
6626 a new <literal>form</literal> keyword.
6635 <title>Portability</title>
6638 Although only GHC implements arrow notation directly,
6639 there is also a preprocessor
6641 <ulink url="http://www.haskell.org/arrows/">arrows web page</ulink>)
6642 that translates arrow notation into Haskell 98
6643 for use with other Haskell systems.
6644 You would still want to check arrow programs with GHC;
6645 tracing type errors in the preprocessor output is not easy.
6646 Modules intended for both GHC and the preprocessor must observe some
6647 additional restrictions:
6652 The module must import
6653 <ulink url="../libraries/base/Control-Arrow.html"><literal>Control.Arrow</literal></ulink>.
6659 The preprocessor cannot cope with other Haskell extensions.
6660 These would have to go in separate modules.
6666 Because the preprocessor targets Haskell (rather than Core),
6667 <literal>let</literal>-bound variables are monomorphic.
6678 <!-- ==================== BANG PATTERNS ================= -->
6680 <sect1 id="bang-patterns">
6681 <title>Bang patterns
6682 <indexterm><primary>Bang patterns</primary></indexterm>
6684 <para>GHC supports an extension of pattern matching called <emphasis>bang
6685 patterns</emphasis>, written <literal>!<replaceable>pat</replaceable></literal>.
6686 Bang patterns are under consideration for Haskell Prime.
6688 url="http://hackage.haskell.org/trac/haskell-prime/wiki/BangPatterns">Haskell
6689 prime feature description</ulink> contains more discussion and examples
6690 than the material below.
6693 The key change is the addition of a new rule to the
6694 <ulink url="http://haskell.org/onlinereport/exps.html#sect3.17.2">semantics of pattern matching in the Haskell 98 report</ulink>.
6695 Add new bullet 10, saying: Matching the pattern <literal>!</literal><replaceable>pat</replaceable>
6696 against a value <replaceable>v</replaceable> behaves as follows:
6698 <listitem><para>if <replaceable>v</replaceable> is bottom, the match diverges</para></listitem>
6699 <listitem><para>otherwise, <replaceable>pat</replaceable> is matched against <replaceable>v</replaceable> </para></listitem>
6703 Bang patterns are enabled by the flag <option>-XBangPatterns</option>.
6706 <sect2 id="bang-patterns-informal">
6707 <title>Informal description of bang patterns
6710 The main idea is to add a single new production to the syntax of patterns:
6714 Matching an expression <literal>e</literal> against a pattern <literal>!p</literal> is done by first
6715 evaluating <literal>e</literal> (to WHNF) and then matching the result against <literal>p</literal>.
6720 This definition makes <literal>f1</literal> is strict in <literal>x</literal>,
6721 whereas without the bang it would be lazy.
6722 Bang patterns can be nested of course:
6726 Here, <literal>f2</literal> is strict in <literal>x</literal> but not in
6727 <literal>y</literal>.
6728 A bang only really has an effect if it precedes a variable or wild-card pattern:
6733 Here, <literal>f3</literal> and <literal>f4</literal> are identical;
6734 putting a bang before a pattern that
6735 forces evaluation anyway does nothing.
6738 There is one (apparent) exception to this general rule that a bang only
6739 makes a difference when it precedes a variable or wild-card: a bang at the
6740 top level of a <literal>let</literal> or <literal>where</literal>
6741 binding makes the binding strict, regardless of the pattern. For example:
6745 is a strict binding: operationally, it evaluates <literal>e</literal>, matches
6746 it against the pattern <literal>[x,y]</literal>, and then evaluates <literal>b</literal>.
6747 (We say "apparent" exception because the Right Way to think of it is that the bang
6748 at the top of a binding is not part of the <emphasis>pattern</emphasis>; rather it
6749 is part of the syntax of the <emphasis>binding</emphasis>.)
6750 Nested bangs in a pattern binding behave uniformly with all other forms of
6751 pattern matching. For example
6753 let (!x,[y]) = e in b
6755 is equivalent to this:
6757 let { t = case e of (x,[y]) -> x `seq` (x,y)
6762 The binding is lazy, but when either <literal>x</literal> or <literal>y</literal> is
6763 evaluated by <literal>b</literal> the entire pattern is matched, including forcing the
6764 evaluation of <literal>x</literal>.
6767 Bang patterns work in <literal>case</literal> expressions too, of course:
6769 g5 x = let y = f x in body
6770 g6 x = case f x of { y -> body }
6771 g7 x = case f x of { !y -> body }
6773 The functions <literal>g5</literal> and <literal>g6</literal> mean exactly the same thing.
6774 But <literal>g7</literal> evaluates <literal>(f x)</literal>, binds <literal>y</literal> to the
6775 result, and then evaluates <literal>body</literal>.
6780 <sect2 id="bang-patterns-sem">
6781 <title>Syntax and semantics
6785 We add a single new production to the syntax of patterns:
6789 There is one problem with syntactic ambiguity. Consider:
6793 Is this a definition of the infix function "<literal>(!)</literal>",
6794 or of the "<literal>f</literal>" with a bang pattern? GHC resolves this
6795 ambiguity in favour of the latter. If you want to define
6796 <literal>(!)</literal> with bang-patterns enabled, you have to do so using
6801 The semantics of Haskell pattern matching is described in <ulink
6802 url="http://www.haskell.org/onlinereport/exps.html#sect3.17.2">
6803 Section 3.17.2</ulink> of the Haskell Report. To this description add
6804 one extra item 10, saying:
6805 <itemizedlist><listitem><para>Matching
6806 the pattern <literal>!pat</literal> against a value <literal>v</literal> behaves as follows:
6807 <itemizedlist><listitem><para>if <literal>v</literal> is bottom, the match diverges</para></listitem>
6808 <listitem><para>otherwise, <literal>pat</literal> is matched against
6809 <literal>v</literal></para></listitem>
6811 </para></listitem></itemizedlist>
6812 Similarly, in Figure 4 of <ulink url="http://www.haskell.org/onlinereport/exps.html#sect3.17.3">
6813 Section 3.17.3</ulink>, add a new case (t):
6815 case v of { !pat -> e; _ -> e' }
6816 = v `seq` case v of { pat -> e; _ -> e' }
6819 That leaves let expressions, whose translation is given in
6820 <ulink url="http://www.haskell.org/onlinereport/exps.html#sect3.12">Section
6822 of the Haskell Report.
6823 In the translation box, first apply
6824 the following transformation: for each pattern <literal>pi</literal> that is of
6825 form <literal>!qi = ei</literal>, transform it to <literal>(xi,!qi) = ((),ei)</literal>, and and replace <literal>e0</literal>
6826 by <literal>(xi `seq` e0)</literal>. Then, when none of the left-hand-side patterns
6827 have a bang at the top, apply the rules in the existing box.
6829 <para>The effect of the let rule is to force complete matching of the pattern
6830 <literal>qi</literal> before evaluation of the body is begun. The bang is
6831 retained in the translated form in case <literal>qi</literal> is a variable,
6839 The let-binding can be recursive. However, it is much more common for
6840 the let-binding to be non-recursive, in which case the following law holds:
6841 <literal>(let !p = rhs in body)</literal>
6843 <literal>(case rhs of !p -> body)</literal>
6846 A pattern with a bang at the outermost level is not allowed at the top level of
6852 <!-- ==================== ASSERTIONS ================= -->
6854 <sect1 id="assertions">
6856 <indexterm><primary>Assertions</primary></indexterm>
6860 If you want to make use of assertions in your standard Haskell code, you
6861 could define a function like the following:
6867 assert :: Bool -> a -> a
6868 assert False x = error "assertion failed!"
6875 which works, but gives you back a less than useful error message --
6876 an assertion failed, but which and where?
6880 One way out is to define an extended <function>assert</function> function which also
6881 takes a descriptive string to include in the error message and
6882 perhaps combine this with the use of a pre-processor which inserts
6883 the source location where <function>assert</function> was used.
6887 Ghc offers a helping hand here, doing all of this for you. For every
6888 use of <function>assert</function> in the user's source:
6894 kelvinToC :: Double -> Double
6895 kelvinToC k = assert (k >= 0.0) (k+273.15)
6901 Ghc will rewrite this to also include the source location where the
6908 assert pred val ==> assertError "Main.hs|15" pred val
6914 The rewrite is only performed by the compiler when it spots
6915 applications of <function>Control.Exception.assert</function>, so you
6916 can still define and use your own versions of
6917 <function>assert</function>, should you so wish. If not, import
6918 <literal>Control.Exception</literal> to make use
6919 <function>assert</function> in your code.
6923 GHC ignores assertions when optimisation is turned on with the
6924 <option>-O</option><indexterm><primary><option>-O</option></primary></indexterm> flag. That is, expressions of the form
6925 <literal>assert pred e</literal> will be rewritten to
6926 <literal>e</literal>. You can also disable assertions using the
6927 <option>-fignore-asserts</option>
6928 option<indexterm><primary><option>-fignore-asserts</option></primary>
6929 </indexterm>.</para>
6932 Assertion failures can be caught, see the documentation for the
6933 <literal>Control.Exception</literal> library for the details.
6939 <!-- =============================== PRAGMAS =========================== -->
6941 <sect1 id="pragmas">
6942 <title>Pragmas</title>
6944 <indexterm><primary>pragma</primary></indexterm>
6946 <para>GHC supports several pragmas, or instructions to the
6947 compiler placed in the source code. Pragmas don't normally affect
6948 the meaning of the program, but they might affect the efficiency
6949 of the generated code.</para>
6951 <para>Pragmas all take the form
6953 <literal>{-# <replaceable>word</replaceable> ... #-}</literal>
6955 where <replaceable>word</replaceable> indicates the type of
6956 pragma, and is followed optionally by information specific to that
6957 type of pragma. Case is ignored in
6958 <replaceable>word</replaceable>. The various values for
6959 <replaceable>word</replaceable> that GHC understands are described
6960 in the following sections; any pragma encountered with an
6961 unrecognised <replaceable>word</replaceable> is
6962 ignored. The layout rule applies in pragmas, so the closing <literal>#-}</literal>
6963 should start in a column to the right of the opening <literal>{-#</literal>. </para>
6965 <para>Certain pragmas are <emphasis>file-header pragmas</emphasis>:
6969 pragma must precede the <literal>module</literal> keyword in the file.
6972 There can be as many file-header pragmas as you please, and they can be
6973 preceded or followed by comments.
6976 File-header pragmas are read once only, before
6977 pre-processing the file (e.g. with cpp).
6980 The file-header pragmas are: <literal>{-# LANGUAGE #-}</literal>,
6981 <literal>{-# OPTIONS_GHC #-}</literal>, and
6982 <literal>{-# INCLUDE #-}</literal>.
6987 <sect2 id="language-pragma">
6988 <title>LANGUAGE pragma</title>
6990 <indexterm><primary>LANGUAGE</primary><secondary>pragma</secondary></indexterm>
6991 <indexterm><primary>pragma</primary><secondary>LANGUAGE</secondary></indexterm>
6993 <para>The <literal>LANGUAGE</literal> pragma allows language extensions to be enabled
6995 It is the intention that all Haskell compilers support the
6996 <literal>LANGUAGE</literal> pragma with the same syntax, although not
6997 all extensions are supported by all compilers, of
6998 course. The <literal>LANGUAGE</literal> pragma should be used instead
6999 of <literal>OPTIONS_GHC</literal>, if possible.</para>
7001 <para>For example, to enable the FFI and preprocessing with CPP:</para>
7003 <programlisting>{-# LANGUAGE ForeignFunctionInterface, CPP #-}</programlisting>
7005 <para><literal>LANGUAGE</literal> is a file-header pragma (see <xref linkend="pragmas"/>).</para>
7007 <para>Every language extension can also be turned into a command-line flag
7008 by prefixing it with "<literal>-X</literal>"; for example <option>-XForeignFunctionInterface</option>.
7009 (Similarly, all "<literal>-X</literal>" flags can be written as <literal>LANGUAGE</literal> pragmas.
7012 <para>A list of all supported language extensions can be obtained by invoking
7013 <literal>ghc --supported-languages</literal> (see <xref linkend="modes"/>).</para>
7015 <para>Any extension from the <literal>Extension</literal> type defined in
7017 url="../libraries/Cabal/Language-Haskell-Extension.html"><literal>Language.Haskell.Extension</literal></ulink>
7018 may be used. GHC will report an error if any of the requested extensions are not supported.</para>
7022 <sect2 id="options-pragma">
7023 <title>OPTIONS_GHC pragma</title>
7024 <indexterm><primary>OPTIONS_GHC</primary>
7026 <indexterm><primary>pragma</primary><secondary>OPTIONS_GHC</secondary>
7029 <para>The <literal>OPTIONS_GHC</literal> pragma is used to specify
7030 additional options that are given to the compiler when compiling
7031 this source file. See <xref linkend="source-file-options"/> for
7034 <para>Previous versions of GHC accepted <literal>OPTIONS</literal> rather
7035 than <literal>OPTIONS_GHC</literal>, but that is now deprecated.</para>
7038 <para><literal>OPTIONS_GHC</literal> is a file-header pragma (see <xref linkend="pragmas"/>).</para>
7040 <sect2 id="include-pragma">
7041 <title>INCLUDE pragma</title>
7043 <para>The <literal>INCLUDE</literal> pragma is for specifying the names
7044 of C header files that should be <literal>#include</literal>'d into
7045 the C source code generated by the compiler for the current module (if
7046 compiling via C). For example:</para>
7049 {-# INCLUDE "foo.h" #-}
7050 {-# INCLUDE <stdio.h> #-}</programlisting>
7052 <para><literal>INCLUDE</literal> is a file-header pragma (see <xref linkend="pragmas"/>).</para>
7054 <para>An <literal>INCLUDE</literal> pragma is the preferred alternative
7055 to the <option>-#include</option> option (<xref
7056 linkend="options-C-compiler" />), because the
7057 <literal>INCLUDE</literal> pragma is understood by other
7058 compilers. Yet another alternative is to add the include file to each
7059 <literal>foreign import</literal> declaration in your code, but we
7060 don't recommend using this approach with GHC.</para>
7063 <sect2 id="warning-deprecated-pragma">
7064 <title>WARNING and DEPRECATED pragmas</title>
7065 <indexterm><primary>WARNING</primary></indexterm>
7066 <indexterm><primary>DEPRECATED</primary></indexterm>
7068 <para>The WARNING pragma allows you to attach an arbitrary warning
7069 to a particular function, class, or type.
7070 A DEPRECATED pragma lets you specify that
7071 a particular function, class, or type is deprecated.
7072 There are two ways of using these pragmas.
7076 <para>You can work on an entire module thus:</para>
7078 module Wibble {-# DEPRECATED "Use Wobble instead" #-} where
7083 module Wibble {-# WARNING "This is an unstable interface." #-} where
7086 <para>When you compile any module that import
7087 <literal>Wibble</literal>, GHC will print the specified
7092 <para>You can attach a warning to a function, class, type, or data constructor, with the
7093 following top-level declarations:</para>
7095 {-# DEPRECATED f, C, T "Don't use these" #-}
7096 {-# WARNING unsafePerformIO "This is unsafe; I hope you know what you're doing" #-}
7098 <para>When you compile any module that imports and uses any
7099 of the specified entities, GHC will print the specified
7101 <para> You can only attach to entities declared at top level in the module
7102 being compiled, and you can only use unqualified names in the list of
7103 entities. A capitalised name, such as <literal>T</literal>
7104 refers to <emphasis>either</emphasis> the type constructor <literal>T</literal>
7105 <emphasis>or</emphasis> the data constructor <literal>T</literal>, or both if
7106 both are in scope. If both are in scope, there is currently no way to
7107 specify one without the other (c.f. fixities
7108 <xref linkend="infix-tycons"/>).</para>
7111 Warnings and deprecations are not reported for
7112 (a) uses within the defining module, and
7113 (b) uses in an export list.
7114 The latter reduces spurious complaints within a library
7115 in which one module gathers together and re-exports
7116 the exports of several others.
7118 <para>You can suppress the warnings with the flag
7119 <option>-fno-warn-warnings-deprecations</option>.</para>
7122 <sect2 id="inline-noinline-pragma">
7123 <title>INLINE and NOINLINE pragmas</title>
7125 <para>These pragmas control the inlining of function
7128 <sect3 id="inline-pragma">
7129 <title>INLINE pragma</title>
7130 <indexterm><primary>INLINE</primary></indexterm>
7132 <para>GHC (with <option>-O</option>, as always) tries to
7133 inline (or “unfold”) functions/values that are
7134 “small enough,” thus avoiding the call overhead
7135 and possibly exposing other more-wonderful optimisations.
7136 Normally, if GHC decides a function is “too
7137 expensive” to inline, it will not do so, nor will it
7138 export that unfolding for other modules to use.</para>
7140 <para>The sledgehammer you can bring to bear is the
7141 <literal>INLINE</literal><indexterm><primary>INLINE
7142 pragma</primary></indexterm> pragma, used thusly:</para>
7145 key_function :: Int -> String -> (Bool, Double)
7146 {-# INLINE key_function #-}
7149 <para>The major effect of an <literal>INLINE</literal> pragma
7150 is to declare a function's “cost” to be very low.
7151 The normal unfolding machinery will then be very keen to
7152 inline it. However, an <literal>INLINE</literal> pragma for a
7153 function "<literal>f</literal>" has a number of other effects:
7156 No functions are inlined into <literal>f</literal>. Otherwise
7157 GHC might inline a big function into <literal>f</literal>'s right hand side,
7158 making <literal>f</literal> big; and then inline <literal>f</literal> blindly.
7161 The float-in, float-out, and common-sub-expression transformations are not
7162 applied to the body of <literal>f</literal>.
7165 An INLINE function is not worker/wrappered by strictness analysis.
7166 It's going to be inlined wholesale instead.
7169 All of these effects are aimed at ensuring that what gets inlined is
7170 exactly what you asked for, no more and no less.
7172 <para>GHC ensures that inlining cannot go on forever: every mutually-recursive
7173 group is cut by one or more <emphasis>loop breakers</emphasis> that is never inlined
7174 (see <ulink url="http://research.microsoft.com/%7Esimonpj/Papers/inlining/index.htm">
7175 Secrets of the GHC inliner, JFP 12(4) July 2002</ulink>).
7176 GHC tries not to select a function with an INLINE pragma as a loop breaker, but
7177 when there is no choice even an INLINE function can be selected, in which case
7178 the INLINE pragma is ignored.
7179 For example, for a self-recursive function, the loop breaker can only be the function
7180 itself, so an INLINE pragma is always ignored.</para>
7182 <para>Syntactically, an <literal>INLINE</literal> pragma for a
7183 function can be put anywhere its type signature could be
7186 <para><literal>INLINE</literal> pragmas are a particularly
7188 <literal>then</literal>/<literal>return</literal> (or
7189 <literal>bind</literal>/<literal>unit</literal>) functions in
7190 a monad. For example, in GHC's own
7191 <literal>UniqueSupply</literal> monad code, we have:</para>
7194 {-# INLINE thenUs #-}
7195 {-# INLINE returnUs #-}
7198 <para>See also the <literal>NOINLINE</literal> pragma (<xref
7199 linkend="noinline-pragma"/>).</para>
7201 <para>Note: the HBC compiler doesn't like <literal>INLINE</literal> pragmas,
7202 so if you want your code to be HBC-compatible you'll have to surround
7203 the pragma with C pre-processor directives
7204 <literal>#ifdef __GLASGOW_HASKELL__</literal>...<literal>#endif</literal>.</para>
7208 <sect3 id="noinline-pragma">
7209 <title>NOINLINE pragma</title>
7211 <indexterm><primary>NOINLINE</primary></indexterm>
7212 <indexterm><primary>NOTINLINE</primary></indexterm>
7214 <para>The <literal>NOINLINE</literal> pragma does exactly what
7215 you'd expect: it stops the named function from being inlined
7216 by the compiler. You shouldn't ever need to do this, unless
7217 you're very cautious about code size.</para>
7219 <para><literal>NOTINLINE</literal> is a synonym for
7220 <literal>NOINLINE</literal> (<literal>NOINLINE</literal> is
7221 specified by Haskell 98 as the standard way to disable
7222 inlining, so it should be used if you want your code to be
7226 <sect3 id="phase-control">
7227 <title>Phase control</title>
7229 <para> Sometimes you want to control exactly when in GHC's
7230 pipeline the INLINE pragma is switched on. Inlining happens
7231 only during runs of the <emphasis>simplifier</emphasis>. Each
7232 run of the simplifier has a different <emphasis>phase
7233 number</emphasis>; the phase number decreases towards zero.
7234 If you use <option>-dverbose-core2core</option> you'll see the
7235 sequence of phase numbers for successive runs of the
7236 simplifier. In an INLINE pragma you can optionally specify a
7240 <para>"<literal>INLINE[k] f</literal>" means: do not inline
7241 <literal>f</literal>
7242 until phase <literal>k</literal>, but from phase
7243 <literal>k</literal> onwards be very keen to inline it.
7246 <para>"<literal>INLINE[~k] f</literal>" means: be very keen to inline
7247 <literal>f</literal>
7248 until phase <literal>k</literal>, but from phase
7249 <literal>k</literal> onwards do not inline it.
7252 <para>"<literal>NOINLINE[k] f</literal>" means: do not inline
7253 <literal>f</literal>
7254 until phase <literal>k</literal>, but from phase
7255 <literal>k</literal> onwards be willing to inline it (as if
7256 there was no pragma).
7259 <para>"<literal>NOINLINE[~k] f</literal>" means: be willing to inline
7260 <literal>f</literal>
7261 until phase <literal>k</literal>, but from phase
7262 <literal>k</literal> onwards do not inline it.
7265 The same information is summarised here:
7267 -- Before phase 2 Phase 2 and later
7268 {-# INLINE [2] f #-} -- No Yes
7269 {-# INLINE [~2] f #-} -- Yes No
7270 {-# NOINLINE [2] f #-} -- No Maybe
7271 {-# NOINLINE [~2] f #-} -- Maybe No
7273 {-# INLINE f #-} -- Yes Yes
7274 {-# NOINLINE f #-} -- No No
7276 By "Maybe" we mean that the usual heuristic inlining rules apply (if the
7277 function body is small, or it is applied to interesting-looking arguments etc).
7278 Another way to understand the semantics is this:
7280 <listitem><para>For both INLINE and NOINLINE, the phase number says
7281 when inlining is allowed at all.</para></listitem>
7282 <listitem><para>The INLINE pragma has the additional effect of making the
7283 function body look small, so that when inlining is allowed it is very likely to
7288 <para>The same phase-numbering control is available for RULES
7289 (<xref linkend="rewrite-rules"/>).</para>
7293 <sect2 id="annotation-pragmas">
7294 <title>ANN pragmas</title>
7296 <para>GHC offers the ability to annotate various code constructs with additional
7297 data by using three pragmas. This data can then be inspected at a later date by
7298 using GHC-as-a-library.</para>
7300 <sect3 id="ann-pragma">
7301 <title>Annotating values</title>
7303 <indexterm><primary>ANN</primary></indexterm>
7305 <para>Any expression that has both <literal>Typeable</literal> and <literal>Data</literal> instances may be attached to a top-level value
7306 binding using an <literal>ANN</literal> pragma. In particular, this means you can use <literal>ANN</literal>
7307 to annotate data constructors (e.g. <literal>Just</literal>) as well as normal values (e.g. <literal>take</literal>).
7308 By way of example, to annotate the function <literal>foo</literal> with the annotation <literal>Just "Hello"</literal>
7309 you would do this:</para>
7312 {-# ANN foo (Just "Hello") #-}
7317 A number of restrictions apply to use of annotations:
7319 <listitem><para>The binder being annotated must be at the top level (i.e. no nested binders)</para></listitem>
7320 <listitem><para>The binder being annotated must be declared in the current module</para></listitem>
7321 <listitem><para>The expression you are annotating with must have a type with <literal>Typeable</literal> and <literal>Data</literal> instances</para></listitem>
7322 <listitem><para>The <ulink linkend="using-template-haskell">Template Haskell staging restrictions</ulink> apply to the
7323 expression being annotated with, so for example you cannot run a function from the module being compiled.</para>
7325 <para>To be precise, the annotation <literal>{-# ANN x e #-}</literal> is well staged if and only if <literal>$(e)</literal> would be
7326 (disregarding the usual type restrictions of the splice syntax, and the usual restriction on splicing inside a splice - <literal>$([|1|])</literal> is fine as an annotation, albeit redundant).</para></listitem>
7329 If you feel strongly that any of these restrictions are too onerous, <ulink url="http://hackage.haskell.org/trac/ghc/wiki/MailingListsAndIRC">
7330 please give the GHC team a shout</ulink>.
7333 <para>However, apart from these restrictions, many things are allowed, including expressions which are not fully evaluated!
7334 Annotation expressions will be evaluated by the compiler just like Template Haskell splices are. So, this annotation is fine:</para>
7337 {-# ANN f SillyAnnotation { foo = (id 10) + $([| 20 |]), bar = 'f } #-}
7342 <sect3 id="typeann-pragma">
7343 <title>Annotating types</title>
7345 <indexterm><primary>ANN type</primary></indexterm>
7346 <indexterm><primary>ANN</primary></indexterm>
7348 <para>You can annotate types with the <literal>ANN</literal> pragma by using the <literal>type</literal> keyword. For example:</para>
7351 {-# ANN type Foo (Just "A `Maybe String' annotation") #-}
7356 <sect3 id="modann-pragma">
7357 <title>Annotating modules</title>
7359 <indexterm><primary>ANN module</primary></indexterm>
7360 <indexterm><primary>ANN</primary></indexterm>
7362 <para>You can annotate modules with the <literal>ANN</literal> pragma by using the <literal>module</literal> keyword. For example:</para>
7365 {-# ANN module (Just "A `Maybe String' annotation") #-}
7370 <sect2 id="line-pragma">
7371 <title>LINE pragma</title>
7373 <indexterm><primary>LINE</primary><secondary>pragma</secondary></indexterm>
7374 <indexterm><primary>pragma</primary><secondary>LINE</secondary></indexterm>
7375 <para>This pragma is similar to C's <literal>#line</literal>
7376 pragma, and is mainly for use in automatically generated Haskell
7377 code. It lets you specify the line number and filename of the
7378 original code; for example</para>
7380 <programlisting>{-# LINE 42 "Foo.vhs" #-}</programlisting>
7382 <para>if you'd generated the current file from something called
7383 <filename>Foo.vhs</filename> and this line corresponds to line
7384 42 in the original. GHC will adjust its error messages to refer
7385 to the line/file named in the <literal>LINE</literal>
7390 <title>RULES pragma</title>
7392 <para>The RULES pragma lets you specify rewrite rules. It is
7393 described in <xref linkend="rewrite-rules"/>.</para>
7396 <sect2 id="specialize-pragma">
7397 <title>SPECIALIZE pragma</title>
7399 <indexterm><primary>SPECIALIZE pragma</primary></indexterm>
7400 <indexterm><primary>pragma, SPECIALIZE</primary></indexterm>
7401 <indexterm><primary>overloading, death to</primary></indexterm>
7403 <para>(UK spelling also accepted.) For key overloaded
7404 functions, you can create extra versions (NB: more code space)
7405 specialised to particular types. Thus, if you have an
7406 overloaded function:</para>
7409 hammeredLookup :: Ord key => [(key, value)] -> key -> value
7412 <para>If it is heavily used on lists with
7413 <literal>Widget</literal> keys, you could specialise it as
7417 {-# SPECIALIZE hammeredLookup :: [(Widget, value)] -> Widget -> value #-}
7420 <para>A <literal>SPECIALIZE</literal> pragma for a function can
7421 be put anywhere its type signature could be put.</para>
7423 <para>A <literal>SPECIALIZE</literal> has the effect of generating
7424 (a) a specialised version of the function and (b) a rewrite rule
7425 (see <xref linkend="rewrite-rules"/>) that rewrites a call to the
7426 un-specialised function into a call to the specialised one.</para>
7428 <para>The type in a SPECIALIZE pragma can be any type that is less
7429 polymorphic than the type of the original function. In concrete terms,
7430 if the original function is <literal>f</literal> then the pragma
7432 {-# SPECIALIZE f :: <type> #-}
7434 is valid if and only if the definition
7436 f_spec :: <type>
7439 is valid. Here are some examples (where we only give the type signature
7440 for the original function, not its code):
7442 f :: Eq a => a -> b -> b
7443 {-# SPECIALISE f :: Int -> b -> b #-}
7445 g :: (Eq a, Ix b) => a -> b -> b
7446 {-# SPECIALISE g :: (Eq a) => a -> Int -> Int #-}
7448 h :: Eq a => a -> a -> a
7449 {-# SPECIALISE h :: (Eq a) => [a] -> [a] -> [a] #-}
7451 The last of these examples will generate a
7452 RULE with a somewhat-complex left-hand side (try it yourself), so it might not fire very
7453 well. If you use this kind of specialisation, let us know how well it works.
7456 <para>A <literal>SPECIALIZE</literal> pragma can optionally be followed with a
7457 <literal>INLINE</literal> or <literal>NOINLINE</literal> pragma, optionally
7458 followed by a phase, as described in <xref linkend="inline-noinline-pragma"/>.
7459 The <literal>INLINE</literal> pragma affects the specialised version of the
7460 function (only), and applies even if the function is recursive. The motivating
7463 -- A GADT for arrays with type-indexed representation
7465 ArrInt :: !Int -> ByteArray# -> Arr Int
7466 ArrPair :: !Int -> Arr e1 -> Arr e2 -> Arr (e1, e2)
7468 (!:) :: Arr e -> Int -> e
7469 {-# SPECIALISE INLINE (!:) :: Arr Int -> Int -> Int #-}
7470 {-# SPECIALISE INLINE (!:) :: Arr (a, b) -> Int -> (a, b) #-}
7471 (ArrInt _ ba) !: (I# i) = I# (indexIntArray# ba i)
7472 (ArrPair _ a1 a2) !: i = (a1 !: i, a2 !: i)
7474 Here, <literal>(!:)</literal> is a recursive function that indexes arrays
7475 of type <literal>Arr e</literal>. Consider a call to <literal>(!:)</literal>
7476 at type <literal>(Int,Int)</literal>. The second specialisation will fire, and
7477 the specialised function will be inlined. It has two calls to
7478 <literal>(!:)</literal>,
7479 both at type <literal>Int</literal>. Both these calls fire the first
7480 specialisation, whose body is also inlined. The result is a type-based
7481 unrolling of the indexing function.</para>
7482 <para>Warning: you can make GHC diverge by using <literal>SPECIALISE INLINE</literal>
7483 on an ordinarily-recursive function.</para>
7485 <para>Note: In earlier versions of GHC, it was possible to provide your own
7486 specialised function for a given type:
7489 {-# SPECIALIZE hammeredLookup :: [(Int, value)] -> Int -> value = intLookup #-}
7492 This feature has been removed, as it is now subsumed by the
7493 <literal>RULES</literal> pragma (see <xref linkend="rule-spec"/>).</para>
7497 <sect2 id="specialize-instance-pragma">
7498 <title>SPECIALIZE instance pragma
7502 <indexterm><primary>SPECIALIZE pragma</primary></indexterm>
7503 <indexterm><primary>overloading, death to</primary></indexterm>
7504 Same idea, except for instance declarations. For example:
7507 instance (Eq a) => Eq (Foo a) where {
7508 {-# SPECIALIZE instance Eq (Foo [(Int, Bar)]) #-}
7512 The pragma must occur inside the <literal>where</literal> part
7513 of the instance declaration.
7516 Compatible with HBC, by the way, except perhaps in the placement
7522 <sect2 id="unpack-pragma">
7523 <title>UNPACK pragma</title>
7525 <indexterm><primary>UNPACK</primary></indexterm>
7527 <para>The <literal>UNPACK</literal> indicates to the compiler
7528 that it should unpack the contents of a constructor field into
7529 the constructor itself, removing a level of indirection. For
7533 data T = T {-# UNPACK #-} !Float
7534 {-# UNPACK #-} !Float
7537 <para>will create a constructor <literal>T</literal> containing
7538 two unboxed floats. This may not always be an optimisation: if
7539 the <function>T</function> constructor is scrutinised and the
7540 floats passed to a non-strict function for example, they will
7541 have to be reboxed (this is done automatically by the
7544 <para>Unpacking constructor fields should only be used in
7545 conjunction with <option>-O</option>, in order to expose
7546 unfoldings to the compiler so the reboxing can be removed as
7547 often as possible. For example:</para>
7551 f (T f1 f2) = f1 + f2
7554 <para>The compiler will avoid reboxing <function>f1</function>
7555 and <function>f2</function> by inlining <function>+</function>
7556 on floats, but only when <option>-O</option> is on.</para>
7558 <para>Any single-constructor data is eligible for unpacking; for
7562 data T = T {-# UNPACK #-} !(Int,Int)
7565 <para>will store the two <literal>Int</literal>s directly in the
7566 <function>T</function> constructor, by flattening the pair.
7567 Multi-level unpacking is also supported:
7570 data T = T {-# UNPACK #-} !S
7571 data S = S {-# UNPACK #-} !Int {-# UNPACK #-} !Int
7574 will store two unboxed <literal>Int#</literal>s
7575 directly in the <function>T</function> constructor. The
7576 unpacker can see through newtypes, too.</para>
7578 <para>If a field cannot be unpacked, you will not get a warning,
7579 so it might be an idea to check the generated code with
7580 <option>-ddump-simpl</option>.</para>
7582 <para>See also the <option>-funbox-strict-fields</option> flag,
7583 which essentially has the effect of adding
7584 <literal>{-# UNPACK #-}</literal> to every strict
7585 constructor field.</para>
7588 <sect2 id="source-pragma">
7589 <title>SOURCE pragma</title>
7591 <indexterm><primary>SOURCE</primary></indexterm>
7592 <para>The <literal>{-# SOURCE #-}</literal> pragma is used only in <literal>import</literal> declarations,
7593 to break a module loop. It is described in detail in <xref linkend="mutual-recursion"/>.
7599 <!-- ======================= REWRITE RULES ======================== -->
7601 <sect1 id="rewrite-rules">
7602 <title>Rewrite rules
7604 <indexterm><primary>RULES pragma</primary></indexterm>
7605 <indexterm><primary>pragma, RULES</primary></indexterm>
7606 <indexterm><primary>rewrite rules</primary></indexterm></title>
7609 The programmer can specify rewrite rules as part of the source program
7615 "map/map" forall f g xs. map f (map g xs) = map (f.g) xs
7620 Use the debug flag <option>-ddump-simpl-stats</option> to see what rules fired.
7621 If you need more information, then <option>-ddump-rule-firings</option> shows you
7622 each individual rule firing in detail.
7626 <title>Syntax</title>
7629 From a syntactic point of view:
7635 There may be zero or more rules in a <literal>RULES</literal> pragma, separated by semicolons (which
7636 may be generated by the layout rule).
7642 The layout rule applies in a pragma.
7643 Currently no new indentation level
7644 is set, so if you put several rules in single RULES pragma and wish to use layout to separate them,
7645 you must lay out the starting in the same column as the enclosing definitions.
7648 "map/map" forall f g xs. map f (map g xs) = map (f.g) xs
7649 "map/append" forall f xs ys. map f (xs ++ ys) = map f xs ++ map f ys
7652 Furthermore, the closing <literal>#-}</literal>
7653 should start in a column to the right of the opening <literal>{-#</literal>.
7659 Each rule has a name, enclosed in double quotes. The name itself has
7660 no significance at all. It is only used when reporting how many times the rule fired.
7666 A rule may optionally have a phase-control number (see <xref linkend="phase-control"/>),
7667 immediately after the name of the rule. Thus:
7670 "map/map" [2] forall f g xs. map f (map g xs) = map (f.g) xs
7673 The "[2]" means that the rule is active in Phase 2 and subsequent phases. The inverse
7674 notation "[~2]" is also accepted, meaning that the rule is active up to, but not including,
7683 Each variable mentioned in a rule must either be in scope (e.g. <function>map</function>),
7684 or bound by the <literal>forall</literal> (e.g. <function>f</function>, <function>g</function>, <function>xs</function>). The variables bound by
7685 the <literal>forall</literal> are called the <emphasis>pattern</emphasis> variables. They are separated
7686 by spaces, just like in a type <literal>forall</literal>.
7692 A pattern variable may optionally have a type signature.
7693 If the type of the pattern variable is polymorphic, it <emphasis>must</emphasis> have a type signature.
7694 For example, here is the <literal>foldr/build</literal> rule:
7697 "fold/build" forall k z (g::forall b. (a->b->b) -> b -> b) .
7698 foldr k z (build g) = g k z
7701 Since <function>g</function> has a polymorphic type, it must have a type signature.
7708 The left hand side of a rule must consist of a top-level variable applied
7709 to arbitrary expressions. For example, this is <emphasis>not</emphasis> OK:
7712 "wrong1" forall e1 e2. case True of { True -> e1; False -> e2 } = e1
7713 "wrong2" forall f. f True = True
7716 In <literal>"wrong1"</literal>, the LHS is not an application; in <literal>"wrong2"</literal>, the LHS has a pattern variable
7723 A rule does not need to be in the same module as (any of) the
7724 variables it mentions, though of course they need to be in scope.
7730 All rules are implicitly exported from the module, and are therefore
7731 in force in any module that imports the module that defined the rule, directly
7732 or indirectly. (That is, if A imports B, which imports C, then C's rules are
7733 in force when compiling A.) The situation is very similar to that for instance
7741 Inside a RULE "<literal>forall</literal>" is treated as a keyword, regardless of
7742 any other flag settings. Furthermore, inside a RULE, the language extension
7743 <option>-XScopedTypeVariables</option> is automatically enabled; see
7744 <xref linkend="scoped-type-variables"/>.
7750 Like other pragmas, RULE pragmas are always checked for scope errors, and
7751 are typechecked. Typechecking means that the LHS and RHS of a rule are typechecked,
7752 and must have the same type. However, rules are only <emphasis>enabled</emphasis>
7753 if the <option>-fenable-rewrite-rules</option> flag is
7754 on (see <xref linkend="rule-semantics"/>).
7763 <sect2 id="rule-semantics">
7764 <title>Semantics</title>
7767 From a semantic point of view:
7772 Rules are enabled (that is, used during optimisation)
7773 by the <option>-fenable-rewrite-rules</option> flag.
7774 This flag is implied by <option>-O</option>, and may be switched
7775 off (as usual) by <option>-fno-enable-rewrite-rules</option>.
7776 (NB: enabling <option>-fenable-rewrite-rules</option> without <option>-O</option>
7777 may not do what you expect, though, because without <option>-O</option> GHC
7778 ignores all optimisation information in interface files;
7779 see <option>-fignore-interface-pragmas</option>, <xref linkend="options-f"/>.)
7780 Note that <option>-fenable-rewrite-rules</option> is an <emphasis>optimisation</emphasis> flag, and
7781 has no effect on parsing or typechecking.
7787 Rules are regarded as left-to-right rewrite rules.
7788 When GHC finds an expression that is a substitution instance of the LHS
7789 of a rule, it replaces the expression by the (appropriately-substituted) RHS.
7790 By "a substitution instance" we mean that the LHS can be made equal to the
7791 expression by substituting for the pattern variables.
7798 GHC makes absolutely no attempt to verify that the LHS and RHS
7799 of a rule have the same meaning. That is undecidable in general, and
7800 infeasible in most interesting cases. The responsibility is entirely the programmer's!
7807 GHC makes no attempt to make sure that the rules are confluent or
7808 terminating. For example:
7811 "loop" forall x y. f x y = f y x
7814 This rule will cause the compiler to go into an infinite loop.
7821 If more than one rule matches a call, GHC will choose one arbitrarily to apply.
7827 GHC currently uses a very simple, syntactic, matching algorithm
7828 for matching a rule LHS with an expression. It seeks a substitution
7829 which makes the LHS and expression syntactically equal modulo alpha
7830 conversion. The pattern (rule), but not the expression, is eta-expanded if
7831 necessary. (Eta-expanding the expression can lead to laziness bugs.)
7832 But not beta conversion (that's called higher-order matching).
7836 Matching is carried out on GHC's intermediate language, which includes
7837 type abstractions and applications. So a rule only matches if the
7838 types match too. See <xref linkend="rule-spec"/> below.
7844 GHC keeps trying to apply the rules as it optimises the program.
7845 For example, consider:
7854 The expression <literal>s (t xs)</literal> does not match the rule <literal>"map/map"</literal>, but GHC
7855 will substitute for <varname>s</varname> and <varname>t</varname>, giving an expression which does match.
7856 If <varname>s</varname> or <varname>t</varname> was (a) used more than once, and (b) large or a redex, then it would
7857 not be substituted, and the rule would not fire.
7864 Ordinary inlining happens at the same time as rule rewriting, which may lead to unexpected
7865 results. Consider this (artificial) example
7868 {-# RULES "f" f True = False #-}
7874 Since <literal>f</literal>'s right-hand side is small, it is inlined into <literal>g</literal>,
7879 Now <literal>g</literal> is inlined into <literal>h</literal>, but <literal>f</literal>'s RULE has
7881 If instead GHC had first inlined <literal>g</literal> into <literal>h</literal> then there
7882 would have been a better chance that <literal>f</literal>'s RULE might fire.
7885 The way to get predictable behaviour is to use a NOINLINE
7886 pragma on <literal>f</literal>, to ensure
7887 that it is not inlined until its RULEs have had a chance to fire.
7897 <title>List fusion</title>
7900 The RULES mechanism is used to implement fusion (deforestation) of common list functions.
7901 If a "good consumer" consumes an intermediate list constructed by a "good producer", the
7902 intermediate list should be eliminated entirely.
7906 The following are good producers:
7918 Enumerations of <literal>Int</literal> and <literal>Char</literal> (e.g. <literal>['a'..'z']</literal>).
7924 Explicit lists (e.g. <literal>[True, False]</literal>)
7930 The cons constructor (e.g <literal>3:4:[]</literal>)
7936 <function>++</function>
7942 <function>map</function>
7948 <function>take</function>, <function>filter</function>
7954 <function>iterate</function>, <function>repeat</function>
7960 <function>zip</function>, <function>zipWith</function>
7969 The following are good consumers:
7981 <function>array</function> (on its second argument)
7987 <function>++</function> (on its first argument)
7993 <function>foldr</function>
7999 <function>map</function>
8005 <function>take</function>, <function>filter</function>
8011 <function>concat</function>
8017 <function>unzip</function>, <function>unzip2</function>, <function>unzip3</function>, <function>unzip4</function>
8023 <function>zip</function>, <function>zipWith</function> (but on one argument only; if both are good producers, <function>zip</function>
8024 will fuse with one but not the other)
8030 <function>partition</function>
8036 <function>head</function>
8042 <function>and</function>, <function>or</function>, <function>any</function>, <function>all</function>
8048 <function>sequence_</function>
8054 <function>msum</function>
8060 <function>sortBy</function>
8069 So, for example, the following should generate no intermediate lists:
8072 array (1,10) [(i,i*i) | i <- map (+ 1) [0..9]]
8078 This list could readily be extended; if there are Prelude functions that you use
8079 a lot which are not included, please tell us.
8083 If you want to write your own good consumers or producers, look at the
8084 Prelude definitions of the above functions to see how to do so.
8089 <sect2 id="rule-spec">
8090 <title>Specialisation
8094 Rewrite rules can be used to get the same effect as a feature
8095 present in earlier versions of GHC.
8096 For example, suppose that:
8099 genericLookup :: Ord a => Table a b -> a -> b
8100 intLookup :: Table Int b -> Int -> b
8103 where <function>intLookup</function> is an implementation of
8104 <function>genericLookup</function> that works very fast for
8105 keys of type <literal>Int</literal>. You might wish
8106 to tell GHC to use <function>intLookup</function> instead of
8107 <function>genericLookup</function> whenever the latter was called with
8108 type <literal>Table Int b -> Int -> b</literal>.
8109 It used to be possible to write
8112 {-# SPECIALIZE genericLookup :: Table Int b -> Int -> b = intLookup #-}
8115 This feature is no longer in GHC, but rewrite rules let you do the same thing:
8118 {-# RULES "genericLookup/Int" genericLookup = intLookup #-}
8121 This slightly odd-looking rule instructs GHC to replace
8122 <function>genericLookup</function> by <function>intLookup</function>
8123 <emphasis>whenever the types match</emphasis>.
8124 What is more, this rule does not need to be in the same
8125 file as <function>genericLookup</function>, unlike the
8126 <literal>SPECIALIZE</literal> pragmas which currently do (so that they
8127 have an original definition available to specialise).
8130 <para>It is <emphasis>Your Responsibility</emphasis> to make sure that
8131 <function>intLookup</function> really behaves as a specialised version
8132 of <function>genericLookup</function>!!!</para>
8134 <para>An example in which using <literal>RULES</literal> for
8135 specialisation will Win Big:
8138 toDouble :: Real a => a -> Double
8139 toDouble = fromRational . toRational
8141 {-# RULES "toDouble/Int" toDouble = i2d #-}
8142 i2d (I# i) = D# (int2Double# i) -- uses Glasgow prim-op directly
8145 The <function>i2d</function> function is virtually one machine
8146 instruction; the default conversion—via an intermediate
8147 <literal>Rational</literal>—is obscenely expensive by
8154 <title>Controlling what's going on</title>
8162 Use <option>-ddump-rules</option> to see what transformation rules GHC is using.
8168 Use <option>-ddump-simpl-stats</option> to see what rules are being fired.
8169 If you add <option>-dppr-debug</option> you get a more detailed listing.
8175 The definition of (say) <function>build</function> in <filename>GHC/Base.lhs</filename> looks like this:
8178 build :: forall a. (forall b. (a -> b -> b) -> b -> b) -> [a]
8179 {-# INLINE build #-}
8183 Notice the <literal>INLINE</literal>! That prevents <literal>(:)</literal> from being inlined when compiling
8184 <literal>PrelBase</literal>, so that an importing module will “see” the <literal>(:)</literal>, and can
8185 match it on the LHS of a rule. <literal>INLINE</literal> prevents any inlining happening
8186 in the RHS of the <literal>INLINE</literal> thing. I regret the delicacy of this.
8193 In <filename>libraries/base/GHC/Base.lhs</filename> look at the rules for <function>map</function> to
8194 see how to write rules that will do fusion and yet give an efficient
8195 program even if fusion doesn't happen. More rules in <filename>GHC/List.lhs</filename>.
8205 <sect2 id="core-pragma">
8206 <title>CORE pragma</title>
8208 <indexterm><primary>CORE pragma</primary></indexterm>
8209 <indexterm><primary>pragma, CORE</primary></indexterm>
8210 <indexterm><primary>core, annotation</primary></indexterm>
8213 The external core format supports <quote>Note</quote> annotations;
8214 the <literal>CORE</literal> pragma gives a way to specify what these
8215 should be in your Haskell source code. Syntactically, core
8216 annotations are attached to expressions and take a Haskell string
8217 literal as an argument. The following function definition shows an
8221 f x = ({-# CORE "foo" #-} show) ({-# CORE "bar" #-} x)
8224 Semantically, this is equivalent to:
8232 However, when external core is generated (via
8233 <option>-fext-core</option>), there will be Notes attached to the
8234 expressions <function>show</function> and <varname>x</varname>.
8235 The core function declaration for <function>f</function> is:
8239 f :: %forall a . GHCziShow.ZCTShow a ->
8240 a -> GHCziBase.ZMZN GHCziBase.Char =
8241 \ @ a (zddShow::GHCziShow.ZCTShow a) (eta::a) ->
8243 %case zddShow %of (tpl::GHCziShow.ZCTShow a)
8245 (tpl1::GHCziBase.Int ->
8247 GHCziBase.ZMZN GHCziBase.Char -> GHCziBase.ZMZN GHCziBase.Cha
8249 (tpl2::a -> GHCziBase.ZMZN GHCziBase.Char)
8250 (tpl3::GHCziBase.ZMZN a ->
8251 GHCziBase.ZMZN GHCziBase.Char -> GHCziBase.ZMZN GHCziBase.Cha
8259 Here, we can see that the function <function>show</function> (which
8260 has been expanded out to a case expression over the Show dictionary)
8261 has a <literal>%note</literal> attached to it, as does the
8262 expression <varname>eta</varname> (which used to be called
8263 <varname>x</varname>).
8270 <sect1 id="special-ids">
8271 <title>Special built-in functions</title>
8272 <para>GHC has a few built-in functions with special behaviour. These
8273 are now described in the module <ulink
8274 url="../libraries/ghc-prim/GHC-Prim.html"><literal>GHC.Prim</literal></ulink>
8275 in the library documentation.</para>
8279 <sect1 id="generic-classes">
8280 <title>Generic classes</title>
8283 The ideas behind this extension are described in detail in "Derivable type classes",
8284 Ralf Hinze and Simon Peyton Jones, Haskell Workshop, Montreal Sept 2000, pp94-105.
8285 An example will give the idea:
8293 fromBin :: [Int] -> (a, [Int])
8295 toBin {| Unit |} Unit = []
8296 toBin {| a :+: b |} (Inl x) = 0 : toBin x
8297 toBin {| a :+: b |} (Inr y) = 1 : toBin y
8298 toBin {| a :*: b |} (x :*: y) = toBin x ++ toBin y
8300 fromBin {| Unit |} bs = (Unit, bs)
8301 fromBin {| a :+: b |} (0:bs) = (Inl x, bs') where (x,bs') = fromBin bs
8302 fromBin {| a :+: b |} (1:bs) = (Inr y, bs') where (y,bs') = fromBin bs
8303 fromBin {| a :*: b |} bs = (x :*: y, bs'') where (x,bs' ) = fromBin bs
8304 (y,bs'') = fromBin bs'
8307 This class declaration explains how <literal>toBin</literal> and <literal>fromBin</literal>
8308 work for arbitrary data types. They do so by giving cases for unit, product, and sum,
8309 which are defined thus in the library module <literal>Generics</literal>:
8313 data a :+: b = Inl a | Inr b
8314 data a :*: b = a :*: b
8317 Now you can make a data type into an instance of Bin like this:
8319 instance (Bin a, Bin b) => Bin (a,b)
8320 instance Bin a => Bin [a]
8322 That is, just leave off the "where" clause. Of course, you can put in the
8323 where clause and over-ride whichever methods you please.
8327 <title> Using generics </title>
8328 <para>To use generics you need to</para>
8331 <para>Use the flags <option>-fglasgow-exts</option> (to enable the extra syntax),
8332 <option>-XGenerics</option> (to generate extra per-data-type code),
8333 and <option>-package lang</option> (to make the <literal>Generics</literal> library
8337 <para>Import the module <literal>Generics</literal> from the
8338 <literal>lang</literal> package. This import brings into
8339 scope the data types <literal>Unit</literal>,
8340 <literal>:*:</literal>, and <literal>:+:</literal>. (You
8341 don't need this import if you don't mention these types
8342 explicitly; for example, if you are simply giving instance
8343 declarations.)</para>
8348 <sect2> <title> Changes wrt the paper </title>
8350 Note that the type constructors <literal>:+:</literal> and <literal>:*:</literal>
8351 can be written infix (indeed, you can now use
8352 any operator starting in a colon as an infix type constructor). Also note that
8353 the type constructors are not exactly as in the paper (Unit instead of 1, etc).
8354 Finally, note that the syntax of the type patterns in the class declaration
8355 uses "<literal>{|</literal>" and "<literal>|}</literal>" brackets; curly braces
8356 alone would ambiguous when they appear on right hand sides (an extension we
8357 anticipate wanting).
8361 <sect2> <title>Terminology and restrictions</title>
8363 Terminology. A "generic default method" in a class declaration
8364 is one that is defined using type patterns as above.
8365 A "polymorphic default method" is a default method defined as in Haskell 98.
8366 A "generic class declaration" is a class declaration with at least one
8367 generic default method.
8375 Alas, we do not yet implement the stuff about constructor names and
8382 A generic class can have only one parameter; you can't have a generic
8383 multi-parameter class.
8389 A default method must be defined entirely using type patterns, or entirely
8390 without. So this is illegal:
8393 op :: a -> (a, Bool)
8394 op {| Unit |} Unit = (Unit, True)
8397 However it is perfectly OK for some methods of a generic class to have
8398 generic default methods and others to have polymorphic default methods.
8404 The type variable(s) in the type pattern for a generic method declaration
8405 scope over the right hand side. So this is legal (note the use of the type variable ``p'' in a type signature on the right hand side:
8409 op {| p :*: q |} (x :*: y) = op (x :: p)
8417 The type patterns in a generic default method must take one of the forms:
8423 where "a" and "b" are type variables. Furthermore, all the type patterns for
8424 a single type constructor (<literal>:*:</literal>, say) must be identical; they
8425 must use the same type variables. So this is illegal:
8429 op {| a :+: b |} (Inl x) = True
8430 op {| p :+: q |} (Inr y) = False
8432 The type patterns must be identical, even in equations for different methods of the class.
8433 So this too is illegal:
8437 op1 {| a :*: b |} (x :*: y) = True
8440 op2 {| p :*: q |} (x :*: y) = False
8442 (The reason for this restriction is that we gather all the equations for a particular type constructor
8443 into a single generic instance declaration.)
8449 A generic method declaration must give a case for each of the three type constructors.
8455 The type for a generic method can be built only from:
8457 <listitem> <para> Function arrows </para> </listitem>
8458 <listitem> <para> Type variables </para> </listitem>
8459 <listitem> <para> Tuples </para> </listitem>
8460 <listitem> <para> Arbitrary types not involving type variables </para> </listitem>
8462 Here are some example type signatures for generic methods:
8465 op2 :: Bool -> (a,Bool)
8466 op3 :: [Int] -> a -> a
8469 Here, op1, op2, op3 are OK, but op4 is rejected, because it has a type variable
8473 This restriction is an implementation restriction: we just haven't got around to
8474 implementing the necessary bidirectional maps over arbitrary type constructors.
8475 It would be relatively easy to add specific type constructors, such as Maybe and list,
8476 to the ones that are allowed.</para>
8481 In an instance declaration for a generic class, the idea is that the compiler
8482 will fill in the methods for you, based on the generic templates. However it can only
8487 The instance type is simple (a type constructor applied to type variables, as in Haskell 98).
8492 No constructor of the instance type has unboxed fields.
8496 (Of course, these things can only arise if you are already using GHC extensions.)
8497 However, you can still give an instance declarations for types which break these rules,
8498 provided you give explicit code to override any generic default methods.
8506 The option <option>-ddump-deriv</option> dumps incomprehensible stuff giving details of
8507 what the compiler does with generic declarations.
8512 <sect2> <title> Another example </title>
8514 Just to finish with, here's another example I rather like:
8518 nCons {| Unit |} _ = 1
8519 nCons {| a :*: b |} _ = 1
8520 nCons {| a :+: b |} _ = nCons (bot::a) + nCons (bot::b)
8523 tag {| Unit |} _ = 1
8524 tag {| a :*: b |} _ = 1
8525 tag {| a :+: b |} (Inl x) = tag x
8526 tag {| a :+: b |} (Inr y) = nCons (bot::a) + tag y
8532 <sect1 id="monomorphism">
8533 <title>Control over monomorphism</title>
8535 <para>GHC supports two flags that control the way in which generalisation is
8536 carried out at let and where bindings.
8540 <title>Switching off the dreaded Monomorphism Restriction</title>
8541 <indexterm><primary><option>-XNoMonomorphismRestriction</option></primary></indexterm>
8543 <para>Haskell's monomorphism restriction (see
8544 <ulink url="http://www.haskell.org/onlinereport/decls.html#sect4.5.5">Section
8546 of the Haskell Report)
8547 can be completely switched off by
8548 <option>-XNoMonomorphismRestriction</option>.
8553 <title>Monomorphic pattern bindings</title>
8554 <indexterm><primary><option>-XNoMonoPatBinds</option></primary></indexterm>
8555 <indexterm><primary><option>-XMonoPatBinds</option></primary></indexterm>
8557 <para> As an experimental change, we are exploring the possibility of
8558 making pattern bindings monomorphic; that is, not generalised at all.
8559 A pattern binding is a binding whose LHS has no function arguments,
8560 and is not a simple variable. For example:
8562 f x = x -- Not a pattern binding
8563 f = \x -> x -- Not a pattern binding
8564 f :: Int -> Int = \x -> x -- Not a pattern binding
8566 (g,h) = e -- A pattern binding
8567 (f) = e -- A pattern binding
8568 [x] = e -- A pattern binding
8570 Experimentally, GHC now makes pattern bindings monomorphic <emphasis>by
8571 default</emphasis>. Use <option>-XNoMonoPatBinds</option> to recover the
8580 ;;; Local Variables: ***
8582 ;;; sgml-parent-document: ("users_guide.xml" "book" "chapter" "sect1") ***
8583 ;;; ispell-local-dictionary: "british" ***