March 2000 - XEmacs-Beta - XEmacs Mailing List Archives

Re: crash when using W3 25 years, 7 months

Jan Vroonhof

Hrvoje Niksic <hniksic(a)iskon.hr> writes: > > ... what happens when the buffer gets killed during the loop? > > You lose. I didn't really pay much attention to that case, > considering it highly extreme and even contrived. Well lisp is allowed to be evil and extreme unfortunately. > How do we get in the situation where after-change-functions kill the > buffer? Can that really be intentional? In this case it is was not the after-change-functions themselves. It is the fact that we process events during after change functions. It is normal to delete buffers in process sentinels. Maybe we should disallow killing the buffer in an after-change-function (and enforce this somehow) and inhibit all event processing? Jan

2 participants
1 comments

Re: crash when using W3 25 years, 7 months

Jan Vroonhof

Be afraid, be very afraid.... Are we going to have to audit the whole code for stuff like this? :-( > I think that whole code should be audited for race conditions.... And its clients.. Index: process.c =================================================================== RCS file: /usr/CVSroot/XEmacs/xemacs/src/process.c,v retrieving revision 1.39.2.12 diff -u -u -r1.39.2.12 process.c --- process.c 2000/02/21 23:06:59 1.39.2.12 +++ process.c 2000/03/18 01:19:03 ＠＠ -921,12 +921,16 ＠＠ buffer_insert_raw_string (buf, chars, nbytes); #endif - Fset_marker (p->mark, make_int (BUF_PT (buf)), p->buffer); + /* buffer_insert_raw_string_1 can call lisp code! */ - MARK_MODELINE_CHANGED; + if (BUFFER_LIVE_P(buf)) + { + Fset_marker (p->mark, make_int (BUF_PT (buf)), p->buffer); + + MARK_MODELINE_CHANGED; - /* If the restriction isn't what it should be, set it. */ - if (old_begv != BUF_BEGV (buf) || old_zv != BUF_ZV (buf)) + /* If the restriction isn't what it should be, set it. */ + if (old_begv != BUF_BEGV (buf) || old_zv != BUF_ZV (buf)) { Fwiden(p->buffer); old_begv = bufpos_clip_to_bounds (BUF_BEG (buf), ＠＠ -939,14 +943,15 ＠＠ p->buffer); } - /* Handling the process output should not deactivate the mark. */ - zmacs_region_stays = old_zmacs_region_stays; - buf->read_only = old_read_only; - old_point = bufpos_clip_to_bounds (BUF_BEGV (buf), - old_point, - BUF_ZV (buf)); - BUF_SET_PT (buf, old_point); - + /* Handling the process output should not deactivate the mark. */ + zmacs_region_stays = old_zmacs_region_stays; + buf->read_only = old_read_only; + old_point = bufpos_clip_to_bounds (BUF_BEGV (buf), + old_point, + BUF_ZV (buf)); + BUF_SET_PT (buf, old_point); + } + UNGCPRO; } return nchars;

2 participants
1 comments

Intent to implement Emacs 20 find-file-literally 25 years, 7 months

Martin Buchholz

Is there some reason we did not implement this? Is this function perhaps ideologically blasphemous? Kirill, you deleted the related function find-file-binary, but it's used in hexl.el. I intend to implement find-file-literally, and use it in hexl.el. Martin

3 participants
3 comments

Progress gauge height 25 years, 7 months

Hrvoje Niksic

I've now noticed that the one thing that makes our new progress gauges, uh, less pretty than they should be, is the fact that they take up so much screen height. The raising of the modeline that gauge causes is too large a flicker to be visually acceptable. Proposal: could it be possible to place the message to the right of the gauge, instead of above it? I think it would make the gauge experience much nicer. Again, thanks Andy for doing this. I enjoy showing off the gauges to my friends.

2 participants
1 comments

Re: .xemacs/init.el 25 years, 7 months

Hrvoje Niksic

sperber(a)informatik.uni-tuebingen.de (Michael Sperber [Mr. Preprocessor]) writes: > Huh? As far as packages are concerned, it is already all over the > documentation. I believe our package installation tools > automatically put stuff there. Nothing has created a `~/.xemacs/' for me, so I assumed I didn't need to have it. And, except for the package installation tools, it's far from all over the place. I'm pretty sure many users aren't aware of the possibility or use of a `~/.xemacs/'.

3 participants
2 comments

Re: set-face-stipple 25 years, 7 months

Hrvoje Niksic

Katsumi Yamaoka <yamaoka(a)jpl.org> writes: > In the file lisp/faces.el: > > (defun set-face-stipple (face pixmap &optional frame) > "[...]" > (while (not (find-face face)) > (setq face (signal 'wrong-type-argument (list 'facep face)))) > (locate-file pixmap x-bitmap-file-path '(".xbm" "")) > ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ > why is the line here? It causes an error as follows: The intention was to search for the file over x-bitmap-file-path. However, it didn't work out because it was placed before the check whether PIXMAP is a list. Also, that line of code doesn't setq anything, meaning it doesn't really do anything. Could you try this version, which I think does everything the original function was intended to do. Kyle, please review. 2000-03-23 Hrvoje Niksic <hniksic(a)iskon.hr> * faces.el (set-face-stipple): Rewrite to correctly handle PIXMAP being a list. Actually define `stipple-pixmap-p' which is used as an error predicate. Correctly handle PIXMAP being either relative or absolute file name. --- lisp/faces.el.orig Thu Mar 23 14:39:24 2000 +++ lisp/faces.el Thu Mar 23 14:54:25 2000 ＠＠ -1607,20 +1607,28 ＠＠ in that frame; otherwise change each frame." (while (not (find-face face)) (setq face (signal 'wrong-type-argument (list 'facep face)))) - (locate-file pixmap x-bitmap-file-path '(".xbm" "")) - (while (cond ((stringp pixmap) - (unless (file-readable-p pixmap) - (setq pixmap `[xbm :file ,pixmap])) - nil) - ((and (consp pixmap) (= (length pixmap) 3)) - (setq pixmap `[xbm :data ,pixmap]) - nil) - (t t)) - (setq pixmap (signal 'wrong-type-argument - (list 'stipple-pixmap-p pixmap)))) - (while (and frame (not (framep frame))) - (setq frame (signal 'wrong-type-argument (list 'framep frame)))) - (set-face-background-pixmap face pixmap frame)) + (let (instantiator) + (while + (null + (setq instantiator + (cond ((stringp pixmap) + (let ((file (if (file-name-absolute-p pixmap) + pixmap + (locate-file pixmap x-bitmap-file-path + '(".xbm" ""))))) + (and file + `[xbm :file ,file]))) + ((and (listp pixmap) (= (length pixmap) 3)) + `[xbm :data ,pixmap]) + (t nil)))) + (flet ((stipple-pixmap-p (pixmap) + (or (stringp pixmap) + (and (listp pixmap) (= (length pixmap) 3))))) + (setq pixmap (signal 'wrong-type-argument + (list 'stipple-pixmap-p pixmap))))) + (while (and frame (not (framep frame))) + (setq frame (signal 'wrong-type-argument (list 'framep frame)))) + (set-face-background-pixmap face instantiator frame))) ;; Create the remaining standard faces now. This way, packages that we dump

2 participants
1 comments

FSF key format in edit-utils/foldout.el (PR#1090) 25 years, 7 months

Gunnar Evermann

(load-library "foldout.el) gives: Signaling: (error "Invalid (FSF Emacs) key format (see doc of define-key)" M-C-down-mouse-1) define-key(#<keymap text-mode-map 3 entries 0xa00c> [M-C-down-mouse-1] foldout-mouse-zoom) could somebody fix this (and if possible look at syncing with a later FSF version)? thanks Gunnar

2 participants
1 comments

Re: [bug] Ben: `menu' keysym behaviour. 25 years, 7 months

Karl M. Hegbloom

>>>>> "Ben" == Ben Wing <ben(a)666.com> writes: Ben> what system? Debian GNU/Linux, Potato, XFree-86 2.2.6, on an AMD K6 (x86), XEmacs 21.2 from CVS, with your large menubar stuff in it. There's a button on the lower right of my keyboard with a glyph of a menu on it with a mouse pointer arrow pointing to it. Pressing that key generates the "menu" keysym. Ben> "Karl M. Hegbloom" wrote: >> "menu runs `popup-mode-menu'" >> >> It pops up a menu, always at the mouse pointer, but the arrow keys >> don't work for selecting menu items, and ESC doesn't pop the menu >> down again.

2 participants
1 comments

Progress gauge bug 25 years, 7 months

Hrvoje Niksic

I've found a way to repeat a bug that has annoyed me with progress gauges for some time. Here's how: Make sure font-lock is turned on. Visit a large file, e.g. a ChangeLog. Then type `C-x i ChangeLog RET', i.e. insert the ChangeLog contents into itself. After this, font-lock and progress gauge seem to lock into an endless loop. If you use C-g to break out of the loop, font-lock-pre-idle-hook reports an error. This is all 100% repeatable, with -vanilla, under X (Athena) and Linux.

2 participants
1 comments

unrecognized selection-conversion type 25 years, 7 months

John A. Turner

I've been seeing this one for a long time - this is on Irix, built with gcc, as detailed here: http://www.xemacs.org/list-archives/xemacs-beta/200003/msg00021.html recipe: 1. xemacs -vanilla & 2. Options -> General -> Debug on Error 3. select text w/mouse 4. go to another virtual desktop, and soon this appears in *Backtrace*: Signaling: (error "unrecognized selection-conversion type" nil #<INTERNAL OBJECT (XEmacs bug?) (opaque, size=4) 0x1063db60>) this may very well be specific to the Irix window manager (4DWm) and its strange idea of virtual desktops... I just noticed that it happens in 20.3 and 21.1.3 (that last one running on Tru64U but displaying on Irix). -- John A. Turner, Senior Research Associate Blue Sky Studios, One South Rd, Harrison, NY 10528 http://www.blueskystudios.com/ (914) 825-8319

1 participants
0 comments

2025

2024

2023

2022

2021

2020

2019

2018

2017

2016

2015

2014

2013

2012

2011

2010

2009

2008

2007

2006

2005

2004

2003

2002

2001

2000

1999

1998

XEmacs-Beta March 2000