ghc/docs/users_guide/vs_haskell.lit

   1 %************************************************************************
   2 %*                                                                      *
   3 \section[vs-Haskell-defn]{Haskell~1.3 vs.~Glasgow Haskell~2.01: language non-compliance}
   4 \index{GHC vs the Haskell 1.3 language}
   5 \index{Haskell 1.3 language vs GHC}
   6 %*                                                                      *
   7 %************************************************************************
   8
   9 This section lists Glasgow Haskell infelicities in its implementation
  10 of Haskell~1.3.  See also the ``when things go wrong'' section
  11 (\sectionref{wrong}) for information about crashes, space leaks, and
  12 other undesirable phenomena.
  13
  14 GHC~2.01 does not have a ``Haskell~1.2 compatibility mode.''  If
  15 enough people yelled loudly and long enough, it could conceivably
  16 happen...
  17
  18 The limitations here are listed in Haskell-Report order (roughly).
  19 %Limitations related to Glasgow extensions (unboxed numbers, etc.) are
  20 %given thereafter (\sectionref{infelicities-Glasgow-exts}).
  21
  22 %************************************************************************
  23 %*                                                                      *
  24 \subsection[infelicities-exprs-pats]{Expressions and patterns}
  25 %*                                                                      *
  26 %************************************************************************
  27
  28 \begin{description}
  29 %-------------------------------------------------------------------
  30 \item[No @n+k@ patterns:]
  31 We took them out of GHC, because we thought they weren't going to
  32 survive into Haskell~1.3.  But they did.  And we haven't put them
  33 back.  Yet.
  34
  35 %-------------------------------------------------------------------
  36 \item[Very long @String@ constants:]
  37 May not go through.  If you add a ``string gap'' every
  38 few thousand characters, then the strings can be as long
  39 as you like.
  40
  41 Bear in mind that string gaps and the \tr{-cpp}\index{-cpp option}
  42 option don't mix.  The C-preprocessor may munch the backslashes.
  43
  44 %-------------------------------------------------------------------
  45 \item[Very long literal lists:]
  46 These may tickle a ``yacc stack overflow'' error in the parser.
  47 (It depends on the Yacc used to build your parser.)
  48
  49 %-------------------------------------------------------------------
  50 \item[Single quotes in module names:]
  51 It might work, but it's just begging for trouble.
  52 \end{description}
  53
  54 %************************************************************************
  55 %*                                                                      *
  56 \subsection[infelicities-decls]{Declarations and bindings}
  57 %*                                                                      *
  58 %************************************************************************
  59
  60 \begin{description}
  61 %-------------------------------------------------------------------
  62 \item[Derived instances of @Read@ and @Show@ for infix constructors:]
  63 All the carry-on about derived @readsPrec@ and @showsPrec@ for infix
  64 constructors---we don't do it (yet).  We treat them the same way as
  65 all other constructors.
  66
  67 %-------------------------------------------------------------------
  68 \item[Derived instances for records:] Hmmm.
  69 \end{description}
  70
  71 %************************************************************************
  72 %*                                                                      *
  73 \subsection[infelicities-Modules]{Module system and interface files}
  74 %*                                                                      *
  75 %************************************************************************
  76
  77 \begin{description}
  78 %-------------------------------------------------------------------
  79 \item[Can't export primitives types (e.g., \tr{Int#}):]
  80
  81 Don't even try...
  82 \end{description}
  83
  84 %************************************************************************
  85 %*                                                                      *
  86 \subsection[infelicities-numbers]{Numbers, basic types, and built-in classes}
  87 %*                                                                      *
  88 %************************************************************************
  89
  90 \begin{description}
  91 %-------------------------------------------------------------------
  92 \item[Very large/small fractional constants:]
  93 (i.e., with a decimal point somewhere) GHC does not check that these
  94 are out of range (e.g., for a @Float@), and bad things will inevitably
  95 follow.  (To be corrected?)
  96
  97 This problem does {\em not} exist for integral constants.
  98
  99 For very large/small fractional constants near the limits of your
 100 floating-point precision, things may go wrong.  (It's better than it
 101 used to be.)  Please report any such bugs.
 102
 103 %-------------------------------------------------------------------
 104 \item[Unchecked arithmetic:]
 105 Arguably {\em not} an infelicity, but... Bear in mind that operations
 106 on \tr{Int}, \tr{Float}, and \tr{Double} numbers are {\em unchecked}
 107 for overflow, underflow, and other sad occurrences.
 108
 109 Use \tr{Integer}, \tr{Rational}, etc., numeric types if this stuff
 110 keeps you awake at night.
 111
 112 %-------------------------------------------------------------------
 113 \item[Multiply-defined array elements---not checked:]
 114 This code fragment {\em should} elicit a fatal error, but it does not:
 115 \begin{verbatim}
 116 main = print (array (1,1) [ 1:=2, 1:=3 ])
 117 \end{verbatim}
 118 \end{description}
 119
 120 %************************************************************************
 121 %*                                                                      *
 122 \subsection[infelicities-Prelude]{In Prelude support}
 123 %*                                                                      *
 124 %************************************************************************
 125
 126 \begin{description}
 127 %-------------------------------------------------------------------
 128 \item[Polymorphic @seq@:]
 129 Required by the Haskell~1.3 prelude; not done yet.  Consequently,
 130 the @strict@ function doesn't really do what you want either.
 131
 132 %-------------------------------------------------------------------
 133 \item[Arbitrary-sized tuples:]
 134 Plain old tuples of arbitrary size {\em do} work.  Note that lots of
 135 overloading can give rise to large tuples ``under the hood'' of your
 136 program.
 137
 138 HOWEVER: standard instances for tuples (@Eq@, @Ord@, @Bounded@, @Ix@
 139 @Read@, and @Show@) are available {\em only} up to 5-tuples.
 140
 141 These limitations are easily subvertible, so please ask if you get
 142 stuck on them.
 143 \end{description}