1 <?xml version="1.0" encoding="iso-8859-1"?>
2 <chapter id="bugs-and-infelicities">
3 <title>Known bugs and infelicities</title>
5 <sect1 id="vs-Haskell-defn">
6 <title>Haskell 98 vs. Glasgow Haskell: language non-compliance
9 <indexterm><primary>GHC vs the Haskell 98 language</primary></indexterm>
10 <indexterm><primary>Haskell 98 language vs GHC</primary></indexterm>
12 <para>This section lists Glasgow Haskell infelicities in its
13 implementation of Haskell 98. See also the “when things
14 go wrong” section (<xref linkend="wrong"/>) for information
15 about crashes, space leaks, and other undesirable phenomena.</para>
17 <para>The limitations here are listed in Haskell Report order
20 <sect2 id="haskell98-divergence">
21 <title>Divergence from Haskell 98</title>
24 <sect3 id="infelicities-lexical">
25 <title>Lexical syntax</title>
29 <para>Certain lexical rules regarding qualified identifiers
30 are slightly different in GHC compared to the Haskell
32 <replaceable>module</replaceable><literal>.</literal><replaceable>reservedop</replaceable>,
33 such as <literal>M.\</literal>, GHC will interpret it as a
34 single qualified operator rather than the two lexemes
35 <literal>M</literal> and <literal>.\</literal>.</para>
40 <sect3 id="infelicities-syntax">
41 <title>Context-free syntax</title>
45 <para>GHC is a little less strict about the layout rule when used
46 in <literal>do</literal> expressions. Specifically, the
47 restriction that "a nested context must be indented further to
48 the right than the enclosing context" is relaxed to allow the
49 nested context to be at the same level as the enclosing context,
50 if the enclosing context is a <literal>do</literal>
53 <para>For example, the following code is accepted by GHC:
56 main = do args <- getArgs
57 if null args then return [] else do
58 ps <- mapM process args
59 mapM print ps</programlisting>
65 <para>GHC doesn't do fixity resolution in expressions during
66 parsing. For example, according to the Haskell report, the
67 following expression is legal Haskell:
69 let x = 42 in x == 42 == True</programlisting>
72 (let x = 42 in x == 42) == True</programlisting>
74 because according to the report, the <literal>let</literal>
75 expression <quote>extends as far to the right as
76 possible</quote>. Since it can't extend past the second
77 equals sign without causing a parse error
78 (<literal>==</literal> is non-fix), the
79 <literal>let</literal>-expression must terminate there. GHC
80 simply gobbles up the whole expression, parsing like this:
82 (let x = 42 in x == 42 == True)</programlisting>
84 The Haskell report is arguably wrong here, but nevertheless
85 it's a difference between GHC & Haskell 98.</para>
90 <sect3 id="infelicities-exprs-pats">
91 <title>Expressions and patterns</title>
93 <para>None known.</para>
96 <sect3 id="infelicities-decls">
97 <title>Declarations and bindings</title>
99 <para>GHC's typechecker makes all pattern bindings monomorphic
100 by default; this behaviour can be disabled with
101 <option>-XNoMonoPatBinds</option>. See <xref
102 linkend="options-language" />.</para>
105 <sect3 id="infelicities-Modules">
106 <title>Module system and interface files</title>
108 <para>GHC requires the use of <literal>hs-boot</literal>
109 files to cut the recursive loops among mutually recursive modules
110 as described in <xref linkend="mutual-recursion"/>. This more of an infelicity
111 than a bug: the Haskell Report says
112 (<ulink url="http://haskell.org/onlinereport/modules.html#sect5.7">Section 5.7</ulink>) "Depending on the Haskell
113 implementation used, separate compilation of mutually
114 recursive modules may require that imported modules contain
115 additional information so that they may be referenced before
116 they are compiled. Explicit type signatures for all exported
117 values may be necessary to deal with mutual recursion. The
118 precise details of separate compilation are not defined by
125 <sect3 id="infelicities-numbers">
126 <title>Numbers, basic types, and built-in classes</title>
130 <term>Multiply-defined array elements—not checked:</term>
132 <para>This code fragment should
133 elicit a fatal error, but it does not:
136 main = print (array (1,1) [(1,2), (1,3)])</programlisting>
137 GHC's implementation of <literal>array</literal> takes the value of an
138 array slot from the last (index,value) pair in the list, and does no
139 checking for duplicates. The reason for this is efficiency, pure and simple.
147 <sect3 id="infelicities-Prelude">
148 <title>In <literal>Prelude</literal> support</title>
152 <term>Arbitrary-sized tuples</term>
154 <para>Tuples are currently limited to size 100. HOWEVER:
155 standard instances for tuples (<literal>Eq</literal>,
156 <literal>Ord</literal>, <literal>Bounded</literal>,
157 <literal>Ix</literal> <literal>Read</literal>, and
158 <literal>Show</literal>) are available
159 <emphasis>only</emphasis> up to 16-tuples.</para>
161 <para>This limitation is easily subvertible, so please ask
162 if you get stuck on it.</para>
167 <term><literal>Read</literal>ing integers</term>
169 <para>GHC's implementation of the
170 <literal>Read</literal> class for integral types accepts
171 hexadecimal and octal literals (the code in the Haskell
172 98 report doesn't). So, for example,
173 <programlisting>read "0xf00" :: Int</programlisting>
175 <para>A possible reason for this is that <literal>readLitChar</literal> accepts hex and
176 octal escapes, so it seems inconsistent not to do so for integers too.</para>
181 <term><literal>isAlpha</literal></term>
183 <para>The Haskell 98 definition of <literal>isAlpha</literal>
186 <programlisting>isAlpha c = isUpper c || isLower c</programlisting>
188 <para>GHC's implementation diverges from the Haskell 98
189 definition in the sense that Unicode alphabetic characters which
190 are neither upper nor lower case will still be identified as
191 alphabetic by <literal>isAlpha</literal>.</para>
196 <term><literal>hGetContents</literal></term>
199 Lazy I/O throws an exception if an error is
200 encountered, in contrast to the Haskell 98 spec which
201 requires that errors are discarded (see Section 21.2.2
202 of the Haskell 98 report). The exception thrown is
203 the usual IO exception that would be thrown if the
204 failing IO operation was performed in the IO monad, and can
205 be caught by <literal>System.IO.Error.catch</literal>
206 or <literal>Control.Exception.catch</literal>.
214 <sect2 id="haskell98-undefined">
215 <title>GHC's interpretation of undefined behaviour in
216 Haskell 98</title>
218 <para>This section documents GHC's take on various issues that are
219 left undefined or implementation specific in Haskell 98.</para>
224 The <literal>Char</literal> type
225 <indexterm><primary><literal>Char</literal></primary><secondary>size of</secondary></indexterm>
228 <para>Following the ISO-10646 standard,
229 <literal>maxBound :: Char</literal> in GHC is
230 <literal>0x10FFFF</literal>.</para>
237 <indexterm><primary><literal>Int</literal></primary><secondary>size of</secondary></indexterm>
240 <para>In GHC the <literal>Int</literal> type follows the
241 size of an address on the host architecture; in other words
242 it holds 32 bits on a 32-bit machine, and 64-bits on a
243 64-bit machine.</para>
245 <para>Arithmetic on <literal>Int</literal> is unchecked for
246 overflow<indexterm><primary>overflow</primary><secondary><literal>Int</literal></secondary>
247 </indexterm>, so all operations on <literal>Int</literal> happen
249 2<superscript><replaceable>n</replaceable></superscript>
250 where <replaceable>n</replaceable> is the size in bits of
251 the <literal>Int</literal> type.</para>
253 <para>The <literal>fromInteger</literal><indexterm><primary><literal>fromInteger</literal></primary>
254 </indexterm>function (and hence
255 also <literal>fromIntegral</literal><indexterm><primary><literal>fromIntegral</literal></primary>
256 </indexterm>) is a special case when
257 converting to <literal>Int</literal>. The value of
258 <literal>fromIntegral x :: Int</literal> is given by taking
259 the lower <replaceable>n</replaceable> bits of <literal>(abs
260 x)</literal>, multiplied by the sign of <literal>x</literal>
261 (in 2's complement <replaceable>n</replaceable>-bit
262 arithmetic). This behaviour was chosen so that for example
263 writing <literal>0xffffffff :: Int</literal> preserves the
264 bit-pattern in the resulting <literal>Int</literal>.</para>
267 <para>Negative literals, such as <literal>-3</literal>, are
268 specified by (a careful reading of) the Haskell Report as
269 meaning <literal>Prelude.negate (Prelude.fromInteger 3)</literal>.
270 So <literal>-2147483648</literal> means <literal>negate (fromInteger 2147483648)</literal>.
271 Since <literal>fromInteger</literal> takes the lower 32 bits of the representation,
272 <literal>fromInteger (2147483648::Integer)</literal>, computed at type <literal>Int</literal> is
273 <literal>-2147483648::Int</literal>. The <literal>negate</literal> operation then
274 overflows, but it is unchecked, so <literal>negate (-2147483648::Int)</literal> is just
275 <literal>-2147483648</literal>. In short, one can write <literal>minBound::Int</literal> as
276 a literal with the expected meaning (but that is not in general guaranteed.
279 <para>The <literal>fromIntegral</literal> function also
280 preserves bit-patterns when converting between the sized
281 integral types (<literal>Int8</literal>,
282 <literal>Int16</literal>, <literal>Int32</literal>,
283 <literal>Int64</literal> and the unsigned
284 <literal>Word</literal> variants), see the modules
285 <literal>Data.Int</literal> and <literal>Data.Word</literal>
286 in the library documentation.</para>
291 <term>Unchecked float arithmetic</term>
293 <para>Operations on <literal>Float</literal> and
294 <literal>Double</literal> numbers are
295 <emphasis>unchecked</emphasis> for overflow, underflow, and
296 other sad occurrences. (note, however that some
297 architectures trap floating-point overflow and
298 loss-of-precision and report a floating-point exception,
299 probably terminating the
300 program)<indexterm><primary>floating-point
301 exceptions</primary></indexterm>.</para>
308 <sect2 id="ffi-divergence">
309 <title>Divergence from the FFI specification</title>
313 <term><literal>hs_init()</literal> not allowed
314 after <literal>hs_exit()</literal></term>
316 <para>The FFI spec requires the implementation to support
317 re-initialising itself after being shut down
318 with <literal>hs_exit()</literal>, but GHC does not
319 currently support that.</para>
329 <title>Known bugs or infelicities</title>
331 <para>The bug tracker lists bugs that have been reported in GHC but not
332 yet fixed: see the <ulink url="http://sourceforge.net/projects/ghc/">SourceForge GHC
333 page</ulink>. In addition to those, GHC also has the following known bugs
334 or infelicities. These bugs are more permanent; it is unlikely that
335 any of them will be fixed in the short term.</para>
337 <sect2 id="bugs-ghc">
338 <title>Bugs in GHC</title>
342 <para> GHC can warn about non-exhaustive or overlapping
343 patterns (see <xref linkend="options-sanity"/>), and usually
344 does so correctly. But not always. It gets confused by
345 string patterns, and by guards, and can then emit bogus
346 warnings. The entire overlap-check code needs an overhaul
351 <para>GHC does not allow you to have a data type with a context
352 that mentions type variables that are not data type parameters.
355 data C a b => T a = MkT a
357 so that <literal>MkT</literal>'s type is
359 MkT :: forall a b. C a b => a -> T a
361 In principle, with a suitable class declaration with a functional dependency,
362 it's possible that this type is not ambiguous; but GHC nevertheless rejects
363 it. The type variables mentioned in the context of the data type declaration must
364 be among the type parameters of the data type.</para>
368 <para>GHC's inliner can be persuaded into non-termination
369 using the standard way to encode recursion via a data type:</para>
371 data U = MkU (U -> Bool)
374 russel u@(MkU p) = not $ p u
377 x = russel (MkU russel)
380 <para>We have never found another class of programs, other
381 than this contrived one, that makes GHC diverge, and fixing
382 the problem would impose an extra overhead on every
383 compilation. So the bug remains un-fixed. There is more
385 url="http://research.microsoft.com/~simonpj/Papers/inlining/">
386 Secrets of the GHC inliner</ulink>.</para>
390 <para>GHC does not keep careful track of
391 what instance declarations are 'in scope' if they come from other packages.
392 Instead, all instance declarations that GHC has seen in other
393 packages are all in scope everywhere, whether or not the
394 module from that package is used by the command-line
395 expression. This bug affects only the <option>--make</option> mode and
402 <sect2 id="bugs-ghci">
403 <title>Bugs in GHCi (the interactive GHC)</title>
406 <para>GHCi does not respect the <literal>default</literal>
407 declaration in the module whose scope you are in. Instead,
408 for expressions typed at the command line, you always get the
409 default default-type behaviour; that is,
410 <literal>default(Int,Double)</literal>.</para>
412 <para>It would be better for GHCi to record what the default
413 settings in each module are, and use those of the 'current'
414 module (whatever that is).</para>
418 <para>On Windows, there's a GNU ld/BFD bug
419 whereby it emits bogus PE object files that have more than
420 0xffff relocations. When GHCi tries to load a package affected by this
421 bug, you get an error message of the form
423 Loading package javavm ... linking ... WARNING: Overflown relocation field (# relocs found: 30765)
425 The last time we looked, this bug still
426 wasn't fixed in the BFD codebase, and there wasn't any
427 noticeable interest in fixing it when we reported the bug
430 <para>The workaround is to split up the .o files that make up
431 your package into two or more .o's, along the lines of
432 how the "base" package does it.</para>
441 ;;; Local Variables: ***
443 ;;; sgml-parent-document: ("users_guide.xml" "book" "chapter") ***