May 2006 - XEmacs-Patches - XEmacs Mailing List Archives

[PATCH] Fix a hang in read-from-minibuffer if the X display disappears 17 years, 11 months

Aidan Kehoe

Like many users of XFree86, I can make my X server crash by evaluating this Lisp code: (read-from-minibuffer "\t") I also use XEmacs as a server, connecting to it with gnuclient in my X init scripts. I run this in /etc/rc.local on startup: ( . /home/aidan/.shrc && /usr/pkg/bin/sudo -u aidan /usr/pkg/bin/screen -d -m /usr/local/bin/xemacs ) This should make it possible to resume my XEmacs session after such an X server crash, since it provides a TTY frame, and since the x_IO_error_handler only aborts if there is no other existing XEmacs frame. However, the x_IO_error_handler doesn’t immediately delete the X device, it queues an event to do so, and then throws to top level. If this happens during read-from-minibuffer, read-from-minibuffer tries to restore the old window and frame configuration, before the “delete the X device” event is processed. This calls X11 code, and calling X11 code with a no-longer-valid Display argument in this case hangs. The patch checks whether the device is being deleted (and whether the frame is live, for good measure) and avoids calling X code if so. Tangentally, looking at the X11 code, there are a few places that currently call ABORT() that should be doing roughly what x_IO_error_handler does instead. Does anyone in principle object to my reworking them? src/ChangeLog addition: 2006-05-07 Aidan Kehoe <kehoea(a)parhasard.net> * frame-x.c (x_set_frame_properties): Check that the frame is live and that the device is not being deleted before attempting anything. Avoids calling X code if x_IO_error_handler has seen a broken pipe. XEmacs Trunk source patch: Diff command: cvs -q diff -u Files affected: src/frame-x.c Index: src/frame-x.c =================================================================== RCS file: /pack/xemacscvs/XEmacs/xemacs/src/frame-x.c,v retrieving revision 1.73 diff -u -u -r1.73 frame-x.c --- src/frame-x.c 2006/05/06 17:56:00 1.73 +++ src/frame-x.c 2006/05/07 11:42:24 ＠＠ -745,7 +745,14 ＠＠ Bool y_position_specified_p = False; Bool internal_border_width_specified = False; Lisp_Object tail; - Widget w = FRAME_X_TEXT_WIDGET (f); + Widget w; + + /* We can be called after the X IO error handler has seen a broken pipe on + the relevant display. Don't do anything in that case. */ + if (!FRAME_LIVE_P (f) || DEVICE_X_BEING_DELETED (XDEVICE (FRAME_DEVICE (f)))) + return; + + w = FRAME_X_TEXT_WIDGET (f); for (tail = plist; !NILP (tail); tail = Fcdr (Fcdr (tail))) { -- Aidan Kehoe, http://www.parhasard.net/

2 participants
2 comments

Re: [Q] JIT Unicode code point support, and UTF-8 in compound text 17 years, 11 months

Aidan Kehoe

Ar an dara lá is fiche de mí Bealtaine, scríobh Ben Wing: > > I've done work and it conflicts with what you've done. If, in the > > medium term, you're going to commit what you've done, then it's not > > constructive of me to do further work on it. If you're not, then it is, > > because I'll be better able to maintain and fix bugs in the > > implementation if I wrote it. > > i can't guarantee anything; i don't even quite remember the state that the > code was last in, or how much work is needed, but i suspect a fair amount. > > what does "medium term" mean to you? Some point in 2006, to a first approximation :-) . > > Enjoy Hong Kong! > > thanks ... i'll actually be in various parts of china; not sure which ones, > quite yet. I hear that the Central Asian parts of it are changing really quickly, from frontier towns populated by Uyghurs and other Turkic folk to densely-Han-populated stereotypical China. I’m sure seeing that happening would be fascinating. -- Aidan Kehoe, http://www.parhasard.net/

1 participants
0 comments

Re: [Q] JIT Unicode code point support, and UTF-8 in compound text 17 years, 11 months

Aidan Kehoe

Ar an dara lá is fiche de mí Bealtaine, scríobh Ben Wing: > also, can anyone make any comments about good mailers? google's gmail is > wonderful about threading, which i find makes it much easier to have xemacs > mixed with other mail ... but i'd rather use a local mailer, e.g. > thunderbird. Why, if you don’t mind me asking? Gmail even auto-saves your drafts reliably, the lack of which is the main reason I hate non-Emacs mailers, the UI is as responsive as anything local, the accessibility to it can’t be beaten, you can backup your stuff locally using POP3, you can send and receive from it using non-Gmail addresses. If you’re not going to have Emacs keybindings anyway--which are the reason I’m still using VM and not Gmail--I don’t see the argument against it, to be honest. > does anyone know of mail programs for windows that are able to > *really* handle threads intelligently, similar to the way that gmail does > it? Jamie Zawinski had good things to say about Netscape 3 :-) . Its UTF-8 support sucks, though. > anyway, aidan, if you are interested in doing some work, we should talk to > make sure things don't conflict. I’ve done work and it conflicts with what you’ve done. If, in the medium term, you’re going to commit what you’ve done, then it’s not constructive of me to do further work on it. If you’re not, then it is, because I’ll be better able to maintain and fix bugs in the implementation if I wrote it. > right now i'm in hong kong so it's > unlikely i can do much anything before i get back to the u.s. (probably jul > 3). but if you can relay your plans i'll be able to make some comments; i > check my mail every few days. My short-term plans are currently something like “catch up with my XEmacs backlog whenever I don’t have the stomach for Italo Calvino in German.” I don’t anticipate getting to anything as big as changing the internal character encoding of the editor soon. Enjoy Hong Kong! -- Aidan Kehoe, http://www.parhasard.net/

2 participants
1 comments

[COMMIT] Incorporate Fabrice's fix of my buffer overrun bug 17 years, 11 months

Aidan Kehoe

APPROVE COMMIT NOTE: This patch has been committed. src/ChangeLog addition: 2006-05-21 Aidan Kehoe <kehoea(a)parhasard.net> * doc.c (extract_object_file_name): * doc.c (unparesseuxify_doc_string): Leave sufficient space for the '\0' sentinel when reading into the buffer. The bug in unparesseuxify_doc_string had been there for ten years at least, but it was Fabrice Popineau's investigation of the code on the same model in extract_object_file_name that provoked its discovery. Thank you Fabrice! XEmacs Trunk source patch: Diff command: cvs -q diff -u Files affected: src/doc.c Index: src/doc.c =================================================================== RCS file: /pack/xemacscvs/XEmacs/xemacs/src/doc.c,v retrieving revision 1.37 diff -u -u -r1.37 doc.c --- src/doc.c 2006/05/07 14:20:36 1.37 +++ src/doc.c 2006/05/21 18:30:25 ＠＠ -49,7 +49,7 ＠＠ { Ibyte buf[DOC_MAX_FILENAME_LENGTH+1]; Ibyte *buffer = buf; - int buffer_size = sizeof (buf), space_left; + int buffer_size = sizeof (buf) - 1, space_left; Ibyte *from, *to; REGISTER Ibyte *p = buffer; Lisp_Object return_me; ＠＠ -59,8 +59,8 ＠＠ GCPRO2 (fdstream, instream); - position = doc_pos > DOC_MAX_FILENAME_LENGTH ? - doc_pos - DOC_MAX_FILENAME_LENGTH : 0; + position = doc_pos > buffer_size ? + doc_pos - buffer_size : 0; if (0 > lseek (fd, position, 0)) { ＠＠ -168,7 +168,7 ＠＠ { Ibyte buf[512 * 32 + 1]; Ibyte *buffer = buf; - int buffer_size = sizeof (buf); + int buffer_size = sizeof (buf) - 1; Ibyte *from, *to; REGISTER Ibyte *p = buffer; Lisp_Object return_me; ＠＠ -215,13 +215,15 ＠＠ if (space_left == 0) { Ibyte *old_buffer = buffer; + buffer_size *= 2; + if (buffer == buf) { - buffer = xnew_ibytes (buffer_size *= 2); + buffer = xnew_ibytes (buffer_size + 1); memcpy (buffer, old_buffer, p - old_buffer); } else - XREALLOC_ARRAY (buffer, Ibyte, buffer_size *= 2); + XREALLOC_ARRAY (buffer, Ibyte, buffer_size + 1); p += buffer - old_buffer; space_left = buffer_size - (p - buffer); } -- Aidan Kehoe, http://www.parhasard.net/

1 participants
0 comments

[PATCH (pkgs)] Haskell mode comint fixes 17 years, 11 months

Jerry James

PATCH packages Stefan, I am submitting the following patch to xemacs-patches for inclusion in the packages distributed by the XEmacs project. Neither Emacs nor XEmacs uses comint-input-sentinel anymore, it having been replaced with comint-input-filter-functions. Would you consider this patch for inclusion in the Haskell mode distribution? Thank you. xemacs-packages/haskell-mode/ChangeLog addition: 2006-05-13 Jerry James <james(a)xemacs.org> * haskell-ghci.el (haskell-ghci-start-process): Replace use of comint-input-sentinel with comint-input-filter-functions. * haskell-hugs.el (haskell-hugs-start-process): Ditto. packages source patch: Diff command: cvs -q diff -uN Files affected: xemacs-packages/haskell-mode/haskell-hugs.el xemacs-packages/haskell-mode/haskell-ghci.el Index: xemacs-packages/haskell-mode/haskell-ghci.el =================================================================== RCS file: /pack/xemacscvs/XEmacs/packages/xemacs-packages/haskell-mode/haskell-ghci.el,v retrieving revision 1.4 diff -d -u -r1.4 haskell-ghci.el --- xemacs-packages/haskell-mode/haskell-ghci.el 2006/03/15 05:18:23 1.4 +++ xemacs-packages/haskell-mode/haskell-ghci.el 2006/05/13 18:18:08 ＠＠ -183,7 +183,7 ＠＠ ;; Track directory changes using the `:cd' command. (setq shell-cd-regexp ":cd") (setq shell-dirtrackp t) - (setq comint-input-sentinel 'shell-directory-tracker) + (add-hook 'comint-input-filter-functions 'shell-directory-tracker) ;; GHCi prompt should be of the form `ModuleName> '. (setq comint-prompt-regexp "^\\*?[A-Z][\\._a-zA-Z0-9]*\\( \\*?[A-Z][\\._a-zA-Z0-9]*\\)*> ") Index: xemacs-packages/haskell-mode/haskell-hugs.el =================================================================== RCS file: /pack/xemacscvs/XEmacs/packages/xemacs-packages/haskell-mode/haskell-hugs.el,v retrieving revision 1.3 diff -d -u -r1.3 haskell-hugs.el --- xemacs-packages/haskell-mode/haskell-hugs.el 2006/03/15 05:18:23 1.3 +++ xemacs-packages/haskell-mode/haskell-hugs.el 2006/05/13 18:18:09 ＠＠ -181,7 +181,7 ＠＠ (make-local-variable 'shell-dirtrackp) (setq shell-cd-regexp ":cd") (setq shell-dirtrackp t) - (setq comint-input-sentinel 'shell-directory-tracker) + (add-hook 'comint-input-filter-functions 'shell-directory-tracker) ; ? or module name in Hugs 1.4 (setq comint-prompt-regexp "^\? \\|^[A-Z][_a-zA-Z0-9\.]*> ") ;; comint's history syntax conflicts with Hugs syntax, eg. !! -- Jerry James, Assistant Professor james(a)xemacs.org Computer Science Department http://www.cs.usu.edu/~jerry/ Utah State University

2 participants
2 comments

[PATCH] Unbreak w3.el for 21.5. 17 years, 11 months

Malcolm Purvis

This patch restores the constant font-running-xemacs to font.el in 21.5 core, which was removed in a cleanup of the file back in November. Without this w3.el will not work (M-x w3 will fail trying to load the home page). Is says something about the popularity of w3 on 21.5 if such as glaring bug takes 5 months to be noticed! Malcolm lisp/ChangeLog addition: 2006-05-06 Malcolm Purvis <malcolmp(a)xemacs.org> * font.el (font-running-xemacs): Restore definition. xemacs-21.5-micro-patches source patch: Diff command: cvs -q diff -u Files affected: lisp/font.el Index: lisp/font.el =================================================================== RCS file: /pack/xemacscvs/XEmacs/xemacs/lisp/font.el,v retrieving revision 1.20 diff -u -r1.20 font.el --- lisp/font.el 2006/04/25 14:01:53 1.20 +++ lisp/font.el 2006/05/06 12:24:50 ＠＠ -102,6 +102,9 ＠＠ ;;; Lots of variables / keywords for use later in the program ;;; Not much should need to be modified ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; +(defconst font-running-xemacs (string-match "XEmacs" (emacs-version)) + "Whether we are running in XEmacs or not.") + ;; #### These aren't window system mappings (defconst font-window-system-mappings '((x . (x-font-create-name x-font-create-object)) -- Malcolm Purvis <malcolmp(a)xemacs.org>

5 participants
8 comments

[21.5R21.4] [PATCH] Mention ssh warnings in PROBLEMS. 17 years, 11 months

Malcolm Purvis

21.5 RECOMMEND 21.4 Skip Montanaro's recent problems with X11 errors being generated when using X forwarding over ssh is not the first time that this problem's been raised, so I think that it's a candidate for the PROBLEMS file. ChangeLog addition: 2006-04-27 Malcolm Purvis <malcolmp(a)xemacs.org> * PROBLEMS: Add entry concerning X errors when using ssh. xemacs-21.5-micro-patches source patch: Diff command: cvs -q diff -u Files affected: PROBLEMS Index: PROBLEMS =================================================================== RCS file: /pack/xemacscvs/XEmacs/xemacs/PROBLEMS,v retrieving revision 1.71 diff -u -r1.71 PROBLEMS --- PROBLEMS 2005/11/16 12:12:57 1.71 +++ PROBLEMS 2006/04/27 12:59:40 ＠＠ -1005,6 +1005,12 ＠＠ too old. There is a copy of the MIT X11R6 XKeysymDB file in the emacs `etc' directory. Try using that one. +*** Lots of warnings generated when displaying via ssh X forwarding. + +If you are seeing a significant number of X11 warnings (in particular +BadWindow errors) when using XEmacs via ssh X forwarding (ssh -X) try +using a TrustedX11 connection (ssh -Y) instead. + *** My X resources used to work, and now some of them are being ignored. Check the resources in .../etc/Emacs.ad (which is the same as the file -- Malcolm Purvis <malcolmp(a)xemacs.org>

2 participants
3 comments

[PATCH (pkgs)] tcl.el comint function binding test 17 years, 11 months

Jerry James

The original code isn't necessarily broken; this just seems safer. xemacs-packages/prog-modes/ChangeLog addition: 2006-05-13 Jerry James <james(a)xemacs.org> * tcl.el (tcl-filter): Predicate use of comint-output-filter on a function binding instead of on a version number. packages source patch: Diff command: cvs -q diff -uN Files affected: xemacs-packages/prog-modes/tcl.el Index: xemacs-packages/prog-modes/tcl.el =================================================================== RCS file: /pack/xemacscvs/XEmacs/packages/xemacs-packages/prog-modes/tcl.el,v retrieving revision 1.10 diff -d -u -r1.10 tcl.el --- xemacs-packages/prog-modes/tcl.el 2003/01/06 18:31:23 1.10 +++ xemacs-packages/prog-modes/tcl.el 2006/05/13 22:27:33 ＠＠ -1282,7 +1282,7 ＠＠ (progn (delete-region (point) inferior-tcl-delete-prompt-marker) (set-marker inferior-tcl-delete-prompt-marker nil))))) - (if tcl-using-emacs-19 + (if (fboundp 'comint-output-filter) (comint-output-filter proc string) (funcall comint-output-filter string))) -- Jerry James, Assistant Professor james(a)xemacs.org Computer Science Department http://www.cs.usu.edu/~jerry/ Utah State University

1 participants
1 comments

[PATCH (pkgs)] os-utils comint fix 17 years, 11 months

Jerry James

PATCH packages This is the 5th patch in the comint-fixing series. xemacs-packages/os-utils/ChangeLog addition: 2006-05-13 Jerry James <james(a)xemacs.org> * inf-lisp.el (inferior-lisp-mode): Replace use of defunct comint-input-sentinel with comint-input-filter-functions. packages source patch: Diff command: cvs -q diff -uN Files affected: xemacs-packages/os-utils/inf-lisp.el Index: xemacs-packages/os-utils/inf-lisp.el =================================================================== RCS file: /pack/xemacscvs/XEmacs/packages/xemacs-packages/os-utils/inf-lisp.el,v retrieving revision 1.2 diff -d -u -r1.2 inf-lisp.el --- xemacs-packages/os-utils/inf-lisp.el 2006/02/26 14:31:40 1.2 +++ xemacs-packages/os-utils/inf-lisp.el 2006/05/13 22:23:19 ＠＠ -253,7 +253,7 ＠＠ (use-local-map inferior-lisp-mode-map) ;c-c c-k for "kompile" file (setq comint-get-old-input (function lisp-get-old-input)) (setq comint-input-filter (function lisp-input-filter)) - (setq comint-input-sentinel 'ignore) + (setq comint-input-filter-functions nil) (run-hooks 'inferior-lisp-mode-hook)) (defun lisp-get-old-input () -- Jerry James, Assistant Professor james(a)xemacs.org Computer Science Department http://www.cs.usu.edu/~jerry/ Utah State University

1 participants
1 comments

[PATCH (pkgs)] Ocaml package comint fix 17 years, 11 months

Jerry James

PATCH packages This is the 4th patch in the comint-fixing series of patches. xemacs-packages/ocaml/ChangeLog addition: 2006-05-13 Jerry James <james(a)xemacs.org> * camldebug.el (camldebug-mode): Replace use of the defunct comint-input-sentinel with comint-input-filter-functions. packages source patch: Diff command: cvs -q diff -uN Files affected: xemacs-packages/ocaml/camldebug.el Index: xemacs-packages/ocaml/camldebug.el =================================================================== RCS file: /pack/xemacscvs/XEmacs/packages/xemacs-packages/ocaml/camldebug.el,v retrieving revision 1.1 diff -d -u -r1.1 camldebug.el --- xemacs-packages/ocaml/camldebug.el 2002/07/31 21:03:34 1.1 +++ xemacs-packages/ocaml/camldebug.el 2006/05/13 22:19:21 ＠＠ -115,7 +115,7 ＠＠ camldebug-last-frame-displayed-p t) (make-local-variable 'shell-dirtrackp) (setq shell-dirtrackp t) - (setq comint-input-sentinel 'shell-directory-tracker)) + (add-hook 'comint-input-filter-functions 'shell-directory-tracker)) ;;; Keymaps. -- Jerry James, Assistant Professor james(a)xemacs.org Computer Science Department http://www.cs.usu.edu/~jerry/ Utah State University

1 participants
1 comments

2024

2023

2022

2021

2020

2019

2018

2017

2016

2015

2014

2013

2012

2011

2010

2009

2008

2007

2006

2005

2004

2003

XEmacs-Patches May 2006