2 <indexterm><primary>language, GHC</primary></indexterm>
3 <indexterm><primary>extensions, GHC</primary></indexterm>
4 As with all known Haskell systems, GHC implements some extensions to
5 the language. To use them, you'll need to give a <option>-fglasgow-exts</option>
6 <indexterm><primary>-fglasgow-exts option</primary></indexterm> option.
10 Virtually all of the Glasgow extensions serve to give you access to
11 the underlying facilities with which we implement Haskell. Thus, you
12 can get at the Raw Iron, if you are willing to write some non-standard
13 code at a more primitive level. You need not be “stuck” on
14 performance because of the implementation costs of Haskell's
15 “high-level” features—you can always code “under” them. In an extreme case, you can write all your time-critical code in C, and then just glue it together with Haskell!
19 Executive summary of our extensions:
25 <term>Unboxed types and primitive operations:</Term>
27 <para>You can get right down to the raw machine types and
28 operations; included in this are “primitive
29 arrays” (direct access to Big Wads of Bytes). Please
30 see <XRef LinkEnd="glasgow-unboxed"> and following.</para>
35 <term>Type system extensions:</term>
37 <para> GHC supports a large number of extensions to Haskell's
38 type system. Specifically:</para>
42 <term>Multi-parameter type classes:</term>
44 <para><xref LinkEnd="multi-param-type-classes"></para>
49 <term>Functional dependencies:</term>
51 <para><xref LinkEnd="functional-dependencies"></para>
56 <term>Implicit parameters:</term>
58 <para><xref LinkEnd="implicit-parameters"></para>
63 <term>Local universal quantification:</term>
65 <para><xref LinkEnd="universal-quantification"></para>
70 <term>Extistentially quantification in data types:</term>
72 <para><xref LinkEnd="existential-quantification"></para>
77 <term>Scoped type variables:</term>
79 <para>Scoped type variables enable the programmer to
80 supply type signatures for some nested declarations,
81 where this would not be legal in Haskell 98. Details in
82 <xref LinkEnd="scoped-type-variables">.</para>
90 <term>Pattern guards</term>
92 <para>Instead of being a boolean expression, a guard is a list
93 of qualifiers, exactly as in a list comprehension. See <xref
94 LinkEnd="pattern-guards">.</para>
99 <term>Data types with no constructors</term>
101 <para>See <xref LinkEnd="nullary-types">.</para>
106 <term>Parallel list comprehensions</term>
108 <para>An extension to the list comprehension syntax to support
109 <literal>zipWith</literal>-like functionality. See <xref
110 linkend="parallel-list-comprehensions">.</para>
115 <term>Foreign calling:</term>
117 <para>Just what it sounds like. We provide
118 <emphasis>lots</emphasis> of rope that you can dangle around
119 your neck. Please see <xref LinkEnd="ffi">.</para>
126 <para>Pragmas are special instructions to the compiler placed
127 in the source file. The pragmas GHC supports are described in
128 <xref LinkEnd="pragmas">.</para>
133 <term>Rewrite rules:</term>
135 <para>The programmer can specify rewrite rules as part of the
136 source program (in a pragma). GHC applies these rewrite rules
137 wherever it can. Details in <xref
138 LinkEnd="rewrite-rules">.</para>
143 <term>Generic classes:</term>
145 <para>Generic class declarations allow you to define a class
146 whose methods say how to work over an arbitrary data type.
147 Then it's really easy to make any new type into an instance of
148 the class. This generalises the rather ad-hoc "deriving"
149 feature of Haskell 98. Details in <xref
150 LinkEnd="generic-classes">.</para>
156 Before you get too carried away working at the lowest level (e.g.,
157 sloshing <literal>MutableByteArray#</literal>s around your
158 program), you may wish to check if there are libraries that provide a
159 “Haskellised veneer” over the features you want. See
160 <xref linkend="book-hslibs">.
163 <sect1 id="options-language">
164 <title>Language options</title>
166 <indexterm><primary>language</primary><secondary>option</secondary>
168 <indexterm><primary>options</primary><secondary>language</secondary>
170 <indexterm><primary>extensions</primary><secondary>options controlling</secondary>
173 <para> These flags control what variation of the language are
174 permitted. Leaving out all of them gives you standard Haskell
180 <term><option>-fglasgow-exts</option>:</term>
181 <indexterm><primary><option>-fglasgow-exts</option></primary></indexterm>
183 <para>This simultaneously enables all of the extensions to
184 Haskell 98 described in <xref
185 linkend="ghc-language-features">, except where otherwise
191 <term><option>-fno-monomorphism-restriction</option>:</term>
192 <indexterm><primary><option>-fno-monomorphism-restriction</option></primary></indexterm>
194 <para> Switch off the Haskell 98 monomorphism restriction.
195 Independent of the <option>-fglasgow-exts</option>
201 <term><option>-fallow-overlapping-instances</option></term>
202 <term><option>-fallow-undecidable-instances</option></term>
203 <term><option>-fcontext-stack</option></term>
204 <indexterm><primary><option>-fallow-overlapping-instances</option></primary></indexterm>
205 <indexterm><primary><option>-fallow-undecidable-instances</option></primary></indexterm>
206 <indexterm><primary><option>-fcontext-stack</option></primary></indexterm>
208 <para> See <xref LinkEnd="instance-decls">. Only relevant
209 if you also use <option>-fglasgow-exts</option>.</para>
214 <term><option>-finline-phase</option></term>
215 <indexterm><primary><option>-finline-phase</option></primary></indexterm>
217 <para>See <xref LinkEnd="rewrite-rules">. Only relevant if
218 you also use <option>-fglasgow-exts</option>.</para>
223 <term><option>-fgenerics</option></term>
224 <indexterm><primary><option>-fgenerics</option></primary></indexterm>
226 <para>See <xref LinkEnd="generic-classes">. Independent of
227 <option>-fglasgow-exts</option>.</para>
232 <term><option>-fno-implicit-prelude</option></term>
234 <para><indexterm><primary>-fno-implicit-prelude
235 option</primary></indexterm> GHC normally imports
236 <filename>Prelude.hi</filename> files for you. If you'd
237 rather it didn't, then give it a
238 <option>-fno-implicit-prelude</option> option. The idea
239 is that you can then import a Prelude of your own. (But
240 don't call it <literal>Prelude</literal>; the Haskell
241 module namespace is flat, and you must not conflict with
242 any Prelude module.)</para>
244 <para>Even though you have not imported the Prelude, all
245 the built-in syntax still refers to the built-in Haskell
246 Prelude types and values, as specified by the Haskell
247 Report. For example, the type <literal>[Int]</literal>
248 still means <literal>Prelude.[] Int</literal>; tuples
249 continue to refer to the standard Prelude tuples; the
250 translation for list comprehensions continues to use
251 <literal>Prelude.map</literal> etc.</para>
253 <para> With one group of exceptions! You may want to
254 define your own numeric class hierarchy. It completely
255 defeats that purpose if the literal "1" means
256 "<literal>Prelude.fromInteger 1</literal>", which is what
257 the Haskell Report specifies. So the
258 <option>-fno-implicit-prelude</option> flag causes the
259 following pieces of built-in syntax to refer to <emphasis>whatever
260 is in scope</emphasis>, not the Prelude versions:</para>
264 <para>Integer and fractional literals mean
265 "<literal>fromInteger 1</literal>" and
266 "<literal>fromRational 3.2</literal>", not the
267 Prelude-qualified versions; both in expressions and in
272 <para>Negation (e.g. "<literal>- (f x)</literal>")
273 means "<literal>negate (f x)</literal>" (not
274 <literal>Prelude.negate</literal>).</para>
278 <para>In an n+k pattern, the standard Prelude
279 <literal>Ord</literal> class is still used for comparison,
280 but the necessary subtraction uses whatever
281 "<literal>(-)</literal>" is in scope (not
282 "<literal>Prelude.(-)</literal>").</para>
286 <para>Note: Negative literals, such as <literal>-3</literal>, are
287 specified by (a careful reading of) the Haskell Report as
288 meaning <literal>Prelude.negate (Prelude.fromInteger 3)</literal>.
289 However, GHC deviates from this slightly, and treats them as meaning
290 <literal>fromInteger (-3)</literal>. One particular effect of this
291 slightly-non-standard reading is that there is no difficulty with
292 the literal <literal>-2147483648</literal> at type <literal>Int</literal>;
293 it means <literal>fromInteger (-2147483648)</literal>. The strict interpretation
294 would be <literal>negate (fromInteger 2147483648)</literal>,
295 and the call to <literal>fromInteger</literal> would overflow
296 (at type <literal>Int</literal>, remember).
305 <!-- UNBOXED TYPES AND PRIMITIVE OPERATIONS -->
308 <sect1 id="glasgow-ST-monad">
309 <title>Primitive state-transformer monad</title>
312 <indexterm><primary>state transformers (Glasgow extensions)</primary></indexterm>
313 <indexterm><primary>ST monad (Glasgow extension)</primary></indexterm>
317 This monad underlies our implementation of arrays, mutable and
318 immutable, and our implementation of I/O, including “C calls”.
322 The <literal>ST</literal> library, which provides access to the
323 <function>ST</function> monad, is described in <xref
329 <sect1 id="glasgow-prim-arrays">
330 <title>Primitive arrays, mutable and otherwise
334 <indexterm><primary>primitive arrays (Glasgow extension)</primary></indexterm>
335 <indexterm><primary>arrays, primitive (Glasgow extension)</primary></indexterm>
339 GHC knows about quite a few flavours of Large Swathes of Bytes.
343 First, GHC distinguishes between primitive arrays of (boxed) Haskell
344 objects (type <literal>Array# obj</literal>) and primitive arrays of bytes (type
345 <literal>ByteArray#</literal>).
349 Second, it distinguishes between…
353 <term>Immutable:</term>
356 Arrays that do not change (as with “standard” Haskell arrays); you
357 can only read from them. Obviously, they do not need the care and
358 attention of the state-transformer monad.
363 <term>Mutable:</term>
366 Arrays that may be changed or “mutated.” All the operations on them
367 live within the state-transformer monad and the updates happen
368 <emphasis>in-place</emphasis>.
373 <term>“Static” (in C land):</term>
376 A C routine may pass an <literal>Addr#</literal> pointer back into Haskell land. There
377 are then primitive operations with which you may merrily grab values
378 over in C land, by indexing off the “static” pointer.
383 <term>“Stable” pointers:</term>
386 If, for some reason, you wish to hand a Haskell pointer (i.e.,
387 <emphasis>not</emphasis> an unboxed value) to a C routine, you first make the
388 pointer “stable,” so that the garbage collector won't forget that it
389 exists. That is, GHC provides a safe way to pass Haskell pointers to
394 Please see <xref LinkEnd="sec-stable-pointers"> for more details.
399 <term>“Foreign objects”:</term>
402 A “foreign object” is a safe way to pass an external object (a
403 C-allocated pointer, say) to Haskell and have Haskell do the Right
404 Thing when it no longer references the object. So, for example, C
405 could pass a large bitmap over to Haskell and say “please free this
406 memory when you're done with it.”
410 Please see <xref LinkEnd="sec-ForeignObj"> for more details.
418 The libraries documentatation gives more details on all these
419 “primitive array” types and the operations on them.
425 <sect1 id="nullary-types">
426 <title>Data types with no constructors</title>
428 <para>With the <option>-fglasgow-exts</option> flag, GHC lets you declare
429 a data type with no constructors. For example:</para>
432 data T a -- T :: * -> *
434 <para>Syntactically, the declaration lacks the "= constrs" part. The
435 type can be parameterised, but only over ordinary types, of kind *; since
436 Haskell does not have kind signatures, you cannot parameterise over higher-kinded
439 <para>Such data types have only one value, namely bottom.
440 Nevertheless, they can be useful when defining "phantom types".</para>
443 <sect1 id="pattern-guards">
444 <title>Pattern guards</title>
447 <indexterm><primary>Pattern guards (Glasgow extension)</primary></indexterm>
448 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.)
452 Suppose we have an abstract data type of finite maps, with a
456 lookup :: FiniteMap -> Int -> Maybe Int
459 The lookup returns <function>Nothing</function> if the supplied key is not in the domain of the mapping, and <function>(Just v)</function> otherwise,
460 where <VarName>v</VarName> is the value that the key maps to. Now consider the following definition:
464 clunky env var1 var2 | ok1 && ok2 = val1 + val2
465 | otherwise = var1 + var2
476 The auxiliary functions are
480 maybeToBool :: Maybe a -> Bool
481 maybeToBool (Just x) = True
482 maybeToBool Nothing = False
484 expectJust :: Maybe a -> a
485 expectJust (Just x) = x
486 expectJust Nothing = error "Unexpected Nothing"
490 What is <function>clunky</function> doing? The guard <literal>ok1 &&
491 ok2</literal> checks that both lookups succeed, using
492 <function>maybeToBool</function> to convert the <function>Maybe</function>
493 types to booleans. The (lazily evaluated) <function>expectJust</function>
494 calls extract the values from the results of the lookups, and binds the
495 returned values to <VarName>val1</VarName> and <VarName>val2</VarName>
496 respectively. If either lookup fails, then clunky takes the
497 <literal>otherwise</literal> case and returns the sum of its arguments.
501 This is certainly legal Haskell, but it is a tremendously verbose and
502 un-obvious way to achieve the desired effect. Arguably, a more direct way
503 to write clunky would be to use case expressions:
507 clunky env var1 var1 = case lookup env var1 of
509 Just val1 -> case lookup env var2 of
511 Just val2 -> val1 + val2
517 This is a bit shorter, but hardly better. Of course, we can rewrite any set
518 of pattern-matching, guarded equations as case expressions; that is
519 precisely what the compiler does when compiling equations! The reason that
520 Haskell provides guarded equations is because they allow us to write down
521 the cases we want to consider, one at a time, independently of each other.
522 This structure is hidden in the case version. Two of the right-hand sides
523 are really the same (<function>fail</function>), and the whole expression
524 tends to become more and more indented.
528 Here is how I would write clunky:
533 | Just val1 <- lookup env var1
534 , Just val2 <- lookup env var2
536 ...other equations for clunky...
540 The semantics should be clear enough. The qualifers are matched in order.
541 For a <literal><-</literal> qualifier, which I call a pattern guard, the
542 right hand side is evaluated and matched against the pattern on the left.
543 If the match fails then the whole guard fails and the next equation is
544 tried. If it succeeds, then the appropriate binding takes place, and the
545 next qualifier is matched, in the augmented environment. Unlike list
546 comprehensions, however, the type of the expression to the right of the
547 <literal><-</literal> is the same as the type of the pattern to its
548 left. The bindings introduced by pattern guards scope over all the
549 remaining guard qualifiers, and over the right hand side of the equation.
553 Just as with list comprehensions, boolean expressions can be freely mixed
554 with among the pattern guards. For example:
565 Haskell's current guards therefore emerge as a special case, in which the
566 qualifier list has just one element, a boolean expression.
570 <sect1 id="parallel-list-comprehensions">
571 <title>Parallel List Comprehensions</title>
572 <indexterm><primary>list comprehensions</primary><secondary>parallel</secondary>
574 <indexterm><primary>parallel list comprehensions</primary>
577 <para>Parallel list comprehensions are a natural extension to list
578 comprehensions. List comprehensions can be thought of as a nice
579 syntax for writing maps and filters. Parallel comprehensions
580 extend this to include the zipWith family.</para>
582 <para>A parallel list comprehension has multiple independent
583 branches of qualifier lists, each separated by a `|' symbol. For
584 example, the following zips together two lists:</para>
587 [ (x, y) | x <- xs | y <- ys ]
590 <para>The behavior of parallel list comprehensions follows that of
591 zip, in that the resulting list will have the same length as the
592 shortest branch.</para>
594 <para>We can define parallel list comprehensions by translation to
595 regular comprehensions. Here's the basic idea:</para>
597 <para>Given a parallel comprehension of the form: </para>
600 [ e | p1 <- e11, p2 <- e12, ...
601 | q1 <- e21, q2 <- e22, ...
606 <para>This will be translated to: </para>
609 [ e | ((p1,p2), (q1,q2), ...) <- zipN [(p1,p2) | p1 <- e11, p2 <- e12, ...]
610 [(q1,q2) | q1 <- e21, q2 <- e22, ...]
615 <para>where `zipN' is the appropriate zip for the given number of
621 <title>The foreign interface</title>
623 <para>The foreign interface consists of the following components:</para>
627 <para>The Foreign Function Interface language specification
628 (included in this manual, in <xref linkend="ffi">).
629 You must use the <option>-fglasgow-exts</option> command-line option
630 to make GHC understand the <literal>foreign</literal> declarations
631 defined by the FFI.</para>
635 <para>The <literal>Foreign</literal> module (see <xref
636 linkend="sec-Foreign">) collects together several interfaces
637 which are useful in specifying foreign language
638 interfaces, including the following:</para>
642 <para>The <literal>ForeignObj</literal> module (see <xref
643 linkend="sec-ForeignObj">), for managing pointers from
644 Haskell into the outside world.</para>
648 <para>The <literal>StablePtr</literal> module (see <xref
649 linkend="sec-stable-pointers">), for managing pointers
650 into Haskell from the outside world.</para>
654 <para>The <literal>CTypes</literal> module (see <xref
655 linkend="sec-CTypes">) gives Haskell equivalents for the
656 standard C datatypes, for use in making Haskell bindings
657 to existing C libraries.</para>
661 <para>The <literal>CTypesISO</literal> module (see <xref
662 linkend="sec-CTypesISO">) gives Haskell equivalents for C
663 types defined by the ISO C standard.</para>
667 <para>The <literal>Storable</literal> library, for
668 primitive marshalling of data types between Haskell and
669 the foreign language.</para>
676 <para>The following sections also give some hints and tips on the use
677 of the foreign function interface in GHC.</para>
679 <sect2 id="glasgow-foreign-headers">
680 <title>Using function headers
684 <indexterm><primary>C calls, function headers</primary></indexterm>
688 When generating C (using the <option>-fvia-C</option> directive), one can assist the
689 C compiler in detecting type errors by using the <option>-#include</option> directive
690 (<xref linkend="options-C-compiler">) to provide <filename>.h</filename> files containing function headers.
702 void initialiseEFS (HsInt size);
703 HsInt terminateEFS (void);
704 HsForeignObj emptyEFS(void);
705 HsForeignObj updateEFS (HsForeignObj a, HsInt i, HsInt x);
706 HsInt lookupEFS (HsForeignObj a, HsInt i);
710 <para>The types <literal>HsInt</literal>,
711 <literal>HsForeignObj</literal> etc. are described in <xref
712 linkend="sec-mapping-table">.</para>
714 <para>Note that this approach is only
715 <emphasis>essential</emphasis> for returning
716 <literal>float</literal>s (or if <literal>sizeof(int) !=
717 sizeof(int *)</literal> on your architecture) but is a Good
718 Thing for anyone who cares about writing solid code. You're
719 crazy not to do it.</para>
725 <sect1 id="multi-param-type-classes">
726 <title>Multi-parameter type classes
730 This section documents GHC's implementation of multi-parameter type
731 classes. There's lots of background in the paper <ULink
732 URL="http://research.microsoft.com/~simonpj/multi.ps.gz" >Type
733 classes: exploring the design space</ULink > (Simon Peyton Jones, Mark
738 I'd like to thank people who reported shorcomings in the GHC 3.02
739 implementation. Our default decisions were all conservative ones, and
740 the experience of these heroic pioneers has given useful concrete
741 examples to support several generalisations. (These appear below as
742 design choices not implemented in 3.02.)
746 I've discussed these notes with Mark Jones, and I believe that Hugs
747 will migrate towards the same design choices as I outline here.
748 Thanks to him, and to many others who have offered very useful
756 There are the following restrictions on the form of a qualified
763 forall tv1..tvn (c1, ...,cn) => type
769 (Here, I write the "foralls" explicitly, although the Haskell source
770 language omits them; in Haskell 1.4, all the free type variables of an
771 explicit source-language type signature are universally quantified,
772 except for the class type variables in a class declaration. However,
773 in GHC, you can give the foralls if you want. See <xref LinkEnd="universal-quantification">).
782 <emphasis>Each universally quantified type variable
783 <literal>tvi</literal> must be mentioned (i.e. appear free) in <literal>type</literal></emphasis>.
785 The reason for this is that a value with a type that does not obey
786 this restriction could not be used without introducing
787 ambiguity. Here, for example, is an illegal type:
791 forall a. Eq a => Int
795 When a value with this type was used, the constraint <literal>Eq tv</literal>
796 would be introduced where <literal>tv</literal> is a fresh type variable, and
797 (in the dictionary-translation implementation) the value would be
798 applied to a dictionary for <literal>Eq tv</literal>. The difficulty is that we
799 can never know which instance of <literal>Eq</literal> to use because we never
800 get any more information about <literal>tv</literal>.
807 <emphasis>Every constraint <literal>ci</literal> must mention at least one of the
808 universally quantified type variables <literal>tvi</literal></emphasis>.
810 For example, this type is OK because <literal>C a b</literal> mentions the
811 universally quantified type variable <literal>b</literal>:
815 forall a. C a b => burble
819 The next type is illegal because the constraint <literal>Eq b</literal> does not
820 mention <literal>a</literal>:
824 forall a. Eq b => burble
828 The reason for this restriction is milder than the other one. The
829 excluded types are never useful or necessary (because the offending
830 context doesn't need to be witnessed at this point; it can be floated
831 out). Furthermore, floating them out increases sharing. Lastly,
832 excluding them is a conservative choice; it leaves a patch of
833 territory free in case we need it later.
843 These restrictions apply to all types, whether declared in a type signature
848 Unlike Haskell 1.4, constraints in types do <emphasis>not</emphasis> have to be of
849 the form <emphasis>(class type-variables)</emphasis>. Thus, these type signatures
856 f :: Eq (m a) => [m a] -> [m a]
863 This choice recovers principal types, a property that Haskell 1.4 does not have.
869 <title>Class declarations</title>
877 <emphasis>Multi-parameter type classes are permitted</emphasis>. For example:
881 class Collection c a where
882 union :: c a -> c a -> c a
893 <emphasis>The class hierarchy must be acyclic</emphasis>. However, the definition
894 of "acyclic" involves only the superclass relationships. For example,
900 op :: D b => a -> b -> b
903 class C a => D a where { ... }
907 Here, <literal>C</literal> is a superclass of <literal>D</literal>, but it's OK for a
908 class operation <literal>op</literal> of <literal>C</literal> to mention <literal>D</literal>. (It
909 would not be OK for <literal>D</literal> to be a superclass of <literal>C</literal>.)
916 <emphasis>There are no restrictions on the context in a class declaration
917 (which introduces superclasses), except that the class hierarchy must
918 be acyclic</emphasis>. So these class declarations are OK:
922 class Functor (m k) => FiniteMap m k where
925 class (Monad m, Monad (t m)) => Transform t m where
926 lift :: m a -> (t m) a
935 <emphasis>In the signature of a class operation, every constraint
936 must mention at least one type variable that is not a class type
943 class Collection c a where
944 mapC :: Collection c b => (a->b) -> c a -> c b
948 is OK because the constraint <literal>(Collection a b)</literal> mentions
949 <literal>b</literal>, even though it also mentions the class variable
950 <literal>a</literal>. On the other hand:
955 op :: Eq a => (a,b) -> (a,b)
959 is not OK because the constraint <literal>(Eq a)</literal> mentions on the class
960 type variable <literal>a</literal>, but not <literal>b</literal>. However, any such
961 example is easily fixed by moving the offending context up to the
966 class Eq a => C a where
971 A yet more relaxed rule would allow the context of a class-op signature
972 to mention only class type variables. However, that conflicts with
973 Rule 1(b) for types above.
980 <emphasis>The type of each class operation must mention <emphasis>all</emphasis> of
981 the class type variables</emphasis>. For example:
987 insert :: s -> a -> s
991 is not OK, because the type of <literal>empty</literal> doesn't mention
992 <literal>a</literal>. This rule is a consequence of Rule 1(a), above, for
993 types, and has the same motivation.
995 Sometimes, offending class declarations exhibit misunderstandings. For
996 example, <literal>Coll</literal> might be rewritten
1000 class Coll s a where
1002 insert :: s a -> a -> s a
1006 which makes the connection between the type of a collection of
1007 <literal>a</literal>'s (namely <literal>(s a)</literal>) and the element type <literal>a</literal>.
1008 Occasionally this really doesn't work, in which case you can split the
1016 class CollE s => Coll s a where
1017 insert :: s -> a -> s
1030 <sect2 id="instance-decls">
1031 <title>Instance declarations</title>
1039 <emphasis>Instance declarations may not overlap</emphasis>. The two instance
1044 instance context1 => C type1 where ...
1045 instance context2 => C type2 where ...
1049 "overlap" if <literal>type1</literal> and <literal>type2</literal> unify
1051 However, if you give the command line option
1052 <option>-fallow-overlapping-instances</option><indexterm><primary>-fallow-overlapping-instances
1053 option</primary></indexterm> then two overlapping instance declarations are permitted
1061 EITHER <literal>type1</literal> and <literal>type2</literal> do not unify
1067 OR <literal>type2</literal> is a substitution instance of <literal>type1</literal>
1068 (but not identical to <literal>type1</literal>)
1081 Notice that these rules
1088 make it clear which instance decl to use
1089 (pick the most specific one that matches)
1096 do not mention the contexts <literal>context1</literal>, <literal>context2</literal>
1097 Reason: you can pick which instance decl
1098 "matches" based on the type.
1105 Regrettably, GHC doesn't guarantee to detect overlapping instance
1106 declarations if they appear in different modules. GHC can "see" the
1107 instance declarations in the transitive closure of all the modules
1108 imported by the one being compiled, so it can "see" all instance decls
1109 when it is compiling <literal>Main</literal>. However, it currently chooses not
1110 to look at ones that can't possibly be of use in the module currently
1111 being compiled, in the interests of efficiency. (Perhaps we should
1112 change that decision, at least for <literal>Main</literal>.)
1119 <emphasis>There are no restrictions on the type in an instance
1120 <emphasis>head</emphasis>, except that at least one must not be a type variable</emphasis>.
1121 The instance "head" is the bit after the "=>" in an instance decl. For
1122 example, these are OK:
1126 instance C Int a where ...
1128 instance D (Int, Int) where ...
1130 instance E [[a]] where ...
1134 Note that instance heads <emphasis>may</emphasis> contain repeated type variables.
1135 For example, this is OK:
1139 instance Stateful (ST s) (MutVar s) where ...
1143 The "at least one not a type variable" restriction is to ensure that
1144 context reduction terminates: each reduction step removes one type
1145 constructor. For example, the following would make the type checker
1146 loop if it wasn't excluded:
1150 instance C a => C a where ...
1154 There are two situations in which the rule is a bit of a pain. First,
1155 if one allows overlapping instance declarations then it's quite
1156 convenient to have a "default instance" declaration that applies if
1157 something more specific does not:
1166 Second, sometimes you might want to use the following to get the
1167 effect of a "class synonym":
1171 class (C1 a, C2 a, C3 a) => C a where { }
1173 instance (C1 a, C2 a, C3 a) => C a where { }
1177 This allows you to write shorter signatures:
1189 f :: (C1 a, C2 a, C3 a) => ...
1193 I'm on the lookout for a simple rule that preserves decidability while
1194 allowing these idioms. The experimental flag
1195 <option>-fallow-undecidable-instances</option><indexterm><primary>-fallow-undecidable-instances
1196 option</primary></indexterm> lifts this restriction, allowing all the types in an
1197 instance head to be type variables.
1204 <emphasis>Unlike Haskell 1.4, instance heads may use type
1205 synonyms</emphasis>. As always, using a type synonym is just shorthand for
1206 writing the RHS of the type synonym definition. For example:
1210 type Point = (Int,Int)
1211 instance C Point where ...
1212 instance C [Point] where ...
1216 is legal. However, if you added
1220 instance C (Int,Int) where ...
1224 as well, then the compiler will complain about the overlapping
1225 (actually, identical) instance declarations. As always, type synonyms
1226 must be fully applied. You cannot, for example, write:
1231 instance Monad P where ...
1235 This design decision is independent of all the others, and easily
1236 reversed, but it makes sense to me.
1243 <emphasis>The types in an instance-declaration <emphasis>context</emphasis> must all
1244 be type variables</emphasis>. Thus
1248 instance C a b => Eq (a,b) where ...
1256 instance C Int b => Foo b where ...
1260 is not OK. Again, the intent here is to make sure that context
1261 reduction terminates.
1263 Voluminous correspondence on the Haskell mailing list has convinced me
1264 that it's worth experimenting with a more liberal rule. If you use
1265 the flag <option>-fallow-undecidable-instances</option> can use arbitrary
1266 types in an instance context. Termination is ensured by having a
1267 fixed-depth recursion stack. If you exceed the stack depth you get a
1268 sort of backtrace, and the opportunity to increase the stack depth
1269 with <option>-fcontext-stack</option><emphasis>N</emphasis>.
1282 <sect1 id="implicit-parameters">
1283 <title>Implicit parameters
1286 <para> Implicit paramters are implemented as described in
1287 "Implicit parameters: dynamic scoping with static types",
1288 J Lewis, MB Shields, E Meijer, J Launchbury,
1289 27th ACM Symposium on Principles of Programming Languages (POPL'00),
1294 There should be more documentation, but there isn't (yet). Yell if you need it.
1298 <para> You can't have an implicit parameter in the context of a class or instance
1299 declaration. For example, both these declarations are illegal:
1301 class (?x::Int) => C a where ...
1302 instance (?x::a) => Foo [a] where ...
1304 Reason: exactly which implicit parameter you pick up depends on exactly where
1305 you invoke a function. But the ``invocation'' of instance declarations is done
1306 behind the scenes by the compiler, so it's hard to figure out exactly where it is done.
1307 Easiest thing is to outlaw the offending types.</para>
1315 <sect1 id="functional-dependencies">
1316 <title>Functional dependencies
1319 <para> Functional dependencies are implemented as described by Mark Jones
1320 in "Type Classes with Functional Dependencies", Mark P. Jones,
1321 In Proceedings of the 9th European Symposium on Programming,
1322 ESOP 2000, Berlin, Germany, March 2000, Springer-Verlag LNCS 1782.
1326 There should be more documentation, but there isn't (yet). Yell if you need it.
1331 <sect1 id="universal-quantification">
1332 <title>Explicit universal quantification
1336 GHC's type system supports explicit universal quantification in
1337 constructor fields and function arguments. This is useful for things
1338 like defining <literal>runST</literal> from the state-thread world.
1339 GHC's syntax for this now agrees with Hugs's, namely:
1345 forall a b. (Ord a, Eq b) => a -> b -> a
1351 The context is, of course, optional. You can't use <literal>forall</literal> as
1352 a type variable any more!
1356 Haskell type signatures are implicitly quantified. The <literal>forall</literal>
1357 allows us to say exactly what this means. For example:
1375 g :: forall b. (b -> b)
1381 The two are treated identically.
1385 <title>Universally-quantified data type fields
1389 In a <literal>data</literal> or <literal>newtype</literal> declaration one can quantify
1390 the types of the constructor arguments. Here are several examples:
1396 data T a = T1 (forall b. b -> b -> b) a
1398 data MonadT m = MkMonad { return :: forall a. a -> m a,
1399 bind :: forall a b. m a -> (a -> m b) -> m b
1402 newtype Swizzle = MkSwizzle (Ord a => [a] -> [a])
1408 The constructors now have so-called <emphasis>rank 2</emphasis> polymorphic
1409 types, in which there is a for-all in the argument types.:
1415 T1 :: forall a. (forall b. b -> b -> b) -> a -> T a
1416 MkMonad :: forall m. (forall a. a -> m a)
1417 -> (forall a b. m a -> (a -> m b) -> m b)
1419 MkSwizzle :: (Ord a => [a] -> [a]) -> Swizzle
1425 Notice that you don't need to use a <literal>forall</literal> if there's an
1426 explicit context. For example in the first argument of the
1427 constructor <function>MkSwizzle</function>, an implicit "<literal>forall a.</literal>" is
1428 prefixed to the argument type. The implicit <literal>forall</literal>
1429 quantifies all type variables that are not already in scope, and are
1430 mentioned in the type quantified over.
1434 As for type signatures, implicit quantification happens for non-overloaded
1435 types too. So if you write this:
1438 data T a = MkT (Either a b) (b -> b)
1441 it's just as if you had written this:
1444 data T a = MkT (forall b. Either a b) (forall b. b -> b)
1447 That is, since the type variable <literal>b</literal> isn't in scope, it's
1448 implicitly universally quantified. (Arguably, it would be better
1449 to <emphasis>require</emphasis> explicit quantification on constructor arguments
1450 where that is what is wanted. Feedback welcomed.)
1456 <title>Construction </title>
1459 You construct values of types <literal>T1, MonadT, Swizzle</literal> by applying
1460 the constructor to suitable values, just as usual. For example,
1466 (T1 (\xy->x) 3) :: T Int
1468 (MkSwizzle sort) :: Swizzle
1469 (MkSwizzle reverse) :: Swizzle
1476 MkMonad r b) :: MonadT Maybe
1482 The type of the argument can, as usual, be more general than the type
1483 required, as <literal>(MkSwizzle reverse)</literal> shows. (<function>reverse</function>
1484 does not need the <literal>Ord</literal> constraint.)
1490 <title>Pattern matching</title>
1493 When you use pattern matching, the bound variables may now have
1494 polymorphic types. For example:
1500 f :: T a -> a -> (a, Char)
1501 f (T1 f k) x = (f k x, f 'c' 'd')
1503 g :: (Ord a, Ord b) => Swizzle -> [a] -> (a -> b) -> [b]
1504 g (MkSwizzle s) xs f = s (map f (s xs))
1506 h :: MonadT m -> [m a] -> m [a]
1507 h m [] = return m []
1508 h m (x:xs) = bind m x $ \y ->
1509 bind m (h m xs) $ \ys ->
1516 In the function <function>h</function> we use the record selectors <literal>return</literal>
1517 and <literal>bind</literal> to extract the polymorphic bind and return functions
1518 from the <literal>MonadT</literal> data structure, rather than using pattern
1523 You cannot pattern-match against an argument that is polymorphic.
1527 newtype TIM s a = TIM (ST s (Maybe a))
1529 runTIM :: (forall s. TIM s a) -> Maybe a
1530 runTIM (TIM m) = runST m
1536 Here the pattern-match fails, because you can't pattern-match against
1537 an argument of type <literal>(forall s. TIM s a)</literal>. Instead you
1538 must bind the variable and pattern match in the right hand side:
1541 runTIM :: (forall s. TIM s a) -> Maybe a
1542 runTIM tm = case tm of { TIM m -> runST m }
1545 The <literal>tm</literal> on the right hand side is (invisibly) instantiated, like
1546 any polymorphic value at its occurrence site, and now you can pattern-match
1553 <title>The partial-application restriction</title>
1556 There is really only one way in which data structures with polymorphic
1557 components might surprise you: you must not partially apply them.
1558 For example, this is illegal:
1564 map MkSwizzle [sort, reverse]
1570 The restriction is this: <emphasis>every subexpression of the program must
1571 have a type that has no for-alls, except that in a function
1572 application (f e1…en) the partial applications are not subject to
1573 this rule</emphasis>. The restriction makes type inference feasible.
1577 In the illegal example, the sub-expression <literal>MkSwizzle</literal> has the
1578 polymorphic type <literal>(Ord b => [b] -> [b]) -> Swizzle</literal> and is not
1579 a sub-expression of an enclosing application. On the other hand, this
1586 map (T1 (\a b -> a)) [1,2,3]
1592 even though it involves a partial application of <function>T1</function>, because
1593 the sub-expression <literal>T1 (\a b -> a)</literal> has type <literal>Int -> T
1600 <title>Type signatures
1604 Once you have data constructors with universally-quantified fields, or
1605 constants such as <Constant>runST</Constant> that have rank-2 types, it isn't long
1606 before you discover that you need more! Consider:
1612 mkTs f x y = [T1 f x, T1 f y]
1618 <function>mkTs</function> is a fuction that constructs some values of type
1619 <literal>T</literal>, using some pieces passed to it. The trouble is that since
1620 <literal>f</literal> is a function argument, Haskell assumes that it is
1621 monomorphic, so we'll get a type error when applying <function>T1</function> to
1622 it. This is a rather silly example, but the problem really bites in
1623 practice. Lots of people trip over the fact that you can't make
1624 "wrappers functions" for <Constant>runST</Constant> for exactly the same reason.
1625 In short, it is impossible to build abstractions around functions with
1630 The solution is fairly clear. We provide the ability to give a rank-2
1631 type signature for <emphasis>ordinary</emphasis> functions (not only data
1632 constructors), thus:
1638 mkTs :: (forall b. b -> b -> b) -> a -> [T a]
1639 mkTs f x y = [T1 f x, T1 f y]
1645 This type signature tells the compiler to attribute <literal>f</literal> with
1646 the polymorphic type <literal>(forall b. b -> b -> b)</literal> when type
1647 checking the body of <function>mkTs</function>, so now the application of
1648 <function>T1</function> is fine.
1652 There are two restrictions:
1661 You can only define a rank 2 type, specified by the following
1666 rank2type ::= [forall tyvars .] [context =>] funty
1667 funty ::= ([forall tyvars .] [context =>] ty) -> funty
1669 ty ::= ...current Haskell monotype syntax...
1673 Informally, the universal quantification must all be right at the beginning,
1674 or at the top level of a function argument.
1681 There is a restriction on the definition of a function whose
1682 type signature is a rank-2 type: the polymorphic arguments must be
1683 matched on the left hand side of the "<literal>=</literal>" sign. You can't
1684 define <function>mkTs</function> like this:
1688 mkTs :: (forall b. b -> b -> b) -> a -> [T a]
1689 mkTs = \ f x y -> [T1 f x, T1 f y]
1694 The same partial-application rule applies to ordinary functions with
1695 rank-2 types as applied to data constructors.
1708 <title>Type synonyms and hoisting
1712 GHC also allows you to write a <literal>forall</literal> in a type synonym, thus:
1714 type Discard a = forall b. a -> b -> a
1719 However, it is often convenient to use these sort of synonyms at the right hand
1720 end of an arrow, thus:
1722 type Discard a = forall b. a -> b -> a
1724 g :: Int -> Discard Int
1727 Simply expanding the type synonym would give
1729 g :: Int -> (forall b. Int -> b -> Int)
1731 but GHC "hoists" the <literal>forall</literal> to give the isomorphic type
1733 g :: forall b. Int -> Int -> b -> Int
1735 In general, the rule is this: <emphasis>to determine the type specified by any explicit
1736 user-written type (e.g. in a type signature), GHC expands type synonyms and then repeatedly
1737 performs the transformation:</emphasis>
1739 <emphasis>type1</emphasis> -> forall a. <emphasis>type2</emphasis>
1741 forall a. <emphasis>type1</emphasis> -> <emphasis>type2</emphasis>
1743 (In fact, GHC tries to retain as much synonym information as possible for use in
1744 error messages, but that is a usability issue.) This rule applies, of course, whether
1745 or not the <literal>forall</literal> comes from a synonym. For example, here is another
1746 valid way to write <literal>g</literal>'s type signature:
1748 g :: Int -> Int -> forall b. b -> Int
1755 <sect1 id="existential-quantification">
1756 <title>Existentially quantified data constructors
1760 The idea of using existential quantification in data type declarations
1761 was suggested by Laufer (I believe, thought doubtless someone will
1762 correct me), and implemented in Hope+. It's been in Lennart
1763 Augustsson's <Command>hbc</Command> Haskell compiler for several years, and
1764 proved very useful. Here's the idea. Consider the declaration:
1770 data Foo = forall a. MkFoo a (a -> Bool)
1777 The data type <literal>Foo</literal> has two constructors with types:
1783 MkFoo :: forall a. a -> (a -> Bool) -> Foo
1790 Notice that the type variable <literal>a</literal> in the type of <function>MkFoo</function>
1791 does not appear in the data type itself, which is plain <literal>Foo</literal>.
1792 For example, the following expression is fine:
1798 [MkFoo 3 even, MkFoo 'c' isUpper] :: [Foo]
1804 Here, <literal>(MkFoo 3 even)</literal> packages an integer with a function
1805 <function>even</function> that maps an integer to <literal>Bool</literal>; and <function>MkFoo 'c'
1806 isUpper</function> packages a character with a compatible function. These
1807 two things are each of type <literal>Foo</literal> and can be put in a list.
1811 What can we do with a value of type <literal>Foo</literal>?. In particular,
1812 what happens when we pattern-match on <function>MkFoo</function>?
1818 f (MkFoo val fn) = ???
1824 Since all we know about <literal>val</literal> and <function>fn</function> is that they
1825 are compatible, the only (useful) thing we can do with them is to
1826 apply <function>fn</function> to <literal>val</literal> to get a boolean. For example:
1833 f (MkFoo val fn) = fn val
1839 What this allows us to do is to package heterogenous values
1840 together with a bunch of functions that manipulate them, and then treat
1841 that collection of packages in a uniform manner. You can express
1842 quite a bit of object-oriented-like programming this way.
1845 <sect2 id="existential">
1846 <title>Why existential?
1850 What has this to do with <emphasis>existential</emphasis> quantification?
1851 Simply that <function>MkFoo</function> has the (nearly) isomorphic type
1857 MkFoo :: (exists a . (a, a -> Bool)) -> Foo
1863 But Haskell programmers can safely think of the ordinary
1864 <emphasis>universally</emphasis> quantified type given above, thereby avoiding
1865 adding a new existential quantification construct.
1871 <title>Type classes</title>
1874 An easy extension (implemented in <Command>hbc</Command>) is to allow
1875 arbitrary contexts before the constructor. For example:
1881 data Baz = forall a. Eq a => Baz1 a a
1882 | forall b. Show b => Baz2 b (b -> b)
1888 The two constructors have the types you'd expect:
1894 Baz1 :: forall a. Eq a => a -> a -> Baz
1895 Baz2 :: forall b. Show b => b -> (b -> b) -> Baz
1901 But when pattern matching on <function>Baz1</function> the matched values can be compared
1902 for equality, and when pattern matching on <function>Baz2</function> the first matched
1903 value can be converted to a string (as well as applying the function to it).
1904 So this program is legal:
1911 f (Baz1 p q) | p == q = "Yes"
1913 f (Baz1 v fn) = show (fn v)
1919 Operationally, in a dictionary-passing implementation, the
1920 constructors <function>Baz1</function> and <function>Baz2</function> must store the
1921 dictionaries for <literal>Eq</literal> and <literal>Show</literal> respectively, and
1922 extract it on pattern matching.
1926 Notice the way that the syntax fits smoothly with that used for
1927 universal quantification earlier.
1933 <title>Restrictions</title>
1936 There are several restrictions on the ways in which existentially-quantified
1937 constructors can be use.
1946 When pattern matching, each pattern match introduces a new,
1947 distinct, type for each existential type variable. These types cannot
1948 be unified with any other type, nor can they escape from the scope of
1949 the pattern match. For example, these fragments are incorrect:
1957 Here, the type bound by <function>MkFoo</function> "escapes", because <literal>a</literal>
1958 is the result of <function>f1</function>. One way to see why this is wrong is to
1959 ask what type <function>f1</function> has:
1963 f1 :: Foo -> a -- Weird!
1967 What is this "<literal>a</literal>" in the result type? Clearly we don't mean
1972 f1 :: forall a. Foo -> a -- Wrong!
1976 The original program is just plain wrong. Here's another sort of error
1980 f2 (Baz1 a b) (Baz1 p q) = a==q
1984 It's ok to say <literal>a==b</literal> or <literal>p==q</literal>, but
1985 <literal>a==q</literal> is wrong because it equates the two distinct types arising
1986 from the two <function>Baz1</function> constructors.
1994 You can't pattern-match on an existentially quantified
1995 constructor in a <literal>let</literal> or <literal>where</literal> group of
1996 bindings. So this is illegal:
2000 f3 x = a==b where { Baz1 a b = x }
2004 You can only pattern-match
2005 on an existentially-quantified constructor in a <literal>case</literal> expression or
2006 in the patterns of a function definition.
2008 The reason for this restriction is really an implementation one.
2009 Type-checking binding groups is already a nightmare without
2010 existentials complicating the picture. Also an existential pattern
2011 binding at the top level of a module doesn't make sense, because it's
2012 not clear how to prevent the existentially-quantified type "escaping".
2013 So for now, there's a simple-to-state restriction. We'll see how
2021 You can't use existential quantification for <literal>newtype</literal>
2022 declarations. So this is illegal:
2026 newtype T = forall a. Ord a => MkT a
2030 Reason: a value of type <literal>T</literal> must be represented as a pair
2031 of a dictionary for <literal>Ord t</literal> and a value of type <literal>t</literal>.
2032 That contradicts the idea that <literal>newtype</literal> should have no
2033 concrete representation. You can get just the same efficiency and effect
2034 by using <literal>data</literal> instead of <literal>newtype</literal>. If there is no
2035 overloading involved, then there is more of a case for allowing
2036 an existentially-quantified <literal>newtype</literal>, because the <literal>data</literal>
2037 because the <literal>data</literal> version does carry an implementation cost,
2038 but single-field existentially quantified constructors aren't much
2039 use. So the simple restriction (no existential stuff on <literal>newtype</literal>)
2040 stands, unless there are convincing reasons to change it.
2048 You can't use <literal>deriving</literal> to define instances of a
2049 data type with existentially quantified data constructors.
2051 Reason: in most cases it would not make sense. For example:#
2054 data T = forall a. MkT [a] deriving( Eq )
2057 To derive <literal>Eq</literal> in the standard way we would need to have equality
2058 between the single component of two <function>MkT</function> constructors:
2062 (MkT a) == (MkT b) = ???
2065 But <VarName>a</VarName> and <VarName>b</VarName> have distinct types, and so can't be compared.
2066 It's just about possible to imagine examples in which the derived instance
2067 would make sense, but it seems altogether simpler simply to prohibit such
2068 declarations. Define your own instances!
2080 <sect1 id="sec-assertions">
2082 <indexterm><primary>Assertions</primary></indexterm>
2086 If you want to make use of assertions in your standard Haskell code, you
2087 could define a function like the following:
2093 assert :: Bool -> a -> a
2094 assert False x = error "assertion failed!"
2101 which works, but gives you back a less than useful error message --
2102 an assertion failed, but which and where?
2106 One way out is to define an extended <function>assert</function> function which also
2107 takes a descriptive string to include in the error message and
2108 perhaps combine this with the use of a pre-processor which inserts
2109 the source location where <function>assert</function> was used.
2113 Ghc offers a helping hand here, doing all of this for you. For every
2114 use of <function>assert</function> in the user's source:
2120 kelvinToC :: Double -> Double
2121 kelvinToC k = assert (k >= 0.0) (k+273.15)
2127 Ghc will rewrite this to also include the source location where the
2134 assert pred val ==> assertError "Main.hs|15" pred val
2140 The rewrite is only performed by the compiler when it spots
2141 applications of <function>Exception.assert</function>, so you can still define and
2142 use your own versions of <function>assert</function>, should you so wish. If not,
2143 import <literal>Exception</literal> to make use <function>assert</function> in your code.
2147 To have the compiler ignore uses of assert, use the compiler option
2148 <option>-fignore-asserts</option>. <indexterm><primary>-fignore-asserts option</primary></indexterm> That is,
2149 expressions of the form <literal>assert pred e</literal> will be rewritten to <literal>e</literal>.
2153 Assertion failures can be caught, see the documentation for the
2154 <literal>Exception</literal> library (<xref linkend="sec-Exception">)
2160 <sect1 id="scoped-type-variables">
2161 <title>Scoped Type Variables
2165 A <emphasis>pattern type signature</emphasis> can introduce a <emphasis>scoped type
2166 variable</emphasis>. For example
2172 f (xs::[a]) = ys ++ ys
2181 The pattern <literal>(xs::[a])</literal> includes a type signature for <VarName>xs</VarName>.
2182 This brings the type variable <literal>a</literal> into scope; it scopes over
2183 all the patterns and right hand sides for this equation for <function>f</function>.
2184 In particular, it is in scope at the type signature for <VarName>y</VarName>.
2188 Pattern type signatures are completely orthogonal to ordinary, separate
2189 type signatures. The two can be used independently or together.
2190 At ordinary type signatures, such as that for <VarName>ys</VarName>, any type variables
2191 mentioned in the type signature <emphasis>that are not in scope</emphasis> are
2192 implicitly universally quantified. (If there are no type variables in
2193 scope, all type variables mentioned in the signature are universally
2194 quantified, which is just as in Haskell 98.) In this case, since <VarName>a</VarName>
2195 is in scope, it is not universally quantified, so the type of <VarName>ys</VarName> is
2196 the same as that of <VarName>xs</VarName>. In Haskell 98 it is not possible to declare
2197 a type for <VarName>ys</VarName>; a major benefit of scoped type variables is that
2198 it becomes possible to do so.
2202 Scoped type variables are implemented in both GHC and Hugs. Where the
2203 implementations differ from the specification below, those differences
2208 So much for the basic idea. Here are the details.
2212 <title>What a pattern type signature means</title>
2214 A type variable brought into scope by a pattern type signature is simply
2215 the name for a type. The restriction they express is that all occurrences
2216 of the same name mean the same type. For example:
2218 f :: [Int] -> Int -> Int
2219 f (xs::[a]) (y::a) = (head xs + y) :: a
2221 The pattern type signatures on the left hand side of
2222 <literal>f</literal> express the fact that <literal>xs</literal>
2223 must be a list of things of some type <literal>a</literal>; and that <literal>y</literal>
2224 must have this same type. The type signature on the expression <literal>(head xs)</literal>
2225 specifies that this expression must have the same type <literal>a</literal>.
2226 <emphasis>There is no requirement that the type named by "<literal>a</literal>" is
2227 in fact a type variable</emphasis>. Indeed, in this case, the type named by "<literal>a</literal>" is
2228 <literal>Int</literal>. (This is a slight liberalisation from the original rather complex
2229 rules, which specified that a pattern-bound type variable should be universally quantified.)
2230 For example, all of these are legal:</para>
2233 t (x::a) (y::a) = x+y*2
2235 f (x::a) (y::b) = [x,y] -- a unifies with b
2237 g (x::a) = x + 1::Int -- a unifies with Int
2239 h x = let k (y::a) = [x,y] -- a is free in the
2240 in k x -- environment
2242 k (x::a) True = ... -- a unifies with Int
2243 k (x::Int) False = ...
2246 w (x::a) = x -- a unifies with [b]
2252 <title>Scope and implicit quantification</title>
2260 All the type variables mentioned in a pattern,
2261 that are not already in scope,
2262 are brought into scope by the pattern. We describe this set as
2263 the <emphasis>type variables bound by the pattern</emphasis>.
2266 f (x::a) = let g (y::(a,b)) = fst y
2270 The pattern <literal>(x::a)</literal> brings the type variable
2271 <literal>a</literal> into scope, as well as the term
2272 variable <literal>x</literal>. The pattern <literal>(y::(a,b))</literal>
2273 contains an occurrence of the already-in-scope type variable <literal>a</literal>,
2274 and brings into scope the type variable <literal>b</literal>.
2280 The type variables thus brought into scope may be mentioned
2281 in ordinary type signatures or pattern type signatures anywhere within
2289 In ordinary type signatures, any type variable mentioned in the
2290 signature that is in scope is <emphasis>not</emphasis> universally quantified.
2298 Ordinary type signatures do not bring any new type variables
2299 into scope (except in the type signature itself!). So this is illegal:
2306 It's illegal because <VarName>a</VarName> is not in scope in the body of <function>f</function>,
2307 so the ordinary signature <literal>x::a</literal> is equivalent to <literal>x::forall a.a</literal>;
2308 and that is an incorrect typing.
2315 There is no implicit universal quantification on pattern type
2316 signatures, nor may one write an explicit <literal>forall</literal> type in a pattern
2317 type signature. The pattern type signature is a monotype.
2325 The type variables in the head of a <literal>class</literal> or <literal>instance</literal> declaration
2326 scope over the methods defined in the <literal>where</literal> part. For example:
2340 (Not implemented in Hugs yet, Dec 98).
2351 <title>Result type signatures</title>
2359 The result type of a function can be given a signature,
2364 f (x::a) :: [a] = [x,x,x]
2368 The final <literal>:: [a]</literal> after all the patterns gives a signature to the
2369 result type. Sometimes this is the only way of naming the type variable
2374 f :: Int -> [a] -> [a]
2375 f n :: ([a] -> [a]) = let g (x::a, y::a) = (y,x)
2376 in \xs -> map g (reverse xs `zip` xs)
2388 Result type signatures are not yet implemented in Hugs.
2394 <title>Where a pattern type signature can occur</title>
2397 A pattern type signature can occur in any pattern, but there
2398 are restrictions on pattern bindings:
2403 A pattern type signature can be on an arbitrary sub-pattern, not
2408 f ((x,y)::(a,b)) = (y,x) :: (b,a)
2417 Pattern type signatures, including the result part, can be used
2418 in lambda abstractions:
2421 (\ (x::a, y) :: a -> x)
2428 Pattern type signatures, including the result part, can be used
2429 in <literal>case</literal> expressions:
2433 case e of { (x::a, y) :: a -> x }
2441 To avoid ambiguity, the type after the “<literal>::</literal>” in a result
2442 pattern signature on a lambda or <literal>case</literal> must be atomic (i.e. a single
2443 token or a parenthesised type of some sort). To see why,
2444 consider how one would parse this:
2458 Pattern type signatures can bind existential type variables.
2463 data T = forall a. MkT [a]
2466 f (MkT [t::a]) = MkT t3
2479 Pattern type signatures that bind new type variables
2480 may not be used in pattern bindings at all.
2485 f x = let (y, z::a) = x in ...
2489 But these are OK, because they do not bind fresh type variables:
2493 f1 x = let (y, z::Int) = x in ...
2494 f2 (x::(Int,a)) = let (y, z::a) = x in ...
2498 However a single variable is considered a degenerate function binding,
2499 rather than a degerate pattern binding, so this is permitted, even
2500 though it binds a type variable:
2504 f :: (b->b) = \(x::b) -> x
2513 Such degnerate function bindings do not fall under the monomorphism
2520 g :: a -> a -> Bool = \x y. x==y
2526 Here <function>g</function> has type <literal>forall a. Eq a => a -> a -> Bool</literal>, just as if
2527 <function>g</function> had a separate type signature. Lacking a type signature, <function>g</function>
2528 would get a monomorphic type.
2536 <sect1 id="pragmas">
2541 GHC supports several pragmas, or instructions to the compiler placed
2542 in the source code. Pragmas don't affect the meaning of the program,
2543 but they might affect the efficiency of the generated code.
2546 <sect2 id="inline-pragma">
2547 <title>INLINE pragma
2549 <indexterm><primary>INLINE pragma</primary></indexterm>
2550 <indexterm><primary>pragma, INLINE</primary></indexterm></title>
2553 GHC (with <option>-O</option>, as always) tries to inline (or “unfold”)
2554 functions/values that are “small enough,” thus avoiding the call
2555 overhead and possibly exposing other more-wonderful optimisations.
2559 You will probably see these unfoldings (in Core syntax) in your
2564 Normally, if GHC decides a function is “too expensive” to inline, it
2565 will not do so, nor will it export that unfolding for other modules to
2570 The sledgehammer you can bring to bear is the
2571 <literal>INLINE</literal><indexterm><primary>INLINE pragma</primary></indexterm> pragma, used thusly:
2574 key_function :: Int -> String -> (Bool, Double)
2576 #ifdef __GLASGOW_HASKELL__
2577 {-# INLINE key_function #-}
2581 (You don't need to do the C pre-processor carry-on unless you're going
2582 to stick the code through HBC—it doesn't like <literal>INLINE</literal> pragmas.)
2586 The major effect of an <literal>INLINE</literal> pragma is to declare a function's
2587 “cost” to be very low. The normal unfolding machinery will then be
2588 very keen to inline it.
2592 An <literal>INLINE</literal> pragma for a function can be put anywhere its type
2593 signature could be put.
2597 <literal>INLINE</literal> pragmas are a particularly good idea for the
2598 <literal>then</literal>/<literal>return</literal> (or <literal>bind</literal>/<literal>unit</literal>) functions in a monad.
2599 For example, in GHC's own <literal>UniqueSupply</literal> monad code, we have:
2602 #ifdef __GLASGOW_HASKELL__
2603 {-# INLINE thenUs #-}
2604 {-# INLINE returnUs #-}
2612 <sect2 id="noinline-pragma">
2613 <title>NOINLINE pragma
2617 <indexterm><primary>NOINLINE pragma</primary></indexterm>
2618 <indexterm><primary>pragma, NOINLINE</primary></indexterm>
2622 The <literal>NOINLINE</literal> pragma does exactly what you'd expect: it stops the
2623 named function from being inlined by the compiler. You shouldn't ever
2624 need to do this, unless you're very cautious about code size.
2629 <sect2 id="specialize-pragma">
2630 <title>SPECIALIZE pragma</title>
2632 <indexterm><primary>SPECIALIZE pragma</primary></indexterm>
2633 <indexterm><primary>pragma, SPECIALIZE</primary></indexterm>
2634 <indexterm><primary>overloading, death to</primary></indexterm>
2636 <para>(UK spelling also accepted.) For key overloaded
2637 functions, you can create extra versions (NB: more code space)
2638 specialised to particular types. Thus, if you have an
2639 overloaded function:</para>
2642 hammeredLookup :: Ord key => [(key, value)] -> key -> value
2645 <para>If it is heavily used on lists with
2646 <literal>Widget</literal> keys, you could specialise it as
2650 {-# SPECIALIZE hammeredLookup :: [(Widget, value)] -> Widget -> value #-}
2653 <para>To get very fancy, you can also specify a named function
2654 to use for the specialised value, as in:</para>
2657 {-# RULES hammeredLookup = blah #-}
2660 <para>where <literal>blah</literal> is an implementation of
2661 <literal>hammerdLookup</literal> written specialy for
2662 <literal>Widget</literal> lookups. It's <emphasis>Your
2663 Responsibility</emphasis> to make sure that
2664 <function>blah</function> really behaves as a specialised
2665 version of <function>hammeredLookup</function>!!!</para>
2667 <para>Note we use the <literal>RULE</literal> pragma here to
2668 indicate that <literal>hammeredLookup</literal> applied at a
2669 certain type should be replaced by <literal>blah</literal>. See
2670 <xref linkend="rules"> for more information on
2671 <literal>RULES</literal>.</para>
2673 <para>An example in which using <literal>RULES</literal> for
2674 specialisation will Win Big:
2677 toDouble :: Real a => a -> Double
2678 toDouble = fromRational . toRational
2680 {-# SPECIALIZE toDouble :: Int -> Double = i2d #-}
2681 i2d (I# i) = D# (int2Double# i) -- uses Glasgow prim-op directly
2684 The <function>i2d</function> function is virtually one machine
2685 instruction; the default conversion—via an intermediate
2686 <literal>Rational</literal>—is obscenely expensive by
2689 <para>A <literal>SPECIALIZE</literal> pragma for a function can
2690 be put anywhere its type signature could be put.</para>
2694 <sect2 id="specialize-instance-pragma">
2695 <title>SPECIALIZE instance pragma
2699 <indexterm><primary>SPECIALIZE pragma</primary></indexterm>
2700 <indexterm><primary>overloading, death to</primary></indexterm>
2701 Same idea, except for instance declarations. For example:
2704 instance (Eq a) => Eq (Foo a) where { ... usual stuff ... }
2706 {-# SPECIALIZE instance Eq (Foo [(Int, Bar)] #-}
2709 Compatible with HBC, by the way.
2714 <sect2 id="line-pragma">
2719 <indexterm><primary>LINE pragma</primary></indexterm>
2720 <indexterm><primary>pragma, LINE</primary></indexterm>
2724 This pragma is similar to C's <literal>#line</literal> pragma, and is mainly for use in
2725 automatically generated Haskell code. It lets you specify the line
2726 number and filename of the original code; for example
2732 {-# LINE 42 "Foo.vhs" #-}
2738 if you'd generated the current file from something called <filename>Foo.vhs</filename>
2739 and this line corresponds to line 42 in the original. GHC will adjust
2740 its error messages to refer to the line/file named in the <literal>LINE</literal>
2747 <title>RULES pragma</title>
2750 The RULES pragma lets you specify rewrite rules. It is described in
2751 <xref LinkEnd="rewrite-rules">.
2756 <sect2 id="deprecated-pragma">
2757 <title>DEPRECATED pragma</title>
2760 The DEPRECATED pragma lets you specify that a particular function, class, or type, is deprecated.
2761 There are two forms.
2765 You can deprecate an entire module thus:</para>
2767 module Wibble {-# DEPRECATED "Use Wobble instead" #-} where
2771 When you compile any module that import <literal>Wibble</literal>, GHC will print
2772 the specified message.</para>
2777 You can deprecate a function, class, or type, with the following top-level declaration:
2780 {-# DEPRECATED f, C, T "Don't use these" #-}
2783 When you compile any module that imports and uses any of the specifed entities,
2784 GHC will print the specified message.
2788 <para>You can suppress the warnings with the flag <option>-fno-warn-deprecations</option>.</para>
2794 <sect1 id="rewrite-rules">
2795 <title>Rewrite rules
2797 <indexterm><primary>RULES pagma</primary></indexterm>
2798 <indexterm><primary>pragma, RULES</primary></indexterm>
2799 <indexterm><primary>rewrite rules</primary></indexterm></title>
2802 The programmer can specify rewrite rules as part of the source program
2803 (in a pragma). GHC applies these rewrite rules wherever it can.
2811 "map/map" forall f g xs. map f (map g xs) = map (f.g) xs
2818 <title>Syntax</title>
2821 From a syntactic point of view:
2827 Each rule has a name, enclosed in double quotes. The name itself has
2828 no significance at all. It is only used when reporting how many times the rule fired.
2834 There may be zero or more rules in a <literal>RULES</literal> pragma.
2840 Layout applies in a <literal>RULES</literal> pragma. Currently no new indentation level
2841 is set, so you must lay out your rules starting in the same column as the
2842 enclosing definitions.
2848 Each variable mentioned in a rule must either be in scope (e.g. <function>map</function>),
2849 or bound by the <literal>forall</literal> (e.g. <function>f</function>, <function>g</function>, <function>xs</function>). The variables bound by
2850 the <literal>forall</literal> are called the <emphasis>pattern</emphasis> variables. They are separated
2851 by spaces, just like in a type <literal>forall</literal>.
2857 A pattern variable may optionally have a type signature.
2858 If the type of the pattern variable is polymorphic, it <emphasis>must</emphasis> have a type signature.
2859 For example, here is the <literal>foldr/build</literal> rule:
2862 "fold/build" forall k z (g::forall b. (a->b->b) -> b -> b) .
2863 foldr k z (build g) = g k z
2866 Since <function>g</function> has a polymorphic type, it must have a type signature.
2873 The left hand side of a rule must consist of a top-level variable applied
2874 to arbitrary expressions. For example, this is <emphasis>not</emphasis> OK:
2877 "wrong1" forall e1 e2. case True of { True -> e1; False -> e2 } = e1
2878 "wrong2" forall f. f True = True
2881 In <literal>"wrong1"</literal>, the LHS is not an application; in <literal>"wrong2"</literal>, the LHS has a pattern variable
2888 A rule does not need to be in the same module as (any of) the
2889 variables it mentions, though of course they need to be in scope.
2895 Rules are automatically exported from a module, just as instance declarations are.
2906 <title>Semantics</title>
2909 From a semantic point of view:
2915 Rules are only applied if you use the <option>-O</option> flag.
2921 Rules are regarded as left-to-right rewrite rules.
2922 When GHC finds an expression that is a substitution instance of the LHS
2923 of a rule, it replaces the expression by the (appropriately-substituted) RHS.
2924 By "a substitution instance" we mean that the LHS can be made equal to the
2925 expression by substituting for the pattern variables.
2932 The LHS and RHS of a rule are typechecked, and must have the
2940 GHC makes absolutely no attempt to verify that the LHS and RHS
2941 of a rule have the same meaning. That is undecideable in general, and
2942 infeasible in most interesting cases. The responsibility is entirely the programmer's!
2949 GHC makes no attempt to make sure that the rules are confluent or
2950 terminating. For example:
2953 "loop" forall x,y. f x y = f y x
2956 This rule will cause the compiler to go into an infinite loop.
2963 If more than one rule matches a call, GHC will choose one arbitrarily to apply.
2969 GHC currently uses a very simple, syntactic, matching algorithm
2970 for matching a rule LHS with an expression. It seeks a substitution
2971 which makes the LHS and expression syntactically equal modulo alpha
2972 conversion. The pattern (rule), but not the expression, is eta-expanded if
2973 necessary. (Eta-expanding the epression can lead to laziness bugs.)
2974 But not beta conversion (that's called higher-order matching).
2978 Matching is carried out on GHC's intermediate language, which includes
2979 type abstractions and applications. So a rule only matches if the
2980 types match too. See <xref LinkEnd="rule-spec"> below.
2986 GHC keeps trying to apply the rules as it optimises the program.
2987 For example, consider:
2996 The expression <literal>s (t xs)</literal> does not match the rule <literal>"map/map"</literal>, but GHC
2997 will substitute for <VarName>s</VarName> and <VarName>t</VarName>, giving an expression which does match.
2998 If <VarName>s</VarName> or <VarName>t</VarName> was (a) used more than once, and (b) large or a redex, then it would
2999 not be substituted, and the rule would not fire.
3006 In the earlier phases of compilation, GHC inlines <emphasis>nothing
3007 that appears on the LHS of a rule</emphasis>, because once you have substituted
3008 for something you can't match against it (given the simple minded
3009 matching). So if you write the rule
3012 "map/map" forall f,g. map f . map g = map (f.g)
3015 this <emphasis>won't</emphasis> match the expression <literal>map f (map g xs)</literal>.
3016 It will only match something written with explicit use of ".".
3017 Well, not quite. It <emphasis>will</emphasis> match the expression
3023 where <function>wibble</function> is defined:
3026 wibble f g = map f . map g
3029 because <function>wibble</function> will be inlined (it's small).
3031 Later on in compilation, GHC starts inlining even things on the
3032 LHS of rules, but still leaves the rules enabled. This inlining
3033 policy is controlled by the per-simplification-pass flag <option>-finline-phase</option><emphasis>n</emphasis>.
3040 All rules are implicitly exported from the module, and are therefore
3041 in force in any module that imports the module that defined the rule, directly
3042 or indirectly. (That is, if A imports B, which imports C, then C's rules are
3043 in force when compiling A.) The situation is very similar to that for instance
3055 <title>List fusion</title>
3058 The RULES mechanism is used to implement fusion (deforestation) of common list functions.
3059 If a "good consumer" consumes an intermediate list constructed by a "good producer", the
3060 intermediate list should be eliminated entirely.
3064 The following are good producers:
3076 Enumerations of <literal>Int</literal> and <literal>Char</literal> (e.g. <literal>['a'..'z']</literal>).
3082 Explicit lists (e.g. <literal>[True, False]</literal>)
3088 The cons constructor (e.g <literal>3:4:[]</literal>)
3094 <function>++</function>
3100 <function>map</function>
3106 <function>filter</function>
3112 <function>iterate</function>, <function>repeat</function>
3118 <function>zip</function>, <function>zipWith</function>
3127 The following are good consumers:
3139 <function>array</function> (on its second argument)
3145 <function>length</function>
3151 <function>++</function> (on its first argument)
3157 <function>map</function>
3163 <function>filter</function>
3169 <function>concat</function>
3175 <function>unzip</function>, <function>unzip2</function>, <function>unzip3</function>, <function>unzip4</function>
3181 <function>zip</function>, <function>zipWith</function> (but on one argument only; if both are good producers, <function>zip</function>
3182 will fuse with one but not the other)
3188 <function>partition</function>
3194 <function>head</function>
3200 <function>and</function>, <function>or</function>, <function>any</function>, <function>all</function>
3206 <function>sequence_</function>
3212 <function>msum</function>
3218 <function>sortBy</function>
3227 So, for example, the following should generate no intermediate lists:
3230 array (1,10) [(i,i*i) | i <- map (+ 1) [0..9]]
3236 This list could readily be extended; if there are Prelude functions that you use
3237 a lot which are not included, please tell us.
3241 If you want to write your own good consumers or producers, look at the
3242 Prelude definitions of the above functions to see how to do so.
3247 <sect2 id="rule-spec">
3248 <title>Specialisation
3252 Rewrite rules can be used to get the same effect as a feature
3253 present in earlier version of GHC:
3256 {-# SPECIALIZE fromIntegral :: Int8 -> Int16 = int8ToInt16 #-}
3259 This told GHC to use <function>int8ToInt16</function> instead of <function>fromIntegral</function> whenever
3260 the latter was called with type <literal>Int8 -> Int16</literal>. That is, rather than
3261 specialising the original definition of <function>fromIntegral</function> the programmer is
3262 promising that it is safe to use <function>int8ToInt16</function> instead.
3266 This feature is no longer in GHC. But rewrite rules let you do the
3271 "fromIntegral/Int8/Int16" fromIntegral = int8ToInt16
3275 This slightly odd-looking rule instructs GHC to replace <function>fromIntegral</function>
3276 by <function>int8ToInt16</function> <emphasis>whenever the types match</emphasis>. Speaking more operationally,
3277 GHC adds the type and dictionary applications to get the typed rule
3280 forall (d1::Integral Int8) (d2::Num Int16) .
3281 fromIntegral Int8 Int16 d1 d2 = int8ToInt16
3285 this rule does not need to be in the same file as fromIntegral,
3286 unlike the <literal>SPECIALISE</literal> pragmas which currently do (so that they
3287 have an original definition available to specialise).
3293 <title>Controlling what's going on</title>
3301 Use <option>-ddump-rules</option> to see what transformation rules GHC is using.
3307 Use <option>-ddump-simpl-stats</option> to see what rules are being fired.
3308 If you add <option>-dppr-debug</option> you get a more detailed listing.
3314 The defintion of (say) <function>build</function> in <FileName>PrelBase.lhs</FileName> looks llike this:
3317 build :: forall a. (forall b. (a -> b -> b) -> b -> b) -> [a]
3318 {-# INLINE build #-}
3322 Notice the <literal>INLINE</literal>! That prevents <literal>(:)</literal> from being inlined when compiling
3323 <literal>PrelBase</literal>, so that an importing module will “see” the <literal>(:)</literal>, and can
3324 match it on the LHS of a rule. <literal>INLINE</literal> prevents any inlining happening
3325 in the RHS of the <literal>INLINE</literal> thing. I regret the delicacy of this.
3332 In <filename>ghc/lib/std/PrelBase.lhs</filename> look at the rules for <function>map</function> to
3333 see how to write rules that will do fusion and yet give an efficient
3334 program even if fusion doesn't happen. More rules in <filename>PrelList.lhs</filename>.
3346 <sect1 id="generic-classes">
3347 <title>Generic classes</title>
3350 The ideas behind this extension are described in detail in "Derivable type classes",
3351 Ralf Hinze and Simon Peyton Jones, Haskell Workshop, Montreal Sept 2000, pp94-105.
3352 An example will give the idea:
3360 fromBin :: [Int] -> (a, [Int])
3362 toBin {| Unit |} Unit = []
3363 toBin {| a :+: b |} (Inl x) = 0 : toBin x
3364 toBin {| a :+: b |} (Inr y) = 1 : toBin y
3365 toBin {| a :*: b |} (x :*: y) = toBin x ++ toBin y
3367 fromBin {| Unit |} bs = (Unit, bs)
3368 fromBin {| a :+: b |} (0:bs) = (Inl x, bs') where (x,bs') = fromBin bs
3369 fromBin {| a :+: b |} (1:bs) = (Inr y, bs') where (y,bs') = fromBin bs
3370 fromBin {| a :*: b |} bs = (x :*: y, bs'') where (x,bs' ) = fromBin bs
3371 (y,bs'') = fromBin bs'
3374 This class declaration explains how <literal>toBin</literal> and <literal>fromBin</literal>
3375 work for arbitrary data types. They do so by giving cases for unit, product, and sum,
3376 which are defined thus in the library module <literal>Generics</literal>:
3380 data a :+: b = Inl a | Inr b
3381 data a :*: b = a :*: b
3384 Now you can make a data type into an instance of Bin like this:
3386 instance (Bin a, Bin b) => Bin (a,b)
3387 instance Bin a => Bin [a]
3389 That is, just leave off the "where" clasuse. Of course, you can put in the
3390 where clause and over-ride whichever methods you please.
3394 <title> Using generics </title>
3395 <para>To use generics you need to</para>
3398 <para>Use the flags <option>-fglasgow-exts</option> (to enable the extra syntax),
3399 <option>-fgenerics</option> (to generate extra per-data-type code),
3400 and <option>-package lang</option> (to make the <literal>Generics</literal> library
3404 <para>Import the module <literal>Generics</literal> from the
3405 <literal>lang</literal> package. This import brings into
3406 scope the data types <literal>Unit</literal>,
3407 <literal>:*:</literal>, and <literal>:+:</literal>. (You
3408 don't need this import if you don't mention these types
3409 explicitly; for example, if you are simply giving instance
3410 declarations.)</para>
3415 <sect2> <title> Changes wrt the paper </title>
3417 Note that the type constructors <literal>:+:</literal> and <literal>:*:</literal>
3418 can be written infix (indeed, you can now use
3419 any operator starting in a colon as an infix type constructor). Also note that
3420 the type constructors are not exactly as in the paper (Unit instead of 1, etc).
3421 Finally, note that the syntax of the type patterns in the class declaration
3422 uses "<literal>{|</literal>" and "<literal>|}</literal>" brackets; curly braces
3423 alone would ambiguous when they appear on right hand sides (an extension we
3424 anticipate wanting).
3428 <sect2> <title>Terminology and restrictions</title>
3430 Terminology. A "generic default method" in a class declaration
3431 is one that is defined using type patterns as above.
3432 A "polymorphic default method" is a default method defined as in Haskell 98.
3433 A "generic class declaration" is a class declaration with at least one
3434 generic default method.
3442 Alas, we do not yet implement the stuff about constructor names and
3449 A generic class can have only one parameter; you can't have a generic
3450 multi-parameter class.
3456 A default method must be defined entirely using type patterns, or entirely
3457 without. So this is illegal:
3460 op :: a -> (a, Bool)
3461 op {| Unit |} Unit = (Unit, True)
3464 However it is perfectly OK for some methods of a generic class to have
3465 generic default methods and others to have polymorphic default methods.
3471 The type variable(s) in the type pattern for a generic method declaration
3472 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:
3476 op {| p :*: q |} (x :*: y) = op (x :: p)
3484 The type patterns in a generic default method must take one of the forms:
3490 where "a" and "b" are type variables. Furthermore, all the type patterns for
3491 a single type constructor (<literal>:*:</literal>, say) must be identical; they
3492 must use the same type variables. So this is illegal:
3496 op {| a :+: b |} (Inl x) = True
3497 op {| p :+: q |} (Inr y) = False
3499 The type patterns must be identical, even in equations for different methods of the class.
3500 So this too is illegal:
3504 op {| a :*: b |} (Inl x) = True
3507 op {| p :*: q |} (Inr y) = False
3509 (The reason for this restriction is that we gather all the equations for a particular type consructor
3510 into a single generic instance declaration.)
3516 A generic method declaration must give a case for each of the three type constructors.
3522 The type for a generic method can be built only from:
3524 <listitem> <para> Function arrows </para> </listitem>
3525 <listitem> <para> Type variables </para> </listitem>
3526 <listitem> <para> Tuples </para> </listitem>
3527 <listitem> <para> Arbitrary types not involving type variables </para> </listitem>
3529 Here are some example type signatures for generic methods:
3532 op2 :: Bool -> (a,Bool)
3533 op3 :: [Int] -> a -> a
3536 Here, op1, op2, op3 are OK, but op4 is rejected, because it has a type variable
3540 This restriction is an implementation restriction: we just havn't got around to
3541 implementing the necessary bidirectional maps over arbitrary type constructors.
3542 It would be relatively easy to add specific type constructors, such as Maybe and list,
3543 to the ones that are allowed.</para>
3548 In an instance declaration for a generic class, the idea is that the compiler
3549 will fill in the methods for you, based on the generic templates. However it can only
3554 The instance type is simple (a type constructor applied to type variables, as in Haskell 98).
3559 No constructor of the instance type has unboxed fields.
3563 (Of course, these things can only arise if you are already using GHC extensions.)
3564 However, you can still give an instance declarations for types which break these rules,
3565 provided you give explicit code to override any generic default methods.
3573 The option <option>-ddump-deriv</option> dumps incomprehensible stuff giving details of
3574 what the compiler does with generic declarations.
3579 <sect2> <title> Another example </title>
3581 Just to finish with, here's another example I rather like:
3585 nCons {| Unit |} _ = 1
3586 nCons {| a :*: b |} _ = 1
3587 nCons {| a :+: b |} _ = nCons (bot::a) + nCons (bot::b)
3590 tag {| Unit |} _ = 1
3591 tag {| a :*: b |} _ = 1
3592 tag {| a :+: b |} (Inl x) = tag x
3593 tag {| a :+: b |} (Inr y) = nCons (bot::a) + tag y
3600 ;;; Local Variables: ***
3602 ;;; sgml-parent-document: ("users_guide.sgml" "book" "chapter" "sect1") ***