[ghc-hetmet.git] / docs / users_guide / bugs.xml

<?xml version="1.0" encoding="iso-8859-1"?>
<chapter id="bugs-and-infelicities">
  <title>Known bugs and infelicities</title>

  <sect1 id="vs-Haskell-defn">
    <title>Haskell&nbsp;98 vs.&nbsp;Glasgow Haskell: language non-compliance
</title>
    
    <indexterm><primary>GHC vs the Haskell 98 language</primary></indexterm>
    <indexterm><primary>Haskell 98 language vs GHC</primary></indexterm>

  <para>This section lists Glasgow Haskell infelicities in its
  implementation of Haskell&nbsp;98.  See also the &ldquo;when things
  go wrong&rdquo; section (<xref linkend="wrong"/>) for information
  about crashes, space leaks, and other undesirable phenomena.</para>

  <para>The limitations here are listed in Haskell Report order
  (roughly).</para>

  <sect2 id="haskell98-divergence">
    <title>Divergence from Haskell&nbsp;98</title>
    
      
    <sect3 id="infelicities-lexical">
      <title>Lexical syntax</title>
      
      <itemizedlist>
        <listitem>
          <para>Certain lexical rules regarding qualified identifiers
          are slightly different in GHC compared to the Haskell
          report.  When you have
          <replaceable>module</replaceable><literal>.</literal><replaceable>reservedop</replaceable>,
          such as <literal>M.\</literal>, GHC will interpret it as a
          single qualified operator rather than the two lexemes
          <literal>M</literal> and <literal>.\</literal>.</para>
        </listitem>
      </itemizedlist>
    </sect3>
      
      <sect3 id="infelicities-syntax">
        <title>Context-free syntax</title>
        
        <itemizedlist>
          <listitem>
            <para>GHC is a little less strict about the layout rule when used
              in <literal>do</literal> expressions.  Specifically, the
              restriction that "a nested context must be indented further to
              the right than the enclosing context" is relaxed to allow the
              nested context to be at the same level as the enclosing context,
              if the enclosing context is a <literal>do</literal>
              expression.</para>

            <para>For example, the following code is accepted by GHC:

<programlisting>
main = do args &lt;- getArgs
          if null args then return [] else do
          ps &lt;- mapM process args
          mapM print ps</programlisting>

              </para>
          </listitem>

        <listitem>
          <para>GHC doesn't do fixity resolution in expressions during
          parsing.  For example, according to the Haskell report, the
          following expression is legal Haskell:
<programlisting>
    let x = 42 in x == 42 == True</programlisting>
        and parses as:
<programlisting>
    (let x = 42 in x == 42) == True</programlisting>

          because according to the report, the <literal>let</literal>
          expression <quote>extends as far to the right as
          possible</quote>.  Since it can't extend past the second
          equals sign without causing a parse error
          (<literal>==</literal> is non-fix), the
          <literal>let</literal>-expression must terminate there.  GHC
          simply gobbles up the whole expression, parsing like this:
<programlisting>
    (let x = 42 in x == 42 == True)</programlisting>

          The Haskell report is arguably wrong here, but nevertheless
          it's a difference between GHC &amp; Haskell 98.</para>
        </listitem>
      </itemizedlist>
    </sect3>

  <sect3 id="infelicities-exprs-pats">
      <title>Expressions and patterns</title>

        <para>None known.</para>
    </sect3>

    <sect3 id="infelicities-decls">
      <title>Declarations and bindings</title>

      <para>GHC's typechecker makes all pattern bindings monomorphic
      by default; this behaviour can be disabled with
      <option>-XNoMonoPatBinds</option>.  See <xref
      linkend="options-language" />.</para>
    </sect3>
      
      <sect3 id="infelicities-Modules">
        <title>Module system and interface files</title>
        
        <para>GHC requires the use of <literal>hs-boot</literal>
          files to cut the recursive loops among mutually recursive modules
          as described in <xref linkend="mutual-recursion"/>.  This more of an infelicity
            than a bug: the Haskell Report says 
          (<ulink url="http://haskell.org/onlinereport/modules.html#sect5.7">Section 5.7</ulink>) "Depending on the Haskell
        implementation used, separate compilation of mutually
        recursive modules may require that imported modules contain
        additional information so that they may be referenced before
        they are compiled. Explicit type signatures for all exported
        values may be necessary to deal with mutual recursion. The
        precise details of separate compilation are not defined by
        this Report."

        </para>

    </sect3>

    <sect3 id="infelicities-numbers">
      <title>Numbers, basic types, and built-in classes</title>

      <variablelist>
        <varlistentry>
          <term>Multiply-defined array elements&mdash;not checked:</term>
          <listitem>
            <para>This code fragment should
            elicit a fatal error, but it does not:

<programlisting>
main = print (array (1,1) [(1,2), (1,3)])</programlisting>
GHC's implementation of <literal>array</literal> takes the value of an
array slot from the last (index,value) pair in the list, and does no
checking for duplicates.  The reason for this is efficiency, pure and simple.
            </para>
          </listitem>
        </varlistentry>
      </variablelist>
      
    </sect3>

      <sect3 id="infelicities-Prelude">
        <title>In <literal>Prelude</literal> support</title>

      <variablelist>
        <varlistentry>
          <term>Arbitrary-sized tuples</term>
          <listitem>
            <para>Tuples are currently limited to size 100.  HOWEVER:
            standard instances for tuples (<literal>Eq</literal>,
            <literal>Ord</literal>, <literal>Bounded</literal>,
            <literal>Ix</literal> <literal>Read</literal>, and
            <literal>Show</literal>) are available
            <emphasis>only</emphasis> up to 16-tuples.</para>

            <para>This limitation is easily subvertible, so please ask
            if you get stuck on it.</para>
            </listitem>
          </varlistentry>

          <varlistentry>
            <term><literal>Read</literal>ing integers</term>
            <listitem>
              <para>GHC's implementation of the
              <literal>Read</literal> class for integral types accepts
              hexadecimal and octal literals (the code in the Haskell
              98 report doesn't).  So, for example,
<programlisting>read "0xf00" :: Int</programlisting>
              works in GHC.</para>
              <para>A possible reason for this is that <literal>readLitChar</literal> accepts hex and
                octal escapes, so it seems inconsistent not to do so for integers too.</para>
            </listitem>
          </varlistentry>

          <varlistentry>
            <term><literal>isAlpha</literal></term>
            <listitem>
              <para>The Haskell 98 definition of <literal>isAlpha</literal>
              is:</para>

<programlisting>isAlpha c = isUpper c || isLower c</programlisting>

              <para>GHC's implementation diverges from the Haskell 98
              definition in the sense that Unicode alphabetic characters which
              are neither upper nor lower case will still be identified as
              alphabetic by <literal>isAlpha</literal>.</para>
            </listitem>
          </varlistentry>

          <varlistentry>
            <term>Strings treated as ISO-8859-1</term>
            <listitem>
              <para>
            Various library functions, such as <literal>putStrLn</literal>,
            treat Strings as if they were ISO-8859-1 rather than UTF-8.
          </para>
            </listitem>
          </varlistentry>
        </variablelist>
    </sect3>
  </sect2>

  <sect2 id="haskell98-undefined">
    <title>GHC's interpretation of undefined behaviour in
    Haskell&nbsp;98</title>

    <para>This section documents GHC's take on various issues that are
    left undefined or implementation specific in Haskell 98.</para>

    <variablelist>
      <varlistentry>
        <term>
          The <literal>Char</literal> type
          <indexterm><primary><literal>Char</literal></primary><secondary>size of</secondary></indexterm>
        </term>
        <listitem>
          <para>Following the ISO-10646 standard,
          <literal>maxBound :: Char</literal> in GHC is
          <literal>0x10FFFF</literal>.</para>
        </listitem>
      </varlistentry>

      <varlistentry>
        <term>
          Sized integral types
          <indexterm><primary><literal>Int</literal></primary><secondary>size of</secondary></indexterm>
        </term>
        <listitem>
          <para>In GHC the <literal>Int</literal> type follows the
          size of an address on the host architecture; in other words
          it holds 32 bits on a 32-bit machine, and 64-bits on a
          64-bit machine.</para>

          <para>Arithmetic on <literal>Int</literal> is unchecked for
          overflow<indexterm><primary>overflow</primary><secondary><literal>Int</literal></secondary>
            </indexterm>, so all operations on <literal>Int</literal> happen
          modulo
          2<superscript><replaceable>n</replaceable></superscript>
          where <replaceable>n</replaceable> is the size in bits of
          the <literal>Int</literal> type.</para>

          <para>The <literal>fromInteger</literal><indexterm><primary><literal>fromInteger</literal></primary>
            </indexterm>function (and hence
          also <literal>fromIntegral</literal><indexterm><primary><literal>fromIntegral</literal></primary>
            </indexterm>) is a special case when
          converting to <literal>Int</literal>.  The value of
          <literal>fromIntegral x :: Int</literal> is given by taking
          the lower <replaceable>n</replaceable> bits of <literal>(abs
          x)</literal>, multiplied by the sign of <literal>x</literal>
          (in 2's complement <replaceable>n</replaceable>-bit
          arithmetic).  This behaviour was chosen so that for example
          writing <literal>0xffffffff :: Int</literal> preserves the
          bit-pattern in the resulting <literal>Int</literal>.</para>


           <para>Negative literals, such as <literal>-3</literal>, are
             specified by (a careful reading of) the Haskell Report as 
             meaning <literal>Prelude.negate (Prelude.fromInteger 3)</literal>.
             So <literal>-2147483648</literal> means <literal>negate (fromInteger 2147483648)</literal>.
             Since <literal>fromInteger</literal> takes the lower 32 bits of the representation,
             <literal>fromInteger (2147483648::Integer)</literal>, computed at type <literal>Int</literal> is
             <literal>-2147483648::Int</literal>.  The <literal>negate</literal> operation then
             overflows, but it is unchecked, so <literal>negate (-2147483648::Int)</literal> is just
             <literal>-2147483648</literal>.  In short, one can write <literal>minBound::Int</literal> as
             a literal with the expected meaning (but that is not in general guaranteed.
             </para>

          <para>The <literal>fromIntegral</literal> function also
          preserves bit-patterns when converting between the sized
          integral types (<literal>Int8</literal>,
          <literal>Int16</literal>, <literal>Int32</literal>,
          <literal>Int64</literal> and the unsigned
          <literal>Word</literal> variants), see the modules
          <literal>Data.Int</literal> and <literal>Data.Word</literal>
          in the library documentation.</para>
        </listitem>
      </varlistentry>

      <varlistentry>
        <term>Unchecked float arithmetic</term>
        <listitem>
          <para>Operations on <literal>Float</literal> and
          <literal>Double</literal> numbers are
          <emphasis>unchecked</emphasis> for overflow, underflow, and
          other sad occurrences.  (note, however that some
          architectures trap floating-point overflow and
          loss-of-precision and report a floating-point exception,
          probably terminating the
          program)<indexterm><primary>floating-point
          exceptions</primary></indexterm>.</para>
        </listitem>
      </varlistentry>
    </variablelist>
      
    </sect2>

  <sect2 id="ffi-divergence">
    <title>Divergence from the FFI specification</title>
    
    <variablelist>
      <varlistentry>
        <term><literal>hs_init()</literal> not allowed
        after <literal>hs_exit()</literal></term>
        <listitem>
          <para>The FFI spec requires the implementation to support
            re-initialising itself after being shut down
            with <literal>hs_exit()</literal>, but GHC does not
            currently support that.</para>
        </listitem>
      </varlistentry>
    </variablelist>
  </sect2>
    
  </sect1>


  <sect1 id="bugs">
    <title>Known bugs or infelicities</title>

    <para>The bug tracker lists bugs that have been reported in GHC but not
      yet fixed: see the <ulink url="http://sourceforge.net/projects/ghc/">SourceForge GHC
    page</ulink>.  In addition to those, GHC also has the following known bugs
      or  infelicities.  These bugs are more permanent; it is unlikely that
      any of them will be fixed in the short term.</para>

  <sect2 id="bugs-ghc">
    <title>Bugs in GHC</title>

    <itemizedlist>
      <listitem>
        <para> GHC can warn about non-exhaustive or overlapping
        patterns (see <xref linkend="options-sanity"/>), and usually
        does so correctly.  But not always.  It gets confused by
        string patterns, and by guards, and can then emit bogus
        warnings.  The entire overlap-check code needs an overhaul
        really.</para>
      </listitem>

      <listitem>
        <para>GHC does not allow you to have a data type with a context 
           that mentions type variables that are not data type parameters.
          For example:
<programlisting>
  data C a b => T a = MkT a
</programlisting>
          so that <literal>MkT</literal>'s type is
<programlisting>
  MkT :: forall a b. C a b => a -> T a
</programlisting>
        In principle, with a suitable class declaration with a functional dependency,
         it's possible that this type is not ambiguous; but GHC nevertheless rejects
          it.  The type variables mentioned in the context of the data type declaration must
        be among the type parameters of the data type.</para>
      </listitem>

      <listitem>
        <para>GHC's inliner can be persuaded into non-termination
        using the standard way to encode recursion via a data type:</para>
<programlisting>
  data U = MkU (U -> Bool)
       
  russel :: U -> Bool
  russel u@(MkU p) = not $ p u
  
  x :: Bool
  x = russel (MkU russel)
</programlisting>

        <para>We have never found another class of programs, other
        than this contrived one, that makes GHC diverge, and fixing
        the problem would impose an extra overhead on every
        compilation.  So the bug remains un-fixed.  There is more
        background in <ulink
        url="http://research.microsoft.com/~simonpj/Papers/inlining/">
        Secrets of the GHC inliner</ulink>.</para>
      </listitem>

      <listitem>
        <para>GHC does not keep careful track of
            what instance declarations are 'in scope' if they come from other packages.
        Instead, all instance declarations that GHC has seen in other
        packages are all in scope everywhere, whether or not the
        module from that package is used by the command-line
        expression.  This bug affects only the <option>--make</option> mode and
          GHCi.</para>
      </listitem>

    </itemizedlist>
  </sect2>

  <sect2 id="bugs-ghci">
    <title>Bugs in GHCi (the interactive GHC)</title>
    <itemizedlist>
      <listitem>
        <para>GHCi does not respect the <literal>default</literal>
        declaration in the module whose scope you are in.  Instead,
        for expressions typed at the command line, you always get the
        default default-type behaviour; that is,
        <literal>default(Int,Double)</literal>.</para>

        <para>It would be better for GHCi to record what the default
        settings in each module are, and use those of the 'current'
        module (whatever that is).</para>
      </listitem>

      <listitem> 
      <para>On Windows, there's a GNU ld/BFD bug
      whereby it emits bogus PE object files that have more than
      0xffff relocations. When GHCi tries to load a package affected by this
      bug, you get an error message of the form
<screen>
Loading package javavm ... linking ... WARNING: Overflown relocation field (# relocs found: 30765)
</screen>
      The last time we looked, this bug still
      wasn't fixed in the BFD codebase, and there wasn't any
      noticeable interest in fixing it when we reported the bug
      back in 2001 or so.
      </para>
      <para>The workaround is to split up the .o files that make up
      your package into two or more .o's, along the lines of
      how the "base" package does it.</para>
      </listitem>
    </itemizedlist>
  </sect2>
  </sect1>

</chapter>

<!-- Emacs stuff:
     ;;; Local Variables: ***
     ;;; mode: xml ***
     ;;; sgml-parent-document: ("users_guide.xml" "book" "chapter") ***
     ;;; End: ***
 -->