Re: PLEASE TEST XEmacs 21.4.9 "Informed Management" Release Candidate 1

Please test the release candidate. Use "cvs update -r r21-4-9rc1" to get the RC sources. I hope to get tarballs up tomorrow or Monday my time. For now CVS is up-to-date and so I'm announcing for those testers who use CVS. Use "cvs update -r candidate-21-4" to get the latest and greatest. Don't even expect checkouts on this tag to build; I commit what I have when I feel like it. Mostly it should work, but if it doesn't, take two aspirin and cvs update in the morning. ;-) "Release candidate" is something of a misnomer, there are a number of patches in the queue that I probably will apply, and I really hope that the syntax.c problems (aka "font-lock slows") will get fixed for 21.4.9. We'll see .... Anyway, expect at least one more candidate before 21.4.9. TESTING PROCEDURE 1. Get the sources. 2. Configure as usual, and if you're feeling especially chipper, try several variants that are new and different. 3. Use M-x build-report to report each build and any problems. If you have problems, please CC xemacs-beta. Successful builds can just go to xemacs-buildreports. 4. Test and send reports to xemacs-beta, positive or negative. WHAT TO TEST 1. I've included the work done by Andy and Nix to support vertical positioning of glyphs and some other redisplay hacking needed to support preview-latex. This is most definitely a feature, and therefore I really should not include it in 21.4. Please try preview-latex (http://preview-latex.sourceforge.net/ IIRC). The more reports and the more detail I get, the more likely I am to include it. Maybe not 21.4.9, maybe 21.4.10. Please beat hard on this one. preview-latex is really cool and I would like to support it in 21.4. 2. I've included Andy's revert-buffer optimization. The idea is to compare the buffer text to the file contents, and not redo expensive buffer analysis (eg, font-lock and semantic bovination) if the text is the same. There are reports on 21.5 that this code trashes markers and extents when you revert. This is probably fixed already, but please be on the watch for such problems, and any anomolies when you revert. Again, this code may not make it in to 21.4.9. 3. Builds on Linux PPC, OpenBSD, and any 64-bit platforms would be nice, I've added some configure tests and s&m pervection to support those. MacOS/X Darwin is already in 21.4.8, but build reports on that platform would be nice, too. The OpenBSD changes include detection of a OpenBSD weirdness where they rename libcurses as libtermcap or something like that. I think it's well-guarded by an "if [ opsys = openbsd ]", but be on the lookout for configure failures and runtime weirdness for TTY stuff. 4. XEmacs on Windows is becoming less broken. Once again I've merged all of the changes from Andy's branch, but with the large number of changes going in to the Windows code still, reports of the demise of the Windows branch are most likely severely exaggerated. :-( The one new fix not in XEmacs 21.4.7 (Windows [1]) is the format-time-string fix from Nick Pakoulin (crash on negative times). So Windows testing isn't terribly important; most of the code in the RC is already in 21.4.7 Windows [1]. However, build reports (especially Cygwin) would be nice, just to make sure the configure hacking hasn't messed anything up. 5. Miscellanea: -- C-x r p -- new key binding for replace-rectangle -- Completion lists should work better now (confusion between minibuffer and completion buffer removed) CHANGELOG DETAILS **** ChangeLog 2002-05-17 Stephen J. Turnbull <stephen(a)xemacs.org&gt; * PROBLEMS: Describe "Hannibal Lecter" crash on Linux. Reorder Linux runtime issues section. * INSTALL: Improve description of mail locking. 2002-05-10 Stephen J. Turnbull <stephen(a)xemacs.org&gt; * etc/OXYMORONS (21.4.15): New. Not a reference to Vin. * build-msw-release.sh: Synch to Andy. 2002-03-18 Gregory Steuck <greg-xemacs-patch(a)nest.cx&gt; * configure.in: Improve OpenBSD support: correct check for socklen_t, check for libtermcap is libncurses. 2002-02-28 Malcolm Purvis <malcolmpurvis(a)optushome.com.au&gt; * configure.in: Protect nocomboreloc linker arguments to avoid problem on the ppc. **** lisp/ChangeLog 2002-05-19 Stephen J. Turnbull <stephen(a)xemacs.org&gt; * files.el (revert-buffer): (revert-buffer-internal): (revert-buffer-insert-file-contents-function): Improve docstrings and comments. 2002-05-08 Andy Piper <andy(a)xemacs.org&gt; * files.el (revert-buffer): observe noconfirm behavior. 2002-04-09 Andy Piper <andy(a)xemacs.org&gt; * files.el (revert-buffer): use revert-buffer-internal if it looks like doing so will not result in any user-visible changes. (revert-buffer-internal): new function. Do the actual process of reversion and then see whether the result is any different to what we have already. If it is not then do nothing. 2001-12-11 Andy Piper <andy(a)xemacs.org&gt; * menubar.el (get-popup-menu-response): re-order so that it works on more sane/facist window systems. 2001-12-03 Andy Piper <andy(a)xemacs.org&gt; * faces.el (frob-face-property): don't infloop in face frobbing from Jan Vroonhof <jan(a)xemacs.org&gt;. 2001-11-30 Jan Vroonhof <jan(a)xemacs.org&gt; * faces.el (frob-face-property): Follow face fall-back hierarchy properly for face properties without an instance. Only do manual copy form 'default in last resort. This handles in particular the case where 'default itself has only a fall-back (which is the case by default on windows). 2001-11-24 Andy Piper <andy(a)xemacs.org&gt; * printer.el (generic-print-region): set default-frame-plist to nil while creating the printer frame so that sizes reflect the printed page. * faces.el (face-complain-about-font): Don't complain on printers. 2002-05-10 Stephen J. Turnbull <stephen(a)xemacs.org&gt; * package-net.el: Andy synch. 2001-12-16 Andy Piper <andy(a)xemacs.org&gt; * package-get.el (package-get-update-all): Make sure installed.db gets updated after updating packages. 2002-03-28 Ben Wing <ben(a)xemacs.org&gt; * msw-glyphs.el: Set glyph-contrib-p to nil on hscroll, continuation, truncation so that we don't get line jumpiness from them being larger than the line height. * msw-font-menu.el (mswindows-font-menu-font-data): Fix bug. 2001-11-30 Jan Vroonhof <jan(a)xemacs.org&gt; * font.el (font-window-system-mappings): Add mapping for Gtk (assume identical to X) 2002-02-04 Andy Piper <andy(a)xemacs.org&gt; * files.el (convert-standard-filename): Fix for short filename Peter Arius <pas(a)methodpark.de&gt; 2001-12-10 Christoph Wedler <Christoph.Wedler(a)sap.com&gt; * list-mode.el (default-choose-completion): completion-base-size should refer to local value in the completion list buffer. 2001-12-03 Didier Verna <didier(a)xemacs.org&gt; * rect.el: autoload `replace-rectangle'. * keydefs.el (global-map): bind it to `C-x r p'. **** src/ChangeLog 2002-04-25 Andy Piper <andy(a)xemacs.org&gt; * redisplay.c (create_text_block): Remove extra comment trailer. 2002-04-24 Andy Piper <andy(a)xemacs.org&gt; * redisplay.c (create_text_block): Don't actually add propagation data if the line ends after we have added a glyph. 2002-04-22 Andy Piper <andy(a)xemacs.org&gt; * extents.c (extent_fragment_update): check for glyphs we have previously displayed. * extents.c (print_extent_1): warning removal. * extents.h: change prototype. * redisplay-output.c (redisplay_normalize_glyph_area): calculate widths correctly for wide glyphs. * redisplay.c (position_redisplay_data_type): add end_glyph_width. * redisplay.c (prop_type): add PROP_GLYPH. * redisplay.c (struct prop_block): add glyph type * redisplay.c (add_glyph_rune): when adding part of a glyph add it to the propagation data. * redisplay.c (create_text_block): if there is a glyph in the propagation data use it to salt extent_fragment_update. * redisplay.c (create_string_text_block): ditto. 2002-04-13 Nix <nix(a)esperi.demon.co.uk&gt; * redisplay.h (struct rune): Add ascent, descent, and yoffset fields. * redisplay-output.c (compare_runes): Compare them. * redisplay.c: Update copyright date. * redisplay.c (pos_data): Add need_baseline_computation field. * redisplay.c (add_glyph_rune): Update ascent, descent, and need_baseline_computation; zero yoffset. Set max_pixmap_height for all pixmaps, not just automatically positioned ones. * redisplay.c (calculate_yoffset): New, compute yoffset values. * redisplay.c (calculate_baseline): New, compute textual baseline. * redisplay.c (add_glyph_rune): Call them. * redisplay.c (create_text_block): Likewise. * redisplay.c (create_overlay_glyph_block): Likewise. * redisplay.c (add_margin_runes): Likewise. * redisplay.c (create_string_text_block): Likewise. Fix tabdamage. * redisplay.h: (redisplay_calculate_display_boxes): Change prototype. * redisplay-output.c (redisplay_calculate_display_boxes): Use yoffset. * redisplay-msw.c (mswindows_output_blank): Pass 0 as yoffset. * redisplay-msw.c (mswindows_output_string): Likewise. * redisplay-msw.c (mswindows_output_display_block): Pass yoffset. * redisplay-gtk.c (gtk_output_display_block): Likewise. * redisplay-x.c (x_output_display_block): Likewise. 2002-03-28 Ben Wing <ben(a)xemacs.org&gt; * redisplay.c: Fixed bug in redisplay w.r.t. hscroll/truncation/continuation glyphs causing jumping up and down of the lines, since they're bigger than the line size. (It was seen most obviously when there's a horizontal scroll bar, e.g. do C-h a glyph or something like that.) The problem was that the glyph-contrib-p setting on glyphs was ignored even if it was set properly, which it wasn't until now. 2002-04-24 Andy Piper <andy(a)xemacs.org&gt; * lisp.h (Dynarr_end): Fix definition. 2002-02-06 Adrian Aichner <adrian(a)xemacs.org&gt; * redisplay.c (mark_redisplay): Remove call to update_frame_window_mirror. 2001-11-15 Andy Piper <andy(a)xemacs.org&gt; * win32.c (Fmswindows_shell_execute): fix handling of URL's under cygwin (again). 2002-02-13 Andy Piper <andy(a)xemacs.org&gt; * event-msw.c (mswindows_wnd_proc): only mark the frame visible if we did in fact enqueue the XM_MAPFRAME event. 2002-01-15 Adrian Aichner <adrian(a)xemacs.org&gt; * event-msw.c (mswindows_wnd_proc): Add handling of WM_SHOWWINDOW to fix problem switching between virtual desktops under virtuawin virtual window manager. 2001-07-30 Adrian Aichner <adrian(a)xemacs.org&gt; * event-msw.c: Typo fix. * event-msw.c (mswindows_wnd_proc): Set FRAME_VISIBLE_P after magic XM_MAPFRAME event has been sent. 2001-11-23 Andy Piper <andy(a)xemacs.org&gt; * event-msw.c (mswindows_wnd_proc): Don't pump mousewheel events. 2001-11-21 Andy Piper <andy(a)xemacs.org&gt; * scrollbar-msw.c (mswindows_handle_mousewheel_event): cope with mouse events outside the frame. 2002-03-20 Andy Piper <andy(a)xemacs.org&gt; * menubar-msw.c (mswindows_popup_menu): warning removal. * dialog-msw.c (dialog_popped_down): ditto. 2001-12-11 Andy Piper <andy(a)xemacs.org&gt; * dialog-msw.c (dialog_popped_down): new function. unset popup_up_p. * dialog-msw.c (mswindows_make_dialog_box_internal): set popup_up_p. * menubar-msw.c (unsafe_handle_wm_initmenupopup_1): ditto. * menubar-msw.c (mswindows_handle_wm_command): ditto. * menubar-msw.c (mswindows_popup_menu): ditto. 2001-10-29 Andy Piper <andy(a)xemacs.org&gt; * dialog-msw.c (handle_directory_dialog_box): quit if the user cancels. 2002-01-03 Andy Piper <andy(a)xemacs.org&gt; * realpath.c (ABS_LENGTH): dtrt for cygwin systems using drive letters. (xrealpath): ditto. 2002-03-29 Jonathan Harris <jonathan(a)xemacs.org&gt; * device-msw.c (plist_get_margin): Add arg specifying mm or inches * device-msw.c (plist_set_margin): Fix multiplicand used for mm * device-msw.c (mswindows_handle_page_setup_dialog_box): Detect and handle case where machine is set up for metric units 2002-04-02 Andy Piper <andy(a)xemacs.org&gt; * dired-msw.c (mswindows_get_files): * nt.c (mswindows_stat): SetErrorMode() so that file errors are completely handled by XEmacs. Suggested by Thomas Vogler <mail(a)thomas-vogler.de&gt;. 2001-11-14 Andy Piper <andy(a)xemacs.org&gt; * nt.c (REG_ROOT): change registry key to XEmacs. 2002-04-27 Andy Piper <andy(a)xemacs.org&gt; * glyphs-msw.c (mswindows_widget_instantiate): remove dead-code. 2002-04-26 Andy Piper <andy(a)xemacs.org&gt; * glyphs-msw.c (mswindows_map_subwindow): observe :initial-focus behavior. 2002-03-14 Mike Alexander <mta(a)arbortext.com&gt; * event-msw.c (mswindows_unwait_process): New, remove process from wait list * process-nt.c (nt_finalize_process_data): Call mswindows_unwait_process * console-msw.h: Declare mswindows_unwait_process 2002-05-14 Stephen J. Turnbull <stephen(a)xemacs.org&gt; * editfns.c (Fdecode_time): (Fformat_time_string): Check for invalid time. Thanks to Nick Pakoulin <npak(a)ispras.ru&gt;. (make_time): Warning elimination (change arg name). 2001-11-24 Andy Piper <andy(a)xemacs.org&gt; * window.c (Fsplit_window): Doc return type. 2002-04-01 Andy Piper <andy(a)xemacs.org&gt; * emacs.c (Fkill_emacs): Only output message box in interactive mode. 2002-04-26 Andy Piper <andy(a)xemacs.org&gt; * config.h.in: pull in 21.5.x change to stop alloca warnings under cygwin. 2002-03-18 Gregory Steuck <greg-xemacs-patch(a)nest.cx&gt; * unexelf.c: Use ELFSIZE or define from _LP64 to indicate a 64 bit platform. **** man/ChangeLog 2002-05-11 Adrian Aichner <adrian(a)xemacs.org&gt; * xemacs-faq.texi (Top): Add Q2.0.15 under "Installation and Trouble Shooting". * xemacs-faq.texi (Installation): Add Q2.0.15 to Installation menu. * xemacs-faq.texi (Q2.0.14): Fix unnumberedsubsec argument. * xemacs-faq.texi (Q2.0.15): New. **** lib-src/ChangeLog 2002-05-10 Stephen J. Turnbull <stephen(a)xemacs.org&gt; * Makefile.in.in: Add winclient targets (Andy synch). -- Institute of Policy and Planning Sciences http://turnbull.sk.tsukuba.ac.jp University of Tsukuba Tennodai 1-1-1 Tsukuba 305-8573 JAPAN My nostalgia for Icon makes me forget about any of the bad things. I don't have much nostalgia for Perl, so its faults I remember. Scott Gilbert c.l.py