[xemacs-hg @ 2001-09-20 06:28:42 by ben] The great integral types renaming. The purpose of this is to rationalize the names used for various integral types, so that they match their intended uses and follow consist conventions, and eliminate types that were not semantically different from each other. The conventions are: -- All integral types that measure quantities of anything are signed. Some people disagree vociferously with this, but their arguments are mostly theoretical, and are vastly outweighed by the practical headaches of mixing signed and unsigned values, and more importantly by the far increased likelihood of inadvertent bugs: Because of the broken "viral" nature of unsigned quantities in C (operations involving mixed signed/unsigned are done unsigned, when exactly the opposite is nearly always wanted), even a single error in declaring a quantity unsigned that should be signed, or even the even more subtle error of comparing signed and unsigned values and forgetting the necessary cast, can be catastrophic, as comparisons will yield wrong results. -Wsign-compare is turned on specifically to catch this, but this tends to result in a great number of warnings when mixing signed and unsigned, and the casts are annoying. More has been written on this elsewhere. -- All such quantity types just mentioned boil down to EMACS_INT, which is 32 bits on 32-bit machines and 64 bits on 64-bit machines. This is guaranteed to be the same size as Lisp objects of type `int', and (as far as I can tell) of size_t (unsigned!) and ssize_t. The only type below that is not an EMACS_INT is Hashcode, which is an unsigned value of the same size as EMACS_INT. -- Type names should be relatively short (no more than 10 characters or so), with the first letter capitalized and no underscores if they can at all be avoided. -- "count" == a zero-based measurement of some quantity. Includes sizes, offsets, and indexes. -- "bpos" == a one-based measurement of a position in a buffer. "Charbpos" and "Bytebpos" count text in the buffer, rather than bytes in memory; thus Bytebpos does not directly correspond to the memory representation. Use "Membpos" for this. -- "Char" refers to internal-format characters, not to the C type "char", which is really a byte. -- For the actual name changes, see the script below. I ran the following script to do the conversion. (NOTE: This script is idempotent. You can safely run it multiple times and it will not screw up previous results -- in fact, it will do nothing if nothing has changed. Thus, it can be run repeatedly as necessary to handle patches coming in from old workspaces, or old branches.) There are two tags, just before and just after the change: `pre-integral-type-rename' and `post-integral-type-rename'. When merging code from the main trunk into a branch, the best thing to do is first merge up to `pre-integral-type-rename', then apply the script and associated changes, then merge from `post-integral-type-change' to the present. (Alternatively, just do the merging in one operation; but you may then have a lot of conflicts needing to be resolved by hand.) Script `fixtypes.sh' follows: ----------------------------------- cut ------------------------------------ files="*.[ch] s/*.h m/*.h config.h.in ../configure.in Makefile.in.in ../lib-src/*.[ch] ../lwlib/*.[ch]" gr Memory_Count Bytecount $files gr Lstream_Data_Count Bytecount $files gr Element_Count Elemcount $files gr Hash_Code Hashcode $files gr extcount bytecount $files gr bufpos charbpos $files gr bytind bytebpos $files gr memind membpos $files gr bufbyte intbyte $files gr Extcount Bytecount $files gr Bufpos Charbpos $files gr Bytind Bytebpos $files gr Memind Membpos $files gr Bufbyte Intbyte $files gr EXTCOUNT BYTECOUNT $files gr BUFPOS CHARBPOS $files gr BYTIND BYTEBPOS $files gr MEMIND MEMBPOS $files gr BUFBYTE INTBYTE $files gr MEMORY_COUNT BYTECOUNT $files gr LSTREAM_DATA_COUNT BYTECOUNT $files gr ELEMENT_COUNT ELEMCOUNT $files gr HASH_CODE HASHCODE $files ----------------------------------- cut ------------------------------------ `fixtypes.sh' is a Bourne-shell script; it uses 'gr': ----------------------------------- cut ------------------------------------ #!/bin/sh # Usage is like this: # gr FROM TO FILES ... # globally replace FROM with TO in FILES. FROM and TO are regular expressions. # backup files are stored in the `backup' directory. from="$1" to="$2" shift 2 echo ${1+"$@"} | xargs global-replace "s/$from/$to/g" ----------------------------------- cut ------------------------------------ `gr' in turn uses a Perl script to do its real work, `global-replace', which follows: ----------------------------------- cut ------------------------------------ : #-*- Perl -*- ### global-modify --- modify the contents of a file by a Perl expression ## Copyright (C) 1999 Martin Buchholz. ## Copyright (C) 2001 Ben Wing. ## Authors: Martin Buchholz <martin@xemacs.org>, Ben Wing <ben@xemacs.org> ## Maintainer: Ben Wing <ben@xemacs.org> ## Current Version: 1.0, May 5, 2001 # This program is free software; you can redistribute it and/or modify # it under the terms of the GNU General Public License as published by # the Free Software Foundation; either version 2, or (at your option) # any later version. # # This program is distributed in the hope that it will be useful, but # WITHOUT ANY WARRANTY; without even the implied warranty of # MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU # General Public License for more details. # # You should have received a copy of the GNU General Public License # along with XEmacs; see the file COPYING. If not, write to the Free # Software Foundation, Inc., 59 Temple Place - Suite 330, Boston, MA # 02111-1307, USA. eval 'exec perl -w -S $0 ${1+"$@"}' if 0; use strict; use FileHandle; use Carp; use Getopt::Long; use File::Basename; (my $myName = $0) =~ s@.*/@@; my $usage=" Usage: $myName [--help] [--backup-dir=DIR] [--line-mode] [--hunk-mode] PERLEXPR FILE ... Globally modify a file, either line by line or in one big hunk. Typical usage is like this: [with GNU print, GNU xargs: guaranteed to handle spaces, quotes, etc. in file names] find . -name '*.[ch]' -print0 | xargs -0 $0 's/\bCONST\b/const/g'\n [with non-GNU print, xargs] find . -name '*.[ch]' -print | xargs $0 's/\bCONST\b/const/g'\n The file is read in, either line by line (with --line-mode specified) or in one big hunk (with --hunk-mode specified; it's the default), and the Perl expression is then evalled with \$_ set to the line or hunk of text, including the terminating newline if there is one. It should destructively modify the value there, storing the changed result in \$_. Files in which any modifications are made are backed up to the directory specified using --backup-dir, or to `backup' by default. To disable this, use --backup-dir= with no argument. Hunk mode is the default because it is MUCH MUCH faster than line-by-line. Use line-by-line only when it matters, e.g. you want to do a replacement only once per line (the default without the `g' argument). Conversely, when using hunk mode, *ALWAYS* use `g'; otherwise, you will only make one replacement in the entire file! "; my %options = (); $Getopt::Long::ignorecase = 0; &GetOptions ( \%options, 'help', 'backup-dir=s', 'line-mode', 'hunk-mode', ); die $usage if $options{"help"} or @ARGV <= 1; my $code = shift; die $usage if grep (-d || ! -w, @ARGV); sub SafeOpen { open ((my $fh = new FileHandle), $_[0]); confess "Can't open $_[0]: $!" if ! defined $fh; return $fh; } sub SafeClose { close $_[0] or confess "Can't close $_[0]: $!"; } sub FileContents { my $fh = SafeOpen ("< $_[0]"); my $olddollarslash = $/; local $/ = undef; my $contents = <$fh>; $/ = $olddollarslash; return $contents; } sub WriteStringToFile { my $fh = SafeOpen ("> $_[0]"); binmode $fh; print $fh $_[1] or confess "$_[0]: $!\n"; SafeClose $fh; } foreach my $file (@ARGV) { my $changed_p = 0; my $new_contents = ""; if ($options{"line-mode"}) { my $fh = SafeOpen $file; while (<$fh>) { my $save_line = $_; eval $code; $changed_p = 1 if $save_line ne $_; $new_contents .= $_; } } else { my $orig_contents = $_ = FileContents $file; eval $code; if ($_ ne $orig_contents) { $changed_p = 1; $new_contents = $_; } } if ($changed_p) { my $backdir = $options{"backup-dir"}; $backdir = "backup" if !defined ($backdir); if ($backdir) { my ($name, $path, $suffix) = fileparse ($file, ""); my $backfulldir = $path . $backdir; my $backfile = "$backfulldir/$name"; mkdir $backfulldir, 0755 unless -d $backfulldir; print "modifying $file (original saved in $backfile)\n"; rename $file, $backfile; } WriteStringToFile ($file, $new_contents); } } ----------------------------------- cut ------------------------------------ In addition to those programs, I needed to fix up a few other things, particularly relating to the duplicate definitions of types, now that some types merged with others. Specifically: 1. in lisp.h, removed duplicate declarations of Bytecount. The changed code should now look like this: (In each code snippet below, the first and last lines are the same as the original, as are all lines outside of those lines. That allows you to locate the section to be replaced, and replace the stuff in that section, verifying that there isn't anything new added that would need to be kept.) --------------------------------- snip ------------------------------------- /* Counts of bytes or chars */ typedef EMACS_INT Bytecount; typedef EMACS_INT Charcount; /* Counts of elements */ typedef EMACS_INT Elemcount; /* Hash codes */ typedef unsigned long Hashcode; /* ------------------------ dynamic arrays ------------------- */ --------------------------------- snip ------------------------------------- 2. in lstream.h, removed duplicate declaration of Bytecount. Rewrote the comment about this type. The changed code should now look like this: --------------------------------- snip ------------------------------------- #endif /* The have been some arguments over the what the type should be that specifies a count of bytes in a data block to be written out or read in, using Lstream_read(), Lstream_write(), and related functions. Originally it was long, which worked fine; Martin "corrected" these to size_t and ssize_t on the grounds that this is theoretically cleaner and is in keeping with the C standards. Unfortunately, this practice is horribly error-prone due to design flaws in the way that mixed signed/unsigned arithmetic happens. In fact, by doing this change, Martin introduced a subtle but fatal error that caused the operation of sending large mail messages to the SMTP server under Windows to fail. By putting all values back to be signed, avoiding any signed/unsigned mixing, the bug immediately went away. The type then in use was Lstream_Data_Count, so that it be reverted cleanly if a vote came to that. Now it is Bytecount. Some earlier comments about why the type must be signed: This MUST BE SIGNED, since it also is used in functions that return the number of bytes actually read to or written from in an operation, and these functions can return -1 to signal error. Note that the standard Unix read() and write() functions define the count going in as a size_t, which is UNSIGNED, and the count going out as an ssize_t, which is SIGNED. This is a horrible design flaw. Not only is it highly likely to lead to logic errors when a -1 gets interpreted as a large positive number, but operations are bound to fail in all sorts of horrible ways when a number in the upper-half of the size_t range is passed in -- this number is unrepresentable as an ssize_t, so code that checks to see how many bytes are actually written (which is mandatory if you are dealing with certain types of devices) will get completely screwed up. --ben */ typedef enum lstream_buffering --------------------------------- snip ------------------------------------- 3. in dumper.c, there are four places, all inside of switch() statements, where XD_BYTECOUNT appears twice as a case tag. In each case, the two case blocks contain identical code, and you should *REMOVE THE SECOND* and leave the first.

/* Line number cache.
   Copyright (C) 1997 Free Software Foundation, Inc.

This file is part of XEmacs.

XEmacs is free software; you can redistribute it and/or modify it
under the terms of the GNU General Public License as published by the
Free Software Foundation; either version 2, or (at your option) any
later version.

XEmacs is distributed in the hope that it will be useful, but WITHOUT
ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License
for more details.

You should have received a copy of the GNU General Public License
along with XEmacs; see the file COPYING.  If not, write to
the Free Software Foundation, Inc., 59 Temple Place - Suite 330,
Boston, MA 02111-1307, USA.  */

/* Synched up with: Not in FSF. */

/* To calculate the line numbers, redisplay must count the newlines
   from a known position.  This used to be BUF_BEGV, but this made the
   line numbering extremely slow for large buffers, because Emacs had
   to rescan the whole buffer at each redisplay.

   To make line numbering efficient, we maintain a buffer-local cache
   of recently used positions and their line numbers.  The cache is
   implemented as a small ring of cache positions.  A cache position
   is either nil or a cons of a buffer position (marker) and the
   corresponding line number.

   When calculating the line numbers, this cache is consulted if it
   would otherwise take too much time to count the newlines in the
   buffer (see the comment to buffer_line_number().)

   Insertion and deletions that contain/delete newlines invalidate the
   cached positions after the insertion point.  This guarantees
   relatively fast line numbers caching (even in buffers where point
   moves a lot), and low memory usage.  All of this is done only in
   the buffers where the cache is actually initialized -- i.e. where
   line-numbering is on, and you move the point farther than
   LINE_NUMBER_FAR from the beginning of buffer.  In this sense, the
   cache is lazy -- if you don't use it, you don't pay for it.

   NOTE: line-number cache should not be confused with line-start
   cache.  Line-start cache (a part of redisplay) works with the
   display lines, whereas this works with the buffer lines (literally
   counting the newlines).  */

#include <config.h>
#include "lisp.h"
#include "buffer.h"

#include "line-number.h"

/* #### The following three values could stand more exploration for
   best performance.  */

/* Size of the ring.  The current code expects this to be a small
   number.  If you make it larger, you should probably optimize the
   code below to keep it sorted. */
#define LINE_NUMBER_RING_SIZE 8

/* How much traversal has to be exceeded for two points to be
   considered "far" from each other.  When two points are far, cache
   will be used.  */
#define LINE_NUMBER_FAR 16384

/* How large a string has to be to give up searching it for newlines,
   before change. */
#define LINE_NUMBER_LARGE_STRING 256

/* To be used only when you *know* the cache has been allocated!  */
#define LINE_NUMBER_RING(b) (XCAR ((b)->text->line_number_cache))
#define LINE_NUMBER_BEGV(b) (XCDR ((b)->text->line_number_cache))


/* Initialize the cache.  Cache is (in pseudo-BNF):

   CACHE		= nil | INITIALIZED-CACHE
   INITIALIZED-CACHE	= cons (RING, BEGV-LINE)
   RING			= vector (*RING-ELEMENT)
   RING-ELEMENT		= nil | RING-PAIR
   RING-PAIR		= cons (marker, integer)
   BEGV-LINE		= integer

   Line number cache should never, ever, be visible to Lisp (because
   destructively modifying its elements can cause crashes.)  Debug it
   using debug_print (current_buffer->text->last_number_cache).  */
static void
allocate_line_number_cache (struct buffer *b)
{
  b->text->line_number_cache = Fcons (make_vector (LINE_NUMBER_RING_SIZE, Qnil),
				      Qzero);
  narrow_line_number_cache (b);
}

/* Flag LINE_NUMBER_BEGV (b) as dirty.  Do it only if the line number
   cache is already initialized.  */
void
narrow_line_number_cache (struct buffer *b)
{
  if (NILP (b->text->line_number_cache))
    return;

  if (BUF_BEG (b) == BUF_BEGV (b))
    /* The is the case Fwiden and save_restriction_restore.  Since we
       know the correct value, we can update it now.  */
    LINE_NUMBER_BEGV (b) = Qzero;
  else
    /* Calculating the line number of BUF_BEGV here is a bad idea,
       because there is absolutely no reason to do it before the next
       redisplay.  We simply mark it as dirty instead.  */
    LINE_NUMBER_BEGV (b) = make_int (-1);
}

/* Invalidate the line number cache positions that lie after POS. */
static void
invalidate_line_number_cache (struct buffer *b, Charbpos pos)
{
  EMACS_INT i, j;
  Lisp_Object *ring = XVECTOR_DATA (LINE_NUMBER_RING (b));

  for (i = 0; i < LINE_NUMBER_RING_SIZE; i++)
    {
      if (!CONSP (ring[i]))
	break;
      /* As the marker stays behind the insertions, this check might
         as well be `>'.  However, Finsert_before_markers can advance
         the marker anyway, which bites in shell buffers.

	 #### This forces recreation of the cached marker (and
	 recalculation of newlines) every time a newline is inserted
	 at point, which is way losing.  Isn't there a way to make a
	 marker impervious to Finsert_before_markers()??  Maybe I
	 should convert the code to use extents.  */
      if (marker_position (XCAR (ring[i])) >= pos)
	{
	  /* Get the marker out of the way.  */
	  Fset_marker (XCAR (ring[i]), Qnil, Qnil);
	  /* ...and shift the ring elements, up to the first nil.  */
	  for (j = i; !NILP (ring[j]) && j < LINE_NUMBER_RING_SIZE - 1; j++)
	    ring[j] = ring[j + 1];
	  ring[j] = Qnil;
	  /* Must recheck position i. */
	  i--;
	}
    }
}

/* Invalidate the cache positions after POS, if the string to be
   inserted contains a newline.  If the string is too large (larger
   than LINE_NUMBER_LARGE_STRING), invalidate the cache positions
   after POS without prior search.

   This will do nothing if the cache is uninitialized.  */
void
insert_invalidate_line_number_cache (struct buffer *b, Charbpos pos,
				     const Intbyte *nonreloc, Bytecount length)
{
  if (NILP (b->text->line_number_cache))
    return;

  if (length > LINE_NUMBER_LARGE_STRING
      ||
      /* We could also count how many newlines there are in the string
         and update the cache accordingly, but it would be too much
         work for too little gain. */
      memchr ((void *)nonreloc, '\n', length))
    invalidate_line_number_cache (b, pos);
}

/* Invalidate the cache positions after FROM, if the region to be
   deleted contains a newline.  If the region-to-be-deleted is larger
   than LINE_NUMBER_LARGE_STRING, invalidate the cache positions after
   FROM without unconditionally.

   This will do nothing if the cache is uninitialized.  */
void
delete_invalidate_line_number_cache (struct buffer *b, Charbpos from, Charbpos to)
{
  if (NILP (b->text->line_number_cache))
    return;

  if ((to - from) > LINE_NUMBER_LARGE_STRING)
    invalidate_line_number_cache (b, from);
  else
    {
      EMACS_INT shortage;
      scan_buffer (b, '\n', from, to, 1, &shortage, 0);
      if (!shortage)
	invalidate_line_number_cache (b, from);
    }
}

/* Get the nearest known position we know the line number of
   (i.e. BUF_BEGV, and cached positions).  The return position will be
   either closer than BEG, or BEG.  The line of this known position
   will be stored in LINE.

   *LINE should be initialized to the line number of BEG (normally,
   BEG will be BUF_BEGV, and *LINE will be XINT (LINE_NUMBER_BEGV).
   This will initialize the cache, if necessary.  */
static void
get_nearest_line_number (struct buffer *b, Charbpos *beg, Charbpos pos,
			 EMACS_INT *line)
{
  EMACS_INT i;
  Lisp_Object *ring = XVECTOR_DATA (LINE_NUMBER_RING (b));
  Charcount length = pos - *beg;

  if (length < 0)
    length = -length;

  /* Find the ring entry closest to POS, if it is closer than BEG. */
  for (i = 0; i < LINE_NUMBER_RING_SIZE && CONSP (ring[i]); i++)
    {
      Charbpos newpos = marker_position (XCAR (ring[i]));
      Charcount howfar = newpos - pos;
      if (howfar < 0)
	howfar = -howfar;
      if (howfar < length)
	{
	  length = howfar;
	  *beg = newpos;
	  *line = XINT (XCDR (ring[i]));
	}
    }
}

/* Add a (POS . LINE) pair to the ring, and rotate it. */
static void
add_position_to_cache (struct buffer *b, Charbpos pos, EMACS_INT line)
{
  Lisp_Object *ring = XVECTOR_DATA (LINE_NUMBER_RING (b));
  int i = LINE_NUMBER_RING_SIZE - 1;

  /* Set the last marker in the ring to point nowhere. */
  if (CONSP (ring[i]))
    Fset_marker (XCAR (ring[i]), Qnil, Qnil);

  /* Rotate the ring... */
  for (; i > 0; i--)
    ring[i] = ring[i - 1];

  /* ...and update it. */
  ring[0] = Fcons (Fset_marker (Fmake_marker (), make_int (pos),
				make_buffer (b)),
		   make_int (line));
}

/* Calculate the line number in buffer B at position POS.  If CACHEP
   is non-zero, initialize and facilitate the line-number cache.  The
   line number of the first line is 0.  If narrowing is in effect,
   count the lines are counted from the beginning of the visible
   portion of the buffer.

   The cache works as follows: To calculate the line number, we need
   two positions: position of point (POS) and the position from which
   to count newlines (BEG).  We start by setting BEG to BUF_BEGV.  If
   this would require too much searching (i.e. pos - BUF_BEGV >
   LINE_NUMBER_FAR), try to find a closer position in the ring.  If it
   is found, use that position for BEG, and increment the line number
   appropriately.

   If the calculation (with or without the cache lookup) required more
   than LINE_NUMBER_FAR characters of traversal, update the cache.  */
EMACS_INT
buffer_line_number (struct buffer *b, Charbpos pos, int cachep)
{
  Charbpos beg = BUF_BEGV (b);
  EMACS_INT cached_lines = 0;
  EMACS_INT shortage, line;

  if ((pos > beg ? pos - beg : beg - pos) <= LINE_NUMBER_FAR)
    cachep = 0;

  if (cachep)
    {
      if (NILP (b->text->line_number_cache))
	allocate_line_number_cache (b);
      /* If we don't know the line number of BUF_BEGV, calculate it now.  */
      if (XINT (LINE_NUMBER_BEGV (b)) == -1)
	{
	  LINE_NUMBER_BEGV (b) = Qzero;
	  /* #### This has a side-effect of changing the cache.  */
	  LINE_NUMBER_BEGV (b) =
	    make_int (buffer_line_number (b, BUF_BEGV (b), 1));
	}
      cached_lines = XINT (LINE_NUMBER_BEGV (b));
      get_nearest_line_number (b, &beg, pos, &cached_lines);
    }

  scan_buffer (b, '\n', beg, pos, pos > beg ? EMACS_INT_MAX : -EMACS_INT_MAX,
	       &shortage, 0);

  line = EMACS_INT_MAX - shortage;
  if (beg > pos)
    line = -line;
  line += cached_lines;

  if (cachep)
    {
      /* If too far, update the cache. */
      if ((pos > beg ? pos - beg : beg - pos) > LINE_NUMBER_FAR)
	add_position_to_cache (b, pos, line);
      /* Account for narrowing.  If cache is not used, this is
	 unnecessary, because we counted from BUF_BEGV anyway.  */
      line -= XINT (LINE_NUMBER_BEGV (b));
    }

  return line;
}