patch request for 21.1.13
24 years, 4 months
Isaac Hollander
Isaac Hollander
Last year I submitted a patch to the main Makefile.in to make tar a macro.
http://www.xemacs.org/list-archives/xemacs-patches/199911/msg00061.html
Jan adapted the patch to Makefile.in.in and applied it to 21.2.x.
Would it be possible to add this patch to 21.1.13? It would make my job
of maintaining xemacs builds on several different unix variants much
easier, since I can then override the system tar with GNU tar with the
options I need (specifically -h)
Thanks
Isaac
- 2 participants
- 1 comments
Problems with alloca on Alpha Linux with Compaq cc
24 years, 4 months
Falk Hueffner
Falk Hueffner
Hi,
with xemacs-21.2 CVS from October 12 and Compaq C V6.2-504 I get
[...]
checking for working alloca.h... no
checking for alloca... no
Defining C_ALLOCA
checking whether alloca needs Cray hooks... no
checking stack direction for C alloca... -1
Defining STACK_DIRECTION = -1
xemacs will be linked with "alloca.o"
[...]
The reason is that ccc won't find alloca when linking if passed
"-std1", which is added by xemacs configure. (ccc picks up gcc's
alloca.h, but the alloca.h check wants to link, too, so it fails.)
Then when building one gets lots of warnings:
ccc -O3 -Demacs -I. -I../src -I/usr/local/src/xemacs-21.2-10.12/lib-src -I/usr/local/src/xemacs-21.2-10.12/src -DHAVE_CONFIG_H -std1 /usr/local/src/xemacs-21.2-10.12/lib-src/wakeup.c -ldb -lgpm -lncurses -L/usr/lib -lesd -laudiofile -lm -lm -lutil -lgcc -lc -lgcc /usr/lib/crtn.o -o wakeup
cc: Warning: /usr/include/alloca.h, line 33: In this declaration, the type of the function "alloca", declared in a system header file, is not compatible with the earlier declaration of "alloca" at line number 43 in file ../src/config.h. (funcredeclext1)
extern __ptr_t alloca __P ((size_t __size));
---------------^
(__ptr_t is a #define for void *, while config.h declares the return
value as char *)
Finally, it fails:
ccc -c -O3 -Demacs -I. -I../src -I/usr/local/src/xemacs-21.2-10.12/lib-src -I/usr/local/src/xemacs-21.2-10.12/src -DHAVE_CONFIG_H -std1 /usr/local/src/xemacs-21.2-10.12/lib-src/../src/alloca.c
cc: Error: /usr/local/src/xemacs-21.2-10.12/lib-src/../src/alloca.c, line 191: In this declaration, the type of "alloca" is not compatible with the type of a previous declaration of "alloca" at line number 43 in file ../src/config.h. (notcompat)
alloca (size)
^
because alloca.c declares alloca as returning "pointer", which is
typedefed to void *.
Now I wonder how to fix this best; one could for example:
* remove -std1 from the ccc flags
* use something like
#if defined(__DECC) && defined(__linux__)
# include <machine/builtins.h>
# define alloca __ALLOCA
#endif
which also works with -std1
* consistentialize xemacs' declarations; regex.c also wants to declare
alloca as returning char *.
And if anybody feels a bit bored, here are a few interesting warnings:
cc: Warning: miscplay.c, line 384: In this statement, the expression "*dest++=(unsigned char)(((int)*((signed char ...)(src++))+(int)*((signed char ...)(src++)))/2)" modifies the variable "src" more than once without an intervening sequence point. This behavior is undefined. (undefvarmod)
*dest++ = (unsigned char)(((int)*((signed char *)(src++)) +
----^
cc: Warning: miscplay.c, line 405: In this statement, the expression "*dest++=(unsigned char)(((int)*((signed char ...)(src++))+(int)*((signed char ...)(src++)))/2)^0X0000000000000080" modifies the variable "src" more than once without an intervening sequence point. This behavior is undefined. (undefvarmod)
*dest++ = (unsigned char)(((int)*((signed char *)(src++)) +
----^
cc: Warning: miscplay.c, line 496: In this statement, the expression "*p++=ulaw_dsp[*p]" modifies "p", and fetches its value in a computation that is not used to produce the modified value without an intervening sequence point. This behavior is undefined. (undefvarfetch)
*p++ = ulaw_dsp[*p];
----^
cc: Warning: miscplay.c, line 570: In this statement, the expression "*p++=((*p&0X00000000000000FF)<<8)|(*p>>8)" modifies "p", and fetches its value in a computation that is not used to produce the modified value without an intervening sequence point. This behavior is undefined. (undefvarfetch)
*p++ = ((*p & 0x00ff) << 8) | (*p >> 8);
----^
cc: Warning: miscplay.c, line 570: In this statement, the expression "*p++=((*p&0X00000000000000FF)<<8)|(*p>>8)" modifies "p", and fetches its value in a computation that is not used to produce the modified value without an intervening sequence point. This behavior is undefined. (undefvarfetch)
*p++ = ((*p & 0x00ff) << 8) | (*p >> 8);
----^
cc: Warning: regex.c, line 5627: This statement uses the type "union declared without a tag" to reference the same storage location as the statement at line number 5627 in file regex.c, which uses the type "long". This does not conform to the ANSI aliasing rules. (badansialias)
POP_FAILURE_POINT (d, p,
----------^
cc: Warning: regex.c, line 5312: This statement uses the type "union declared without a tag" to reference the same storage location as the statement at line number 5312 in file regex.c, which uses the type "long". This does not conform to the ANSI aliasing rules. (badansialias)
POP_FAILURE_POINT (sdummy, pdummy,
------------^
Falk
- 3 participants
- 4 comments
External modules
24 years, 5 months
Jerry James
Jerry James
It appears that the consensus from the discussion following my earlier
modules question was that:
a) I should do my own autoconf magic (which, as somebody mentioned,
wasn't all that hard); and
b) I should structure my module so that it can be compiled outside of
the XEmacs source tree (i.e., make it an external module).
For the information of whoever is responsible for modules (I don't see
anyone listed on <URL:http://xemacs.sourceforge.net/Develop/jobs.html>),
here are the obstacles I have encountered so far in accomplishing (b),
working with the XEmacs 21.2.37 distribution:
1) Missing header files.
a) emodules.h is not installed anywhere when XEmacs is installed,
so I can't very well include it.
b) A bunch of other header files, needed by emodules.h, are also
not installed. These include lisp.h, sysdep.h, window.h,
buffer.h, etc.
Incidentally, speaking of installed header files, why are the
entire s&m directories installed in
.../lib/xemacs-21.2-b37/config-name/include? It's a
machine-specific directory, so why do I want header files for every
machine under the sun?
2) Lack of examples.
a) The examples in the modules directory of the XEmacs distribution
are all of internal modules. That is, they only build inside
the XEmacs source tree. And even then, they aren't right. They
only build inside the source tree *if you built XEmacs itself
inside the source tree*. If you, like me, build XEmacs for
multiple platforms, and therefore use VPATH to build outside of
the source tree, the modules won't build at all, due to
assumptions about the location of ellcc and make-docfile.
b) The example Makefiles (and, strangely, not the README in the
modules directory) all refer to "samples in the sub-directory
'installed'" to show how to write modules to be compiled outside
of the XEmacs source tree. That subdirectory is not part of the
XEmacs 21.2.37 distribution, the comments are in the wrong
place, and I doubt that such examples even exist due to problem
#1.
I am willing to be a guinea pig to help get external modules working,
but I need some direction from the list and whoever is coordinating
module development.
Regards,
--
Jerry James
- 3 participants
- 8 comments
Petition for packages: sml-mode and haskell-mode
24 years, 5 months
José Romildo Malaquias
José Romildo Malaquias
Hello.
Would someone add the packages sml-mode and
haskell-mode to the collection of XEmacs
packages? They are modes for editing source
code written in SML and Haskell. URLs:
ftp://rum.cs.yale.edu/pub/monnier/sml-mode/
http://www.haskell.org/haskell-mode/
Regards,
Romildo
--
Prof. José Romildo Malaquias <romildo(a)iceb.ufop.br>
Departamento de Computação
Universidade Federal de Ouro Preto
Brasil
- 3 participants
- 3 comments
Re: Xemacs Mule + koi8 fonts question.
24 years, 5 months
Hrvoje Niksic
Hrvoje Niksic
"Andrew W. Nosenko" <awn(a)bcs.zp.ua> writes:
> (global-set-key [Cyrillic_shorti] (ru-keys-insert-char-m ?й))
...
Note that they are nicer ways of achieving this. What you could do
is:
(global-set-key 'Cyrillic_shorti 'self-insert-command)
(put 'Cyrillic_shorti 'ascii-character ?й)
...
Then your Russian keys would work with minimal intervention, and with
the regular self-insert code -- no more need to hack overwrite mode
and all that.
Or, if you want to play with a bit more advanced keyboard remapping
hacks, why not take a look at my `latin2.el'? Not Russian, but solves
similar problems.
- 2 participants
- 1 comments
Mule redisplay crash
24 years, 5 months
Gunnar Evermann
Gunnar Evermann
This is one of the oldest crashes on my to do list.
loading the following in a vanilla Mule XEmacs will trigger a crash.
(defun delete-and-split ()
(interactive)
(delete-other-windows)
(split-window))
(split-window-vertically)
(insert "$B$,(B")
(backward-char)
(redraw-frame)
(delete-and-split)
Fatal error: assertion failed, file /deb-huge/home/gunnar/src/XEmacs/xemacs/src/objects.h, line 143, RECORD_TYPEP (obj, &lrecord_font_instance)
#3 0x808eb81 in assert_failed (file=0x8218c60 "/deb-huge/home/gunnar/src/XEmacs/xemacs/src/objects.h", line=143,
expr=0x8218d40 "RECORD_TYPEP (obj, &lrecord_font_instance)")
at /deb-huge/home/gunnar/src/XEmacs/xemacs/src/emacs.c:2825
#4 0x80c9db4 in error_check_font_instance (obj=136776524)
at /deb-huge/home/gunnar/src/XEmacs/xemacs/src/objects.h:143
#5 0x81c730f in x_output_string (w=0x85a8c10, dl=0x8609808, buf=0x85e7640, xpos=4, xoffset=0, clip_start=4,
width=16, findex=0, cursor=1, cursor_start=4, cursor_width=16, cursor_height=13)
at /deb-huge/home/gunnar/src/XEmacs/xemacs/src/redisplay-x.c:883
#6 0x81c60d5 in x_output_display_block (w=0x85a8c10, dl=0x8609808, block=0, start=0, end=1, start_pixpos=4,
cursor_start=4, cursor_width=16, cursor_height=13)
at /deb-huge/home/gunnar/src/XEmacs/xemacs/src/redisplay-x.c:390
#7 0x81799c2 in redisplay_output_display_block (w=0x85a8c10, dl=0x8609808, block=0, start=0, end=1,
start_pixpos=4, cursor_start=4, cursor_width=16, cursor_height=13)
at /deb-huge/home/gunnar/src/XEmacs/xemacs/src/redisplay-output.c:1088
#8 0x8178a28 in output_display_line (w=0x85a8c10, cdla=0x0, ddla=0x85e2668, line=1, force_start=4, force_end=20)
at /deb-huge/home/gunnar/src/XEmacs/xemacs/src/redisplay-output.c:751
#9 0x8179839 in redraw_cursor_in_window (w=0x85a8c10, run_end_begin_meths=1)
at /deb-huge/home/gunnar/src/XEmacs/xemacs/src/redisplay-output.c:1029
#10 0x81798be in redisplay_redraw_cursor (f=0x8577f00, run_end_begin_meths=1)
at /deb-huge/home/gunnar/src/XEmacs/xemacs/src/redisplay-output.c:1054
#11 0x81760e9 in Fforce_cursor_redisplay (frame=139951872)
at /deb-huge/home/gunnar/src/XEmacs/xemacs/src/redisplay.c:8938
#12 0x8176761 in text_cursor_visible_p_changed (specifier=137474936, w=0x85e8148, oldval=136988916)
at /deb-huge/home/gunnar/src/XEmacs/xemacs/src/redisplay.c:9005
#13 0x81999b1 in recompute_one_cached_specifier_in_window (specifier=137474936, w=0x85e8148)
at /deb-huge/home/gunnar/src/XEmacs/xemacs/src/specifier.c:2812
#14 0x8199b67 in recompute_all_cached_specifiers_in_window (w=0x85e8148)
at /deb-huge/home/gunnar/src/XEmacs/xemacs/src/specifier.c:2854
#15 0x81d5e1e in Fset_window_buffer (window=140411208, buffer=137541312)
at /deb-huge/home/gunnar/src/XEmacs/xemacs/src/window.c:3198
#16 0x81d6b4e in Fsplit_window (window=140151824, chsize=136988916, horflag=136988916)
at /deb-huge/home/gunnar/src/XEmacs/xemacs/src/window.c:3503
#17 0x8094ed9 in Feval (form=140010872) at /deb-huge/home/gunnar/src/XEmacs/xemacs/src/eval.c:2975
#18 0x8090870 in Fprogn (args=140010812) at /deb-huge/home/gunnar/src/XEmacs/xemacs/src/eval.c:767
#19 0x8096629 in funcall_lambda (fun=140010896, nargs=0, args=0xbffff048)
at /deb-huge/home/gunnar/src/XEmacs/xemacs/src/eval.c:3497
#20 0x8095522 in Feval (form=140007804) at /deb-huge/home/gunnar/src/XEmacs/xemacs/src/eval.c:3081
#21 0x8146b51 in readevalloop (readcharfun=140386240, sourcename=139813548, evalfun=0x80947b0 <Feval>, printflag=0)
at /deb-huge/home/gunnar/src/XEmacs/xemacs/src/lread.c:1448
[...]
#43 0x808d520 in xemacs_21_2_b23_i586_pc_linux (argc=4, argv=0xbffffc34, envp=0xbffffc48, restart=0)
at /deb-huge/home/gunnar/src/XEmacs/xemacs/src/emacs.c:1898
#44 0x808e0ab in main (argc=4, argv=0xbffffc34, envp=0xbffffc48)
at /deb-huge/home/gunnar/src/XEmacs/xemacs/src/emacs.c:2323
The problem is that the face cache doesn't contain the font instance
for the relevant charset or to be precise that the cache has been
cleared but we do not regenerate the necessary entries.
A quick fix for the crash would be adding a call to
ensure_face_cachel_contains_charset() at the relevant place in
x_output_string like this:
Index: redisplay-x.c
===================================================================
RCS file: /usr/CVSroot/XEmacs/xemacs/src/redisplay-x.c,v
retrieving revision 1.23.2.17
diff -u -r1.23.2.17 redisplay-x.c
--- redisplay-x.c 1999/10/15 10:11:51 1.23.2.17
+++ redisplay-x.c 1999/12/13 22:58:35
＠＠ -879,10 +879,14 ＠＠
for (i = 0; i < nruns; i++)
{
- Lisp_Object font = FACE_CACHEL_FONT (cachel, runs[i].charset);
- struct Lisp_Font_Instance *fi = XFONT_INSTANCE (font);
+ Lisp_Object font;
+ struct Lisp_Font_Instance *fi;
int this_width;
int need_clipping;
+
+ ensure_face_cachel_contains_charset(cachel, window, runs[i].charset);
+ font = FACE_CACHEL_FONT (cachel, runs[i].charset);
+ fi = XFONT_INSTANCE (font);
if (EQ (font, Vthe_null_font_instance))
continue;
This works but is kindof expensive to do, as x_output_string is called
very very often :-)
Maybe our Mule and redisplay experts can try to come up with something
more clever...
Gunnar
- 3 participants
- 2 comments
crash when loading file which contains EUC-JP and ISO-2022-JP
24 years, 5 months
Mike FABIAN
Mike FABIAN
Save the following attached file:
(This file is only gzipped, because Gnus runs into an endless loop
when I attach it without gzipping. That is probably also a bug, but a
different one.).
Gunzip it, start XEmacs and let it load this file:
mfabian＠gregory:~/xemacs-bug$ LANG=ja_JP xemacs -q -vanilla kanji.euc-iso
Fatal error (11).
Your files have been auto-saved.
Use `M-x recover-session' to recover them.
It is important to use "LANG=ja_JP", it will not happen with "LANG=C".
The file "kanji.euc-iso" contains only two lines. Both lines contain
only the Japanese word "漢字" (Kanji). The first line contains this
word in EUC-JP encoding, the second line in ISO-2022-JP encoding.
The order of these lines is also important, if the ISO-2022-JP line
comes first like in the gunzipped version of this file:
XEmacs will not crash (but of course the EUC-JP part is unreadable).
I encounter this problem often when opening Japanese *.patch files,
which often contain both EUC-JP in some hunks and ISO-2022-JP in other
hunks. If the EUC-JP comes first, XEmacs will crash.
I use the following XEmacs version:
In XEmacs 21.1 (patch 10) "Capitol Reef" [Lucid] (i686-pc-linux, Mule) of Mon May 8 2000 on nozomi
configured using `configure --prefix=/opt/stow/xemacs-21.1.10 --with-mule=yes --with-dialogs=athena3d --with-xface=yes --with-wnn=yes --with-canna=yes --with-xim=xlib --with-xfs --with-menubars=lucid --site-includes=/opt/include --site-libraries=/opt/lib '--cflags=-g -O4 -fno-strength-reduce''
(The same problem also occurs with the XEmacs 21.1.12 from the SuSE
rpm).
Here is the lisp backtrace:
And here is the C backtrace:
--
Mike Fabian <mfabian(a)suse.de>
- 3 participants
- 4 comments
Re: Commenting out code in lisp
24 years, 5 months
Adrian Aichner
Adrian Aichner
>>>>> "Stephen" == Stephen J Turnbull <turnbull(a)sk.tsukuba.ac.jp> writes:
Stephen> Deleting XEmacs Review from the CC list.
>>>>> "APA" == Adrian Aichner <adrian(a)xemacs.org> writes:
APA> Using double-semicolon will set up the same re-indentation
APA> trap
Stephen> Only if the surrounding code moves. The double semicolon
Stephen> style should indent the first semicolon to the current
Stephen> indent-column, and push all the blocked-out code 2 (or 3)
Stephen> columns to the right. The lines get a little longer, but
This is the problem I refer to. ;; comments so indented cannot be
uncommented using C-u M-x comment-region. Using ;;; avoid this
short-coming of C-u M-x comment-region.
Do we want to make C-u M-x comment-region smarter?
Stephen> not disasterously so unless the code is very deeply
Stephen> nested in the first place.
APA> and make uncomment-region impossible.
Stephen> Not as long as it can assume that the region is
Stephen> double-semicolon commented.
That's not how C-u M-x comment-region works currently.
Stephen> I don't know how much the existing code would have to
Stephen> change. I don't oppose your proposal on the grounds that
Stephen> ;; is better than ;;;, for sure. I just put it forward
Stephen> as one I like better.
I see your point. Maybe C-u M-x comment-region needs to become smarter.
APA> If my proposal to comment broken or unneeded blocks of code
APA> (if they shouldn't be deleted alltogether) by triple
APA> semicolon makes sense, then we should make it the dafault.
Stephen> Well, that kinda depends on whether we would screw up
Stephen> usage outside of the core Lisp. I don't know anything
Stephen> about that, but if people are used to the single
Stephen> semicolon style, this would be something of a shocker.
Stephen> :-)
Not any more than re-indenting this from, ahem, lisp-mnt.el:
(require 'picture) ; provides move-to-column-force
;(require 'emacsbug) ; XEmacs, not needed for bytecompilation
APA> However, re-indenting XEmacs core lisp files (unless
APA> otherwise noted explicitely) should be harmless and
APA> desirable.
Stephen> Please, no! Reindent code that you have to touch for
Stephen> other reasons, OK. But gratuitous format changes create
Stephen> distracting diffs and conflicts in update.
I agree with you. But having to manually indent functions, just
because someone else had used default XEmacs commenting on it before,
is a real pain.
Best regards,
Adrian
Stephen> --
Stephen> University of Tsukuba Tennodai 1-1-1 Tsukuba 305-8573 JAPAN
Stephen> Institute of Policy and Planning Sciences Tel/fax: +81 (298) 53-5091
Stephen> _________________ _________________ _________________ _________________
Stephen> What are those straight lines for? "XEmacs rules."
--
Adrian Aichner
mailto:adrian＠xemacs.org
http://www.xemacs.org
- 2 participants
- 1 comments
Re: Attention Mule users...
24 years, 5 months
Jan Vroonhof
Jan Vroonhof
npak(a)ispras.ru (Nick V. Pakoulin) writes:
> NVP> contains changes to ps-mule that I made to enable printing in russian.
This looks promising..
> --- - Thu Nov 30 16:27:49 2000
> +++ lisp/ps-print/ps-mule.el Thu Nov 30 15:12:55 2000
> ＠＠ -143,8 +143,9 ＠＠
>
> ;; For Emacs 20.2 and the earlier version.
> (eval-and-compile
> - (if (and (boundp 'mule-version) ; only if mule package is loaded
> - (not (string< mule-version "4.0")))
> +;; (if (and (boundp 'mule-version) ; only if mule package is loaded
> +;; (not (string< mule-version "4.0")))
> + (if (featurep 'mule)
> (progn
> (defalias 'ps-mule-next-point '1+)
> (defalias 'ps-mule-chars-in-string 'length)
This was to be expected
[What was the exact problem here (I don't have a mule-build handy). Is
"mule-version" not defined on our mule build does it not have the
right value]
> +;;; The following commented code does not compile under XEmacs 21.2.36
> +;;; npak(a)ispras.ru
> +
> [ Ethiopic support comment out]
Ok, we simply have no Ethoipic support
> ;; A charset which we are now processing.
> (defvar ps-mule-current-charset nil)
> ＠＠ -934,6 +939,10 ＠＠
> (string-as-unibyte
> (cond ((coding-system-p encoding)
> (encode-coding-string str encoding))
> + ;; XEmacs change npak(a)ispras.ru
> + ((and (symbolp encoding)
> + (find-coding-system encoding))
> + (encode-coding-string str (find-coding-system encoding)))
> ((functionp encoding)
> (funcall encoding str))
Hmm.. Do we have a slight API incompatibility here, i.e. can we not
take a coding system name in encode-coding-string?
Anyway, it just another minor problem.
> (and (boundp 'enable-multibyte-characters)
> - enable-multibyte-characters
> + ;; FIXME
> + ;;enable-multibyte-characters
I think is again just a misguided FSF-specific mule test
> [From another message:]
>
> (define-coding-system-alias 'cyrillic-iso-8bit 'iso-8859-5)
I think was the essential bit... FSF Emacs has a whole bunch of
different names for coding systems.
I think the core bits work.
Could you, just for laughs, try printing using the bdf fonts if
you have the intlfont package installed.
Jan
- 4 participants
- 4 comments
Re: file-truename in Windows native build
24 years, 5 months
Dan Holmsand
Dan Holmsand
Mike Alexander <mta(a)arbortext.com> writes:
> It's annoyed me for some time that the native Windows version of
> XEmacs often ends up with two (or more) buffers visiting the same
> file, even if find-file-compare-truenames or find-file-use-truenames
> is set.
I totally agree. In fact I also started to fix this a while ago...
[...]
> The attached patch changed Ffile_truename to use
> win32_get_long_filename on Windows after changing it to do something
> reasonable when given a path that contains a file or directory that
> doesn't exist. This latter change might cause a problem with code
> that used win32-long-file-name to test for the existence of a file,
> but I can't find any code that uses it at all, let alone for that
> purpose. I also changed xemacs.mak to remove realpath.c from the
> native Windows build since it isn't needed anymore. With this patch I
> haven't had any more problems with multiple buffers visiting the same
> file if I set find-file-compare-truenames or find-file-use-truenames.
I chose the other alternative - making xrealpath work on windows. I
obviously think this is a better choice, mainly because it's mine :-)
My code have some real advantages, though. Mainly, it also works with
cygwin xemacs, which is broken in basically the same way as the native
built one. Besides that the current code doesn't care for case or
short filenames, it is totally broken when you feed it UNC paths.
My code deals with UNC paths by lower-casing the server and share
parts of the filename (i.e. //server/share/...), since case is not
significant here.
It is perhaps also an advantage that the code is shared between unix,
cygwin and native windows xemacsen. It should, for example, be quite
easy to add support for Win2k's reparse points (which basically work
like symlinks).
I also definitely think find-file-compare-truenames should be true by
default on windows.
Anyway, here it is. I've been using it for a while, on both native
xemacs and cygwin. However, I haven't tested the code on unix (the
changes there are mainly trivial, though; mostly *s == '/' ->
IS_DIRECTORY_SEP (*s), etc.).
/dan
--
Dan Ola Holmsand
IB
dan(a)eyebee.com
2000-08-31 Dan Holmsand <dan(a)eyebee.com>
* buffer.c: Make find-file-compare-truenames default to true on
windows.
* realpath.c (win32_abs_start):
(cygwin_readlink):
(win32_readlink): New functions.
(xrealpath): Return really real filenames on windows.
* fileio.c (Ffile_truename): Make file-truename work on windows.
Index: src/buffer.c
===================================================================
RCS file: /usr/CVSroot/XEmacs/xemacs/src/buffer.c,v
retrieving revision 1.36.2.38
diff -u -u -r1.36.2.38 buffer.c
--- buffer.c 2000/08/18 17:16:00 1.36.2.38
+++ buffer.c 2000/08/31 21:19:18
＠＠ -2227,7 +2227,11 ＠＠
See also the variable find-file-use-truenames.
*/ );
+#if defined(CYGWIN) || defined(WIN32_NATIVE)
+ find_file_compare_truenames = 1;
+#else
find_file_compare_truenames = 0;
+#endif
DEFVAR_BOOL ("find-file-use-truenames", &find_file_use_truenames /*
If this is true, then a buffer's visited file-name will always be
Index: src/fileio.c
===================================================================
RCS file: /usr/CVSroot/XEmacs/xemacs/src/fileio.c,v
retrieving revision 1.51.2.26
diff -u -u -r1.51.2.26 fileio.c
--- fileio.c 2000/08/06 09:27:12 1.51.2.26
+++ fileio.c 2000/08/31 21:19:40
＠＠ -1316,14 +1316,30 ＠＠
we just use our own version in realpath.c. */
for (;;)
{
- p = (Extbyte *) memchr (p + 1, '/', elen - (p + 1 - path));
- if (p)
- *p = 0;
+ Extbyte *pos;
+#ifdef WIN32_NATIVE
+ if (IS_DRIVE (p[0]) && IS_DEVICE_SEP (p[1])
+ && IS_DIRECTORY_SEP (p[2]))
+ /* don't test c: on windows */
+ p = p+2;
+ else if (IS_DIRECTORY_SEP (p[0]) && IS_DIRECTORY_SEP (p[1]))
+ /* start after // */
+ p = p+1;
+#endif
+ for (pos = p + 1; pos < path + elen; pos++)
+ if (IS_DIRECTORY_SEP (*pos))
+ {
+ *(p = pos) = 0;
+ break;
+ }
+ if (p != pos)
+ p = 0;
+
if (xrealpath ((char *) path, resolved_path))
{
if (p)
- *p = '/';
+ *p = DIRECTORY_SEP;
else
break;
＠＠ -1342,13 +1358,13 ＠＠
{
int plen = elen - (p - path);
- if (rlen > 1 && resolved_path[rlen - 1] == '/')
+ if (rlen > 1 && IS_DIRECTORY_SEP (resolved_path[rlen - 1]))
rlen = rlen - 1;
if (plen + rlen + 1 > countof (resolved_path))
goto toolong;
- resolved_path[rlen] = '/';
+ resolved_path[rlen] = DIRECTORY_SEP;
memcpy (resolved_path + rlen + 1, p + 1, plen + 1 - 1);
}
break;
＠＠ -1361,12 +1377,12 ＠＠
{
Lisp_Object resolved_name;
int rlen = strlen (resolved_path);
- if (elen > 0 && XSTRING_BYTE (expanded_name, elen - 1) == '/'
- && !(rlen > 0 && resolved_path[rlen - 1] == '/'))
+ if (elen > 0 && IS_DIRECTORY_SEP (XSTRING_BYTE (expanded_name, elen - 1))
+ && !(rlen > 0 && IS_DIRECTORY_SEP (resolved_path[rlen - 1])))
{
if (rlen + 1 > countof (resolved_path))
goto toolong;
- resolved_path[rlen++] = '/';
+ resolved_path[rlen++] = DIRECTORY_SEP;
resolved_path[rlen] = '\0';
}
TO_INTERNAL_FORMAT (DATA, (resolved_path, rlen),
Index: src/realpath.c
===================================================================
RCS file: /usr/CVSroot/XEmacs/xemacs/src/realpath.c,v
retrieving revision 1.6.2.7
diff -u -u -r1.6.2.7 realpath.c
--- realpath.c 2000/06/12 04:18:21 1.6.2.7
+++ realpath.c 2000/08/31 21:19:47
＠＠ -23,42 +23,145 ＠＠
/* Synched up with: Not in FSF. */
#include <config.h>
+#include "lisp.h"
+#include "sysfile.h"
-#include <sys/types.h>
#include <stdio.h>
#include <string.h>
-#include <errno.h>
#include <limits.h>
-#ifdef HAVE_UNISTD_H
-#include <unistd.h>
-#endif
+/* First char after start of absolute filename. */
+#define ABS_START(name) (name + ABS_LENGTH (name))
-#if defined (HAVE_SYS_PARAM_H)
-#include <sys/param.h>
-#endif
+#if defined (WIN32_NATIVE)
+/* Length of start of absolute filename. */
+# define ABS_LENGTH(name) (win32_abs_start (name))
+static int win32_abs_start (const char * name);
+/* System dependent version of readlink. */
+# define sys_readlink(name, buf, size) win32_readlink (name, buf, size)
+#else
+# ifdef CYGWIN
+# define ABS_LENGTH(name) (IS_DIRECTORY_SEP (*name) ? \
+ (IS_DIRECTORY_SEP (name[1]) ? 2 : 1) : 0)
+# define sys_readlink(name, buf, size) cygwin_readlink (name, buf, size)
+# else
+# define ABS_LENGTH(name) (IS_DIRECTORY_SEP (*name) ? 1 : 0)
+# define sys_readlink(name, buf, size) readlink (name, buf, size)
+# endif /* CYGWIN */
+#endif /* WIN32_NATIVE */
+
+#if defined (WIN32_NATIVE) || defined (CYGWIN)
+#include "syswindows.h"
+/* Emulate readlink on win32 - finds real name (i.e. correct case) of
+ a file. UNC servers and shares are lower-cased. Directories must be
+ given without trailing '/'. One day, this could read Win2K's
+ reparse points. */
+static int
+win32_readlink (const char * name, char * buf, int size)
+{
+ WIN32_FIND_DATA find_data;
+ HANDLE dir_handle = NULL;
+ int len = 0;
+ int err = 0;
+ const char* lastname;
+ int count = 0;
+ const char* tmp;
+ char* res = NULL;
+
+ assert (*name);
+
+ /* Sort of check we have a valid filename. */
+ if (strpbrk (name, "*?|<>\"") || strlen (name) >= MAX_PATH)
+ {
+ errno = EIO;
+ return -1;
+ }
+
+ /* Find start of filename */
+ lastname = name + strlen (name);
+ while (lastname > name && !IS_DIRECTORY_SEP (lastname[-1]))
+ --lastname;
+
+ /* Count slashes in unc path */
+ if (ABS_LENGTH (name) == 2)
+ for (tmp = name; *tmp; tmp++)
+ if (IS_DIRECTORY_SEP (*tmp))
+ count++;
+
+ if (count >= 2 && count < 4)
+ {
+ /* UNC server or share name: just copy lowercased name. */
+ res = find_data.cFileName;
+ for (tmp = lastname; *tmp; tmp++)
+ *res++ = tolower (*tmp);
+ *res = '\0';
+ }
+ else
+ dir_handle = FindFirstFile (name, &find_data);
+
+ if (res || dir_handle != INVALID_HANDLE_VALUE)
+ {
+ if ((len = strlen (find_data.cFileName)) < size)
+ {
+ if (strcmp (lastname, find_data.cFileName) == 0)
+ /* Signal that the name is already OK. */
+ err = EINVAL;
+ else
+ memcpy (buf, find_data.cFileName, len + 1);
+ }
+ else
+ err = ENAMETOOLONG;
+ if (!res) FindClose (dir_handle);
+ }
+ else
+ err = ENOENT;
+ errno = err;
+ return err ? -1 : len;
+}
+#endif /* WIN32_NATIVE || CYGWIN */
+
+#ifdef CYGWIN
+/* Call readlink and try to find out the correct case for the file. */
+static int
+cygwin_readlink (const char * name, char * buf, int size)
+{
+ int n = readlink (name, buf, size);
+ if (n < 0)
+ {
+ /* The file may exist, but isn't a symlink. Try to find the
+ right name. */
+ char* tmp = alloca (cygwin_posix_to_win32_path_list_buf_size (name));
+ cygwin_posix_to_win32_path_list (name, tmp);
+ n = win32_readlink (tmp, buf, size);
+ }
+ return n;
+}
+#endif /* CYGWIN */
+
#ifdef WIN32_NATIVE
-#include <direct.h>
+#ifndef ELOOP
+#define ELOOP 10062 /* = WSAELOOP in winsock.h */
#endif
-
-#include <sys/stat.h> /* for S_IFLNK */
+/* Length of start of absolute filename. */
+static int
+win32_abs_start (const char * name)
+{
+ if (isalpha (*name) && IS_DEVICE_SEP (name[1])
+ && IS_DIRECTORY_SEP (name[2]))
+ return 3;
+ else if (IS_DIRECTORY_SEP (*name))
+ return IS_DIRECTORY_SEP (name[1]) ? 2 : 1;
+ else
+ return 0;
+}
+#endif /* WIN32_NATIVE */
#if !defined (HAVE_GETCWD) && defined (HAVE_GETWD)
#undef getcwd
#define getcwd(buffer, len) getwd (buffer)
#endif
-#ifndef PATH_MAX
-# if defined (_POSIX_PATH_MAX)
-# define PATH_MAX _POSIX_PATH_MAX
-# elif defined (MAXPATHLEN)
-# define PATH_MAX MAXPATHLEN
-# else
-# define PATH_MAX 1024
-# endif
-#endif
-
#define MAX_READLINKS 32
char * xrealpath (const char *path, char resolved_path []);
＠＠ -68,78 +171,63 ＠＠
char copy_path[PATH_MAX];
char *new_path = resolved_path;
char *max_path;
-#ifdef S_IFLNK
+#if defined (S_IFLNK) || defined (WIN32_NATIVE)
int readlinks = 0;
char link_path[PATH_MAX];
int n;
+ int abslen = ABS_LENGTH (path);
#endif
/* Make a copy of the source path since we may need to modify it. */
strcpy (copy_path, path);
path = copy_path;
max_path = copy_path + PATH_MAX - 2;
+
#ifdef WIN32_NATIVE
- /*
- ** In NT we have two different cases: (1) the path name begins
- ** with a drive letter, e.g., "C:"; and (2) the path name begins
- ** with just a slash, which roots to the current drive. In the
- ** first case we are going to leave things alone, in the second
- ** case we will prepend the drive letter to the given path.
- ** Note: So far in testing, I'm only seeing case #1, even though
- ** I've tried to get the other cases to happen.
- ** August Hill, 31 Aug 1997.
- **
- ** Check for a driver letter...C:/...
- */
- if (*(path + 1) == ':')
+ /* Check for c:/... or //server/... */
+ if (abslen == 2 || abslen == 3)
{
- strncpy(new_path, path, 3);
- new_path += 3;
- path += 3;
+ strncpy (new_path, path, abslen);
+ new_path += abslen;
+ path += abslen;
}
-
- /*
- ** No drive letter, but a beginning slash? Prepend the drive
- ** letter...
- */
- else if (*path == '/')
+ /* No drive letter, but a beginning slash? Prepend drive letter. */
+ else if (abslen == 1)
{
getcwd (new_path, PATH_MAX - 1);
new_path += 3;
path++;
}
-
- /*
- ** Just a path name, prepend the current directory
- */
+ /* Just a path name, prepend the current directory */
else
{
getcwd (new_path, PATH_MAX - 1);
- new_path += strlen(new_path);
- if (new_path[-1] != '/')
- *new_path++ = '/';
+ new_path += strlen (new_path);
+ if (!IS_DIRECTORY_SEP (new_path[-1]))
+ *new_path++ = DIRECTORY_SEP;
}
-
#else
/* If it's a relative pathname use getcwd for starters. */
- if (*path != '/')
+ if (abslen == 0)
{
getcwd (new_path, PATH_MAX - 1);
- new_path += strlen(new_path);
- if (new_path[-1] != '/')
- *new_path++ = '/';
+ new_path += strlen (new_path);
+ if (!IS_DIRECTORY_SEP (new_path[-1]))
+ *new_path++ = DIRECTORY_SEP;
}
else
{
- *new_path++ = '/';
- path++;
+ /* Copy first directory sep. May have two on cygwin. */
+ strncpy (new_path, path, abslen);
+ new_path += abslen;
+ path += abslen;
}
#endif
/* Expand each slash-separated pathname component. */
while (*path != '\0')
{
/* Ignore stray "/". */
- if (*path == '/')
+ if (IS_DIRECTORY_SEP (*path))
{
path++;
continue;
＠＠ -148,7 +236,7 ＠＠
if (*path == '.')
{
/* Ignore ".". */
- if (path[1] == '\0' || path[1] == '/')
+ if (path[1] == '\0' || IS_DIRECTORY_SEP (path[1]))
{
path++;
continue;
＠＠ -156,23 +244,24 ＠＠
/* Handle ".." */
if (path[1] == '.' &&
- (path[2] == '\0' || path[2] == '/'))
+ (path[2] == '\0' || IS_DIRECTORY_SEP (path[2])))
{
path += 2;
/* Ignore ".." at root. */
- if (new_path == resolved_path + 1)
+ if (new_path == ABS_START (resolved_path))
continue;
/* Handle ".." by backing up. */
- while ((--new_path)[-1] != '/')
- ;
+ --new_path;
+ while (!IS_DIRECTORY_SEP (new_path[-1]))
+ --new_path;
continue;
}
}
/* Safely copy the next pathname component. */
- while (*path != '\0' && *path != '/')
+ while (*path != '\0' && !IS_DIRECTORY_SEP (*path))
{
if (path > max_path)
{
＠＠ -182,10 +271,10 ＠＠
*new_path++ = *path++;
}
-#ifdef S_IFLNK
+#if defined (S_IFLNK) || defined (WIN32_NATIVE)
/* See if latest pathname component is a symlink. */
*new_path = '\0';
- n = readlink (resolved_path, link_path, PATH_MAX - 1);
+ n = sys_readlink (resolved_path, link_path, PATH_MAX - 1);
if (n < 0)
{
＠＠ -204,14 +293,14 ＠＠
/* Note: readlink doesn't add the null byte. */
link_path[n] = '\0';
-
- if (*link_path == '/')
+
+ if (ABS_LENGTH (link_path) > 0)
/* Start over for an absolute symlink. */
- new_path = resolved_path;
+ new_path = resolved_path + ABS_LENGTH (link_path) - 1;
else
/* Otherwise back up over this component. */
- while (*(--new_path) != '/')
- ;
+ for (--new_path; !IS_DIRECTORY_SEP (*new_path); --new_path)
+ assert (new_path > resolved_path);
/* Safe sex check. */
if (strlen(path) + n >= PATH_MAX)
＠＠ -225,15 +314,21 ＠＠
strcpy(copy_path, link_path);
path = copy_path;
}
-#endif /* S_IFLNK */
- *new_path++ = '/';
+#endif /* S_IFLNK || WIN32_NATIVE */
+ *new_path++ = DIRECTORY_SEP;
}
/* Delete trailing slash but don't whomp a lone slash. */
- if (new_path != resolved_path + 1 && new_path[-1] == '/')
+ if (new_path != ABS_START (resolved_path) && IS_DIRECTORY_SEP (new_path[-1]))
new_path--;
/* Make sure it's null terminated. */
*new_path = '\0';
+
+#ifdef WIN32_NATIVE
+ if (ABS_LENGTH (resolved_path) == 3)
+ /* Lowercase drive letter. */
+ *resolved_path = tolower (*resolved_path);
+#endif
return resolved_path;
}
- 3 participants
- 3 comments