view src/README @ 4407:4ee73bbe4f8e

Always use boyer_moore in ASCII or Latin-1 buffers with ASCII search strings. 2007-12-26 Aidan Kehoe <kehoea@parhasard.net> * casetab.c: Extend and correct some case table documentation. * search.c (search_buffer): Correct a bug where only the first entry for a character in the case equivalence table was examined in determining if the Boyer-Moore search algorithm is appropriate. If there are case mappings outside of the charset and row of the characters specified in the search string, those case mappings can be safely ignored (and Boyer-Moore search can be used) if we know from the buffer statistics that the corresponding characters cannot occur. * search.c (boyer_moore): Assert that we haven't been passed a string with varying characters sets or rows within character sets. That's what simple_search is for. In the very rare event that a character in the search string has a canonical case mapping that is not in the same character set and row, don't try to search for the canonical character, search for some other character that is in the the desired character set and row. Assert that the case table isn't corrupt. Do not search for any character case mappings that cannot possibly occur in the buffer, given the buffer metadata about its contents.
author Aidan Kehoe <kehoea@parhasard.net>
date Wed, 26 Dec 2007 17:30:16 +0100
parents abe6d1db359e
children 304aebb79cd3
line wrap: on
line source

This directory contains the source files for the C component of XEmacs.
Nothing in this directory is needed for using XEmacs once it is built
and installed, if the dumped Emacs is copied elsewhere.

See the files ../README and then ../INSTALL for installation instructions.

Under Unix, the file `Makefile.in.in' is used as a template by the script
`../configure' to produce `Makefile.in'.  The same script then uses `cpp'
to produce the machine-dependent `Makefile' from `Makefile.in';
`Makefile' is the file which actually controls the compilation of
Emacs.  Most of this should work transparently to the user; you should
only need to run `../configure', and then type `make'.

General changes for XEmacs:
---------------------------
1. Lisp objects.

   -- XFASTINT has been eliminated.  Use of this expression as an lvalue
      is incompatible with the union form of Lisp objects, and use as
      an rvalue is likely to lead to errors and doesn't really save much
      time.  Expressions of the form `XFASTINT (obj) = num;' get replaced
      by `obj = make_int (num);' or `XSETINT (obj, num);' and
      expressions of the form `num = XFASTINT (obj);' get replaced by
      `num = XINT (obj);'.  Use Qzero in place of `make_int (0)'.

   -- Use of XTYPE gets replaced by the appropriate predicate.  Using
      XTYPE only works for the small number of types that are not stored
      using the Lisp_Record type (int, cons, string, and vector).  For
      example, `(XTYPE (foo) == Lisp_Buffer)' gets replaced by
      `(BUFFERP (foo))'.

   -- `XSET (obj, Lisp_Int, num)' gets replaced by `XSETINT (obj, num)',
      for consistency.

   -- Some occurrences of XSET need to get replaced by XSETR --
      specifically, those where the type is not a primitive type
      (primitive types are int, cons, string, and vector).

   -- References to `XSTRING (obj)->size' get replaced with
      `XSTRING_LENGTH (obj)'.  This is currently for cosmetic reasons
      but there may be other reasons in the future.  (This change is
      currently incomplete in the source files.)


2. Storage classes:

   -- All occurrences of `register' should be replaced by `REGISTER'.
      It interferes with backtraces so we disable it if DEBUG_XEMACS
      is defined.


3. Errors, messages, I18N3 snarfing:

   -- Errors are continuable in XEmacs but are not in FSF Emacs.
      Therefore, it's important that functions do something reasonable
      if an error gets continued.  If you want to signal a non-
      continuable error, the call to Fsignal() gets put inside a
      `while (1)' loop.  To facilitate this, and also for proper I18N3
      message snarfing, most calls to Fsignal() have been replaced by
      calls to signal_error(), signal_simple_error(), etc.  Look at
      eval.c for a classification of various error functions.

   -- Constant strings occurring in source files need to get wrapped
      in a call to GETTEXT (or if inside of a call to `build_string',
      change that function to `build_translated_string') if they don't
      occur in certain places where the I18N3 message snarfer will see
      them.  For a complete discussion of this, see the file
      lib-src/make-msgfile.lex.

      NOTE: I18N3 support is not currently working, so the above may
      or may not apply.  Thus it is not a good idea to add random
      GETTEXTs, unless you really know what you are doing.

   -- Calls to `fprintf (stderr, ...)' and `printf (...)' get replaced
      with calls to `stderr_out' and `stdout_out'.  This is for I18N3
      message snarfing.

4. Initialization:

   -- FSF constructs like `obj = intern ("string"); staticpro (&obj);'
      get replaced by `defsymbol (&obj);'.  This is for code cleanness
      and better purespace usage.
   -- FSF constructs like
        obj = intern ("error");
        Fput (obj, Qerror_message, "message");
	Fput (obj, Qerror_conditions, some list);
      get replaced by calls to deferror().  See the definition of
      deferror() for how the correct arguments to pass.  This is for
      code cleanness and I18N3 message snarfing.
   -- Code in keys_of_foo() functions has been moved into Lisp.