gcc/LANGUAGES

   1 Right now there is no documentation for the GCC tree -> rtl interfaces
   2 (or more generally the interfaces for adding new languages).
   3
   4 Such documentation would be of great benefit to the project.  Until such
   5 time as we can formally start documenting the interface this file will
   6 serve as a repository for information on these interface and any incompatable
   7 changes we've made.
   8
   9 2001-02-26:
  10   A DECL_INITIAL of NULL_TREE or error_mark_node in a VAR_DECL is no longer
  11   taken to signify a tentative definition which should not be emitted until
  12   end-of-file.  Frontends which want that behavior should set
  13   DECL_DEFER_OUTPUT before calling rest_of_decl_compilation.
  14
  15 Aug 31, 1998:
  16   The interface to HANDLE_PRAGMA has changed.  It now takes three arguments.
  17   The first two are pointers to functions that should be used to read characters
  18   from the input stream, and to push them back into the input stream respectively.
  19   The third argument is a pointer to a null terminate string which is the first
  20   word after #pragma.  The expression supplied by HANDLE_PRAGMA should return
  21   nonzero if it parsed and implemented the pragma.  Otherwise it should return
  22   zero, and leave the input stream as it was before the expression was evaluated.
  23
  24   A new back-end definable macro has been added: INSERT_ATTRIBUTES.  This macro
  25   allows backend to add attributes to decls as they are created.
  26
  27 Jun 10, 1998:
  28   The interface to lang_decode_option has changed. It now uses and argc/argv
  29   interface to allow for options that use more than one input string. The new
  30   declaration is: int lang_decode_option (int argc, char** argv). It now
  31   returns the number of input strings processed, or 0 if the option is
  32   unknown.
  33
  34 Jun  7, 1998:
  35   Front-ends must now define lang_init_options.  It is safe for this
  36   function to do nothing.  See c-lang.c.
  37
  38 Apr 21, 1998:
  39   Front ends which link with c-common or other files from the C/C++
  40   front-ends may need to handle TI types.  Look for references to
  41   [unsigned]int_DI_type_node in your front end.  If you have references
  42   to these variables, you'll need up update the front end.
  43
  44   To update the front end you must mirror all the code which currently
  45   deals with intDI_type_node to also handle intTI_type_node.
  46
  47
  48 Apr 7, 1998:
  49   The interface between toplev.c and the language front ends for opening the
  50   source file has changed:
  51
  52   o init_lex() has been renamed to init_parse (char *filename) where filename
  53     is the name of the source file.
  54   o The code in toplev.c which opened the source file should be moved to
  55     the new init_parse function.
  56   o toplev.c now calls finish_parse() instead of closing the source file
  57     using fclose(). This should now be done in finish_parse, if necessary.
  58
  59 Apr 1, 1998:
  60   Front-ends must now define lang_print_xnode.  It is safe for this
  61   function to do nothing.  See c-lang.c.
  62
  63 Feb 1, 1998:
  64
  65   GCC used to store structure sizes & offsets to elements as bitsize
  66   quantities.  This causes problems because a structure can only be
  67   (target memsize / 8) bytes long (this may effect arrays too).  This
  68   is particularly problematical on machines with small address spaces.
  69
  70   So:
  71
  72     All trees that represent sizes in bits should have a TREE_TYPE of
  73     bitsizetype (rather than sizetype).
  74
  75     Accordingly, when such values are computed / initialized, care has to
  76     be takes to use / compute the proper type.
  77
  78     When a size in bits is converted into a size in bytes, which is expressed
  79     in trees, care should be taken to change the tree's type again to sizetype.
  80
  81     We've updated C, C++, Fortran & Objective-C to work with the new
  82     scheme.  Other languages will need to be updated accordingly.
  83     Contact amylaar@cygnus.com for additional information.
  84
  85 ?? 1997:
  86
  87   In an effort to decrease cache thrashing and useless loads we've changed the
  88   third argument to the DEFTREECODE macro to be a single char.  This will
  89   affect languages that defined their own tree codes (usually in a .def file).
  90
  91   Old way:
  92
  93     DEFTREECODE (CLASS_METHOD_DECL, "class_method_decl", "d", 0)
  94
  95   New way:
  96
  97     DEFTREECODE (CLASS_METHOD_DECL, "class_method_decl", 'd', 0)