On Sunday, 29 September 2002, Andy Piper wrote: Strange things are happening with this build. I haven't changed any setting. Running M-x compile RET make realclean pdf RET gives the following: cd c:/kitty/doc/papers/kitty-project/ make realclean pdf The syntax of the command is incorrect. Compilation exited abnormally with code 1 at Mon Sep 30 01:28:48 I don't know which syntax is incorrect. The command runs fine in a shell. I don't know what else to report as I am kind of stumped by this. Also I noticed that when completing filenames using C-x C-f foo TAB, gives a [No match] error and wouldn't show me the completions namely foo.{cpp,h}. I have not touched any setting in my init.el. One more question. How do I revert to a previous version of Windows build ? The installer removed the previous version (21.4.9-3) i.e C:\Program Files\XEmacs\XEmacs-21.4.9 instead of dropping the XEmacs-21.4.10 alongside the old one. Is there some setting that I missed ? Now the setup that I downloaded from the Net doesn't let me choose 21.4.9-3. Next time I will be careful not to install under the same path:-( -kitty. -- Krishnakumar B <kitty at cs dot wustl dot edu> Distributed Object Computing Laboratory, Washington University in St.Louis

This sounds like some bugs that I was seeing before the original syntax table revert - maybe Stephen made the same mistakes I did :) Right now probably the only way is to hand edit the setup.ini. The old versions are on the ftp site, they just don't show up in the netinstaller. Its on my list to write a script that does a merge with previous versions but I haven't got around to it as yet. andy

I've put up some new kits with the dired change reverted. Please try and see if it fixes the problem. andy

If this is 21.4.10rc2, I think setup.ini is broken in it (I think the version should be "21.4.10", not "21.4.10rc2"). Because of this, it won't install from scratch. -- David Masterson dsm(a)rawbw.com

Andy> Stephen, for some reason I don't see this patch in Andy> candidate-21-4. This is the only difference between the Andy> windows and main branches currently. It does now, thanks. It's sitting in my queue because I was going to try to think of a way to automate Adrian's test, but since it depends on the Windows task manager, that's not my job, I think. :-) -- 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

Also failed in Solaris 2.6 and RH7.3. I added the function definition of `retry_close' from sysdep.c in the cvs trunk to the end of sysdep.c in the candidate-21-4 branch, and then succeeded. -- Katsumi Yamaoka <yamaoka(a)jpl.org&gt;

Stephen> Mike's bad. I *did* say something to the effect of "this *should* work", didn't I? -- Cheers =8-} Mike Friede, Völkerverständigung und überhaupt blabla

rendhalver(a)xemacs.org (Rendhalver [Peter Brown]) writes: Hi! This is very interesting. My installation, Oort Gnus v0.08 (CVS from yesterday) and candidate-21-4 as described below, dies on startup with the attached trace. What version of Gnus are you using? Thanks, norbert. --

"Stephen J. Turnbull" <stephen(a)xemacs.org&gt; writes: Hmm, here's a couple of numbers, but I'm pretty sure, you would like to see more ... (gdb) pobj coding_system $22 = (struct lstream *) 0x8a65b80 $23 = {header = {lheader = {type = 15, mark = 0, c_readonly = 0, lisp_readonly = 0}, next = 0x92de880, uid = 25466, free = 0}, imp = 0x81ea040, buffering = LSTREAM_BLOCK_BUFFERED, buffering_size = 512, in_buffer = 0x0, in_buffer_size = 512, in_buffer_current = 0, in_buffer_ind = 0, out_buffer = 0x0, out_buffer_size = 0, out_buffer_ind = 0, unget_buffer = 0x0, unget_buffer_size = 0, unget_buffer_ind = 0, byte_count = 0, flags = 2, data = {{...}}} I can't get through to the data member, at least I don't know a nifty function which would tell me :-) Simon might know something about. He was pretty actively committing during the time this crash first occurred. Seems to: 'mule-ucs-autoloads' is listed in features. norbert.

Simon Josefsson <jas(a)extundo.com&gt; writes: I now read the XEmacs backtrace, and think that these two CCL thingies are unrelated, I don't think you were verifying a OpenPGP mail when it crashed. So I don't know what caused it, Gnus doesn't use CCL coding systems at all AFAIK.

Simon Josefsson <jas(a)extundo.com&gt; writes: Hi! No, it's not pgg related. The crash is older than the pgg's introduction into the code-base. norbert.

R> i also didnt see the make check failures mentioned below maybe R> its another one of those FreeBSDisms I don't think you're running make check (it is _not_ part of make beta, you need to do it explicitly). I don't see any cd ./src && make check which should pass the M-x build-report filter. -- 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

rendhalver(a)xemacs.org (Rendhalver [Peter Brown]) writes: I can second this. If you use the BSD make utility, you don't get the stuff installed that resides in lib-src. I simply can't understand why, though. I think, a make clean all would do, but most of the time I'm using gmake these days. norbert.

I would like to see 2002-09-20 Mike Sperber <mike(a)xemacs.org&gt; (This partially gets back my patch from 2002-07-20 without the bug.) * process-unix.c (child_setup): Remove Ben's comment noting confusion over the file descriptors >= 3: close_process_descs() doesn't close all. Fix fd range to go up to MAXDESC-1 instead of 64. (unix_create_process): Call begin_dont_check_for_quit to inhibit unwanted interaction (and thus breaking of X event synchronicity) in the child. from http://list-archive.xemacs.org/xemacs-patches/200209/msg00087.html in there if Steve Y doesn't object. -- Cheers =8-} Mike Friede, Völkerverständigung und überhaupt blabla

|--==> "SJT" == Stephen J Turnbull <stephen(a)xemacs.org&gt; writes: ms> I would like to see ms> http://list-archive.xemacs.org/xemacs-patches/200209/msg00087.html ms> in there if Steve Y doesn't object. SJT> Right. I'm planning an RC2 (maybe tomorrow, maybe not until the SJT> weekend) and it'll be in that. But if I don't see positive feedback SJT> (especially from Steve Y) it may come back out. + feedback :-) -- |---<Steve Youngs>---------------<GnuPG KeyID: 10D5C9C5>---| | XEmacs - It's not just an editor. | | It's a way of life. | |------------------------------------&lt;youngs(a)xemacs.org&gt;---|

Please test Release Candidate 2 of XEmacs 21.4.10 ================================================= There are only a few specialized changes from RC1. If they are not relevant to your usage, there is no need to build RC2. Continue testing RC1. Report your builds via M-x build-report and any bugs or interesting behavior (including improvements ;-) via M-x report-emacs-bug. o The notable fix is for the XEmacs processes that used to linger after logging out from gdm (and some other session managers), consuming 100% of CPU. Test reports here would be appreciated. o Other changes are detailed below. Test reports on Candidate 1: o Some users have reported substantial speedups with the changes to the code handling syntax-table text properties. I'd still like to hear reports from *** cperl-mode *** users, as the group most likely to be adversely affected. o There are reports of problems _on Windows_ with `file-truename' (native Windows builds), or if `directory-sep-char' is explicitly set to `/'. These often seem to manifest using dired and file name completion. They are believed to be unrelated to the changes between 21.4.9 and 21.4.10. More information about these problems would be very helpful. *** Note *** Due to a glitch in my release script that I inadvertantly fixed, the prerelease location has returned to ftp://ftp.xemacs.org/pub/xemacs/xemacs-21.4/pretest/ (it was .../tmp/ for Candidate 1). The rc1 versions are still available, and have been moved there for consistency. The tag for checkout from CVS is `r21-4-10rc2'. Brief list of changes Candidate 1 -> Candidate 2 ------------------------------------------------ -- Fix: (X11) CPU-eating XEmacs process at gdm logout -- Ben Sigelman -- Fix: Freeze in certain modal dialogs -- Andy Piper -- Fix: (MS Windows) Handle leak -- Mike Alexander -- Fix: (Unix) Subprocess problems -- Mike Sperber, Stephen Turnbull -- Fix: (X11) Crash in external widget -- I. N. Golubev -- Update: Package sites and Texi docs -- Ville Skyttä Change Logs ----------- lisp/ChangeLog: 2002-09-22 Ville Skyttä <ville.skytta(a)xemacs.org&gt; * package-get.el (package-get-download-sites): Bring sites list up to date, prefer xx.xemacs.org addresses. 2002-09-27 Andy Piper <andy(a)xemacs.org&gt; * dialog.el (make-dialog-box): for general dialogs only delete-frame-hook will be run. man/ChangeLog: 2002-09-22 Ville Skyttä <ville.skytta(a)xemacs.org&gt; * xemacs/packages.texi (Available Packages): Add fortran-modes, perl-modes, psgml-dtds, python-modes and ruby-modes. Some consistency tweaks. src/ChangeLog: 2002-08-30 Mike Alexander <mta(a)arbortext.com&gt; * event-msw.c (winsock_closer): Use closesocket, not CloseHandle to close a socket. 2002-10-03 Stephen J. Turnbull <stephen(a)xemacs.org&gt; * callproc.c (child_setup): 21.4 doesn't have retry_close. 2002-09-20 Mike Sperber <mike(a)xemacs.org&gt; (This partially gets back my patch from 2002-07-20 without the bug.) * callproc.c (child_setup): Remove Ben's comment noting confusion over the file descriptors >= 3: close_process_descs() doesn't close those. Fix fd range to go up to MAXDESC-1 instead of 64. * process-unix.c (unix_create_process): Call begin_dont_check_for_quit to inhibit unwanted interaction (and thus breaking of X event synchronicity) in the child. 2002-09-26 Golubev I. N. <gin(a)mo.msk.ru&gt; * frame-x.c (x_delete_frame): do not call XtDisplay on a destroyed widget. 2002-08-20 Ben Sigelman <bhsigelm(a)cs.brown.edu&gt; * event-Xt.c (drain_X_queue): loop over while(XEventsQueued()) instead of while(XtAppPending()&XtIMXEvent). Upon a gdm logout, the xemacs process did not exit, but looped infinitely here. This fixes that problem. Release notes from Candidate 1 ============================== Besides the main issue (syntax optimizations affecting fontlock, indentation, movement, and cperl mode), there are a fair number of dialog fixes. If you're not using the most recent Windows version (which already has the fixes), you may want to check these out. Of special interest: java-mode, cperl-mode, font-lock, indentation ------------------------------------------------------------------- It is important to test RC1 (this candidate) or Release Candidate 2 (there will most likely be an RC2 because there are some important patches pending) because it contains much work on the syntax cache code, which is responsible for seriously degraded performance. This is especially noticable in large (> 2000 line) Java files (on a PIII 650MHz class machine, YMMV). There are three main issues to watch for. o Compare font-lock and indenting performance at the end of large Java files when using RC1 against earlier 21.4 releases. You should see a large speedup in RC1 with `lookup-syntax-properties' set to nil (now the default). There may be speedups in other modes, but java-mode seems to be the worst. If performance still seems unacceptable, please let us know. Also, if some large files seem to have acceptable performance at the end, but others are unacceptable, that is important information. The performance problem depends on the buffer position; for me, it's only perceptible thousands of lines from the beginning of the buffer. o Watch for bugs in font-locking, indentation, and syntax-based movement (forward-comment, forward-sexp, forward-word, etc). To optimize the syntax lookups (ie, `lookup-syntax-properties' set to t), a fair amount of surgery was done on src/syntax.c. I have done my best to avoid changing semantics, but the code is very unclear. Bugs may have been introduced. o cperl users must set `lookup-syntax-properties' to t. Because this performance problem affects everyone, we have changed the default of the Lisp variable `lookup-syntax-properties' to nil. This means that modes that depend on looking up `syntax-table' text properties will not work properly unless you set the variable to t. The only mode I know of that this affects is cperl mode. We will be discussing with the cperl maintainers how best to minimize inconvenience here, but for RC1, cperl users will have to turn on `lookup-syntax-properties' via customize-variable or in their init files. Mode maintainers: on Alan Mackenzie's recommendation, we have aliased parse-sexp-lookup-properties to lookup-syntax-properties. This alias will also be put into the XEmacs 21.5 tree. Other issues: ------------- o Test failures will occur in "make check". Please note that in "make check" 6 regexp tests will fail. This is due to a variance between the ways "\<" and "\b" are implemented; the latter is buggy. This is unrelated to the syntax table changes. It will (I expect) be fixed for RC2. How to get the Release Candidate ================================ ***** Check out from anonymous CVS with cvs -d :pserver:cvs＠cvs.xemacs.org:/pack/xemacscvs checkout -r r21-4-10rc1 _or_ cvs -d :pserver:cvs＠cvs.xemacs.org:/pack/xemacscvs checkout -r candidate-21-4 If you haven't done so already, do a cvs login with password "cvs". Of course, if you already have a checked-out XEmacs tree you can use "cvs update" with the same CVSROOT and -r flags. ***** By FTP from ftp://ftp.xemacs.org/pub/xemacs/xemacs-21.4/pretest/xemacs-21.4.10.tar.gz ***** Patch up from XEmacs 21.4.9 with ftp://ftp.xemacs.org/pub/xemacs/xemacs-21.4/pretest/xemacs-21.4.9-21.4.10... ------------------------------------------------------------------------ Brief summary of Changes to 21.4.10 "Military Intelligence" -- (**) Default lookup-syntax-properties to nil -- Andy Piper, Stephen Turnbull -- Fix: revert-buffer bugs -- Andy Piper -- Fix: Ugly dialog layouts -- Andy Piper -- Fix: "Stale match data" bug -- Stephen Turnbull -- Fix: Code typos -- Stephen Turnbull -- Improve: FontSet defaults -- Ville Skyttä, Stephen Turnbull -- Improve: Optimize update_syntax_cache -- Stephen Turnbull -- Improve: GNU synchs: parse-sexp-lookup-properties -- Alan Mackenzie, Stephen Turnbull -- Docs: New PROBLEMs -- Stephen Turnbull -- Docs: New oxymorons -- Stephen Turnbull -- Docs: New configure.usage -- Stephen Turnbull -- Docs: Update manuals on syntax functions -- Stephen Turnbull -- Docs: Typo fixes -- Jerry James, Ville Skyttä, Stephen Turnbull -- Improve: (Build) New regression tests -- Stephen Turnbull -------- ChangeLog entries from ChangeLog -------- 2002-09-28 Stephen J. Turnbull <stephen(a)xemacs.org&gt; * XEmacs 21.4.10 "Military Intelligence" is released. 2002-09-27 Stephen J. Turnbull <stephen(a)xemacs.org&gt; * PROBLEMS (Running, General): Missing charset in FontSet warnings. * Emacs.ad: Add charsets to *menubar*FontSet and *popup*FontSet. 2002-08-29 Ville Skyttä <ville.skytta(a)xemacs.org&gt; * Emacs.ad: Add *menubar*FontSet and *popup*FontSet entries, (self-)obtained from Red Hat. 2002-09-25 Stephen J. Turnbull <stephen(a)xemacs.org&gt; * etc/OXYMORONS: A couple new ones. 2002-09-03 Stephen J. Turnbull <stephen(a)xemacs.org&gt; * configure.usage: Complete rewrite and reorganization. -------- ChangeLog entries from lib-src/ChangeLog -------- 2002-09-28 Stephen J. Turnbull <stephen(a)xemacs.org&gt; * XEmacs 21.4.10 "Military Intelligence" is released. -------- ChangeLog entries from lisp/ChangeLog -------- 2002-09-28 Stephen J. Turnbull <stephen(a)xemacs.org&gt; * XEmacs 21.4.10 "Military Intelligence" is released. 2002-09-28 Stephen J. Turnbull <stephen(a)xemacs.org&gt; * font-lock.el (font-lock-fontify-syntactically-region): Add comment about lookup-syntax-properties, document LOUDLY as unused. 2002-08-26 Andy Piper <andy(a)xemacs.org&gt; * gutter-items.el (set-progress-feedback-instantiator): Fix bad merge. 2002-09-02 Ville Skyttä <ville.skytta(a)xemacs.org&gt; * bytecomp-runtime.el (make-obsolete): Docstring argument nit. (make-obsolete-variable): Ditto. (make-compatible): Ditto. (make-compatible-variable): Ditto. 2002-09-16 Stephen J. Turnbull <stephen(a)xemacs.org&gt; * obsolete.el (parse-sexp-lookup-properties): new compatible alias for `lookup-syntax-properties'. RFE from Alan Mackenzie. 2002-08-28 Andy Piper <andy(a)xemacs.org&gt; * files.el (revert-buffer): propagate revert-without-query-status. 2002-06-03 Andy Piper <andy(a)xemacs.org&gt; * dialog-items.el (make-search-dialog): align various subcontrols in visually pleasing ways, allow frame to autosize to the dialog size. * dialog.el (make-dialog-box): add a border inside the frame. Add :autosize property, if t then reset the frame size appropriately before the frame is mapped. * gutter-items.el (set-progress-feedback-instantiator): align various subcontrols. * gutter-items.el (set-progress-abort-instantiator): ditto. -------- ChangeLog entries from lwlib/ChangeLog -------- 2002-09-28 Stephen J. Turnbull <stephen(a)xemacs.org&gt; * XEmacs 21.4.10 "Military Intelligence" is released. -------- ChangeLog entries from man/ChangeLog -------- 2002-09-28 Stephen J. Turnbull <stephen(a)xemacs.org&gt; * XEmacs 21.4.10 "Military Intelligence" is released. 2002-09-20 Stephen J. Turnbull <stephen(a)xemacs.org&gt; * internals/internals.texi (Techniques for XEmacs Developers): More performance optimization hints. (Modules for Other Aspects of the Lisp Interpreter and Object System): Describe syntax code internals. * lispref/syntax.texi (Syntax Basics): XEmacs "20" -> "20 and later". (Syntax Class Table): Deprecate SPC as whitespace designator. (Syntax Flags): Rewrite for `8-bit' comment syntax flags. -------- ChangeLog entries from netinstall/ChangeLog -------- 2002-09-28 Stephen J. Turnbull <stephen(a)xemacs.org&gt; * XEmacs 21.4.10 "Military Intelligence" is released. -------- ChangeLog entries from nt/ChangeLog -------- 2002-09-28 Stephen J. Turnbull <stephen(a)xemacs.org&gt; * XEmacs 21.4.10 "Military Intelligence" is released. -------- ChangeLog entries from src/ChangeLog -------- 2002-09-28 Stephen J. Turnbull <stephen(a)xemacs.org&gt; * XEmacs 21.4.10 "Military Intelligence" is released. 2002-09-28 Stephen J. Turnbull <stephen(a)xemacs.org&gt; * font-lock.c (SYNTAX_START_STYLE, SYNTAX_END_STYLE): Quiet GCC. (find_context): Improve comments. * regex.c (re_compile_fastmap): Gratuitous #ifdef comment fixes. (re_match_2): An XCHAR_TABLE operation was missing. Many of the following changes were inspired by Andy Piper's (now obsolete) patch to make syntax-table text property lookup optional. [Documentation] * syntax.h (top level): Document semantics of syntax flags. * syntax.c (comstr_start, Fforward_comment): Improve comments. * syntax.c, syntax.h (top level): Document syntax cache better in comments. [Instrument syntax caching (temporary change)] * syntax.h (enum syntax_cache_statistics_functions, struct syntax_cache_statistics, SCS_STATISTICS_SET_FUNCTION, SYNTAX_CACHE_STATISTICS_COUNT_INIT): New types and macros. * syntax.c (scan_words, Fforward_comment, scan_lists, Fbackward_prefix_characters): * font-lock.c (find_context): Use them to instrument syntax caching. * syntax.c (struct syntax_cache_statistics, scs_statistics, syntax_cache_statistics_function_names): New type and variables. * syntax.c (update_syntax_cache): Update and report statistics. [Major syntax cache optimizations] syntax.c (lookup_syntax_properties): Default off. * syntax.h (UPDATE_SYNTAX_INTERNAL): New, implements UPDATE_SYNTAX*. * syntax.c (update_syntax_cache): Move init test from update_syntax_cache to UPDATE_SYNTAX_INTERNAL. [Minor syntax cache optimizations and code clarifications] * syntax.h (SYNTAX_COMMENT_BITS, SYNTAX_START_P, SYNTAX_END_P, SYNTAX_STYLES_MATCH_START_P, SYNTAX_STYLES_MATCH_1CHAR_P, SYNTAX_STYLES_MATCH_END_P, STYLE_FOUND_1CHAR_P, STYLE_FOUND_P, SYNTAX_COMMENT_MASK_END, SYNTAX_COMMENT_MASK_START, SYNTAX_COMMENT_1CHAR_MASK): Remove redundant argument `table'. (Code clarification.) * syntax.c (scan_words, find_end_of_comment, scan_sexps_forward, scan_lists): Compute code from syncode, use Boolean macros instead of redundant masking operations. (Code clarification; any decent compiler can do this optimization.) * syntax.c (scan_lists): Rearrange switches (case Sstring_fence) for clarity. * syntax.h, syntax.c (NEXT_SINGLE_PROPERTY_CHANGE, PREVIOUS_SINGLE_PROPERTY_CHANGE, update_syntax_cache): Conditionally use F*_single_property_change instead of F*_extent_change. Currently F*_single_property_change is _much_ less efficient. * dired.c (Fdirectory_files): Qt not valid in syntax_cache.object. * syntax.h (SETUP_SYNTAX_CACHE_FOR_OBJECT): Initialize syntax_cache.object properly, don't leave Qt or Qnil. (SYNTAX_CACHE_BYTE_TO_CHAR): Use SETUP_SYNTAX_CACHE_FOR_OBJECT. * syntax.c (update_syntax_cache): Use BUFFERP and STRINGP, not tests for Qt or Qnil. 2002-09-21 Stephen J. Turnbull <stephen(a)xemacs.org&gt; * redisplay.c (point_in_line_start_cache): * extents.c (print_extent_1): Fix warnings. * extents.c (Fset_extent_property): Document `syntax-table' property. * extents.h (Fnext_single_property_change): EXFUN it. 2002-09-09 Stephen J. Turnbull <stephen(a)xemacs.org&gt; * search.c (clear_unused_search_regs): New static function. (search_buffer): (simple_search): (boyer_moore): Use it. Fixes "stale match data" bug reported by Martin Stjernholm. Minor clarifications in comments. * regex.c (re_match_2_internal): Ensure no stale submatches. 2002-09-09 Stephen J. Turnbull <stephen(a)xemacs.org&gt; * event-Xt.c (enqueue_focus_event): Add missing prototype. 2002-08-29 Jerry James <james(a)xemacs.org&gt; * lisp.h: structrures -> structures. 2002-06-03 Andy Piper <andy(a)xemacs.org&gt; * frame.c (frame_conversion_internal): add conversion appropriate for setting frame dimensions in pixels. * console.h (struct console_methods): add device-specific widget layout functions. * emacs.c (main_1): initialize glyph vars. * event-Xt.c (handle_focus_event_1): call handle_focus_event_2 (handle_focus_event_2): new function. Allow focus to be set to a particular window and widget. (enqueue_focus_event): new function. Create a synthetic focus event. (emacs_Xt_handle_magic_event): call handle_focus_event_2. * frame.c (Fset_frame_pixel_height): new function. Allow pixel-based sizing. Does the right thing on character-oriented displays. (Fset_frame_pixel_width): ditto. (Fset_frame_pixel_size): ditto. (syms_of_frame): declare. * general-slots.h: add :vertically-justify and :horizontally-justify. * glyphs-msw.c (xbm_create_bitmap_from_data): warning fix. (mswindows_widget_border_width): new function. (mswindows_combo_box_instantiate): set border widths. (console_type_create_glyphs_mswindows): declare. * glyphs-widget.c: (check_valid_justification): allow vertical and horizontal justification. (widget_border_width): new function. Call device methods. (widget_instance_border_width): ditto. (logical_unit_height): new function. Determine the height of a logical-unit. Logical units are intended to provide grid-like layout for subcontrols in a layout. If the layout is :vertically-justify top then subcontrols will be positioned on logical-unit boundaries, thus allowing adjacent subcontrols to line up. (widget_logical_unit_height): ditto for widgets. (redisplay_widget): misc cleanup. (widget_spacing): new function. Call device specific methods. Intended to be the default gap between adjacent widgets. (widget_query_geometry): use new spacing functions to determine appropriate geometry. (initialize_widget_image_instance): initialize horizontal and vertical justification. (widget_instantiate): use logica-units where appropriate for sizing large widgets. (button_query_geometry): use widget_instance_border_width. (tree_view_query_geometry): ditto. (tab_control_query_geometry): ditto. (edit_field_query_geometry): new function. (layout_update): pick up :justify, :horizontally-justify and :vertically-justify none of which worked previously. :justify is used as a fallback for both horizontal and vertical justification. (layout_query_geometry): use logical units to size appropriately. (layout_layout): ditto. (Fwidget_logical_to_character_width): new function. Allow users to convert from logical units to characters so that frame sizes can be set appropriately. (Fwidget_logical_to_character_height): ditto. (syms_of_glyphs_widget): declare. (image_instantiator_edit_fields): add edit_field_query_geometry. (VALID_LAYOUT_KEYWORDS): declare :vertically-justify and :horizontally-justify. (specifier_vars_of_glyphs_widget): new function. Declare Vwidget_border_width which is not yet used. * glyphs-x.c (x_map_subwindow): enqueue appropriate focus events. (x_widget_border_width): new function. (x_widget_instantiate): create a sensible name for widgets. (console_type_create_glyphs_x): declare x_widget_border_width. * glyphs.h (DEFAULT_WIDGET_BORDER_WIDTH): new macro. (struct Lisp_Image_Instance): add :vertically-justify and :horizontally-justify. (LAYOUT_JUSTIFY_TOP): new. (IMAGE_INSTANCE_SUBWINDOW_H_JUSTIFY): new. (ROUND_UP): new macro. Round argument up to next integral interval. * redisplay-output.c (redisplay_output_layout): misc formatting. * symsinit.h: declare specifier_vars_of_glyphs_widget. -------- ChangeLog entries from tests/ChangeLog -------- 2002-09-28 Stephen J. Turnbull <stephen(a)xemacs.org&gt; * XEmacs 21.4.10 "Military Intelligence" is released. 2002-09-12 Stephen J. Turnbull <stephen(a)xemacs.org&gt; * automated/regexp-tests.el: Add word-boundary regexp tests. 2002-09-09 Stephen J. Turnbull <stephen(a)xemacs.org&gt; * automated/regexp-tests.el: Add test for stale subexpr match-data. Thanks to Martin Stjernholm for the report. * automated/syntax-tests.el: Conditionalize syntax-table property tests on feature. Enable feature if present. -- 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 -- 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

Ville> Stephen, I don't see this in RC2, and looks like other Ville> lisp/ stuff that was announced is missing too. What's up? I don't know. What do you mean you don't see it? cvs diff -r r21-4-10rc1 -r r21-4-10rc2 lisp/package-get.el cvs diff -r r21-4-10rc1 -r candidate-21-4 lisp/package-get.el recover your patch as expected. I haven't unpacked the tarball, but it is in my staging area and it is in the patch, so I expect the tarball is OK. Maybe you're using CVS and have a sticky tag != candidate-21-4? This stuff is NOT in release-21-4; I'm still not confident that the syntax stuff is right (well, I'm pretty close to a release at this point, waiting to see if we can figure out what's going on with completion and dired on Windows). -- 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

Ville> But the new sites list isn't effective in the UI because Ville> lisp/auto-autoloads.el still seems to have the old listing Ville> for package-get-download-sites. Should it be patched too, Ville> or is it supposed to be autogenerated? It is autogenerated. See the target FRC.lisp.auto.autoloads.el in the top-level makefile, and the reports in beta.err: Loading autoload... Updating autoloads for directory /playpen/Projects/XEmacs/Builds/21.5/lisp... etc, etc. Probably it should be cvs remove'd, I think that's been done for the trunk already. -- 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

Ville> On Fri, 2002-10-04 at 12:22, Stephen J. Turnbull wrote: Ville> But the new sites list isn't effective in the UI because Ville> lisp/auto-autoloads.el still seems to have the old listing Ville> for package-get-download-sites. Should it be patched too, Ville> or is it supposed to be autogenerated? Ville> Ok, time for a dumb question: where is beta.err? I can't Ville> find it, not after a "make" or "make beta". Ditto the Ville> files referenced in M-x build-report Ville> (xemacs-make-(install|check|check-temacs|all).err). Hi Ville! M-x build-report in XEmacs core expects beta.err to be present and contain output of the XEmacs make process. You can produce it as follows: make install > beta.err 2>&1 & See also /playpen/Projects/XEmacs/Builds/21.5/etc/aliases.ksh in your setup. My separate build package controls some of the build-report variables and sets them according to the files it produces. The build package generates compilation buffers named according to the make commands executed. That's where xemacs-make-(install|check|check-temacs|all).err or similar come from. Ah, the build package needes more detailed documentation at http://www.dk.xemacs.org/Documentation/packages/html/build_toc.html#SEC_C... Best regards, Adrian Ville> And I've built 21.4.10 RC2 several times now, and Ville> lisp/auto-autoloads.el(c?) has never been updated with the new sites Ville> list from package-get.el. BTW, your example above talks about 21.5, Ville> don't know if that's relevant. Ville> What am I missing? I'm building from the tarball, if that matters. Ville> -- Ville> \/ille Skyttä Ville> ville.skytta at xemacs.org -- Adrian Aichner mailto:adrian＠xemacs.org http://www.xemacs.org/

Please test Release Candidate 3 of XEmacs 21.4.10 ================================================= There are only a few specialized changes from RC1 and RC2. The important ones affect native Windows users, and HP/UX users. If they are not relevant to your usage, there is no need to build RC3. Continue testing RC1 or RC2. Report your builds via M-x build-report and any bugs or interesting behavior (including improvements ;-) via M-x report-emacs-bug. o There are reports of problems _on Windows_ with `file-truename' (native Windows builds), or if `directory-sep-char' is explicitly set to `/'. These often seem to manifest using dired and file name completion. A preliminary patch has been applied to this candidate which improves things. Users of the native Windows build are requested to test Candidate 3. Use dired with wildcard filenames and file completion. It is unlikely that this will help with the `directory-sep-char' problem, but if you set this variable for some reason, reports on anomolies would be appreciated. o HP/UX 11.xx is reported to have a non-functional getaddrinfo. HP/UX users please check your config.log to make sure that use of getaddrinfo is suppressed. (I'm not sure what should happen on HP/UX < 11.xx. If getaddrinfo is not implemented then it won't be used, if it is implemented then it should be used.) o Other changes are detailed below. Test reports on Candidates 1 and 2: o The notable fix is for the XEmacs processes that used to linger after logging out from gdm (and some other session managers), consuming 100% of CPU. This patch seems to be successful. o Some users have reported substantial speedups with the changes to the code handling syntax-table text properties. I'd still like to hear reports from *** cperl-mode *** users, as the group most likely to be adversely affected. *** Note *** The prerelease location is ftp://ftp.xemacs.org/pub/xemacs/xemacs-21.4/pretest/ The rc1 and rc2 versions are still available. The tag for checkout from CVS is `r21-4-10rc3'. Brief list of changes Candidate 2 -> Candidate 3 ------------------------------------------------ -- (**) Add: Check Mule coding cookie in 2d line, too -- Stephen Turnbull -- (**) Add: C-x r d invokes delete-rectangle -- Katsumi Yamaoka -- Fix: Truename bug on native Windows -- Stephen Turnbull -- Fix: Regex crashes; increase stack size for Windows builds -- Andy Piper -- Fix: font-lock-syntax-keywords tromps font-lock-keywords -- Daiki Ueno, Katsumi Yamaoka -- Fix: Don't use getaddrinfo on HP/UX -- Darryl Okahata, Stephen Turnbull -- Fix: Always generate auto-autoloads.el -- Stephen Turnbull -- Improve PROBLEMS, INSTALL, etc/BETA -- Stephen Turnbull -- Improve: Test suite more friendly, some docs -- Stephen Turnbull -- Improve: Crash message -- Stephen Turnbull Change Logs ----------- ChangeLog: 2002-10-09 Stephen J. Turnbull <stephen(a)xemacs.org&gt; * etc/BETA: Improve descriptions of XEmacs Patches; mention Design. 2002-10-08 Stephen J. Turnbull <stephen(a)xemacs.org&gt; * PROBLEMS (MacOS/X): Describe stack limitation. (Digital Unix): Generalize to all regexp-using applications. * INSTALL (PREQUISITES): Mention MacOS/X stack limitation. (PROBLEMS): Point to PROBLEMS file for build notes. Thanks to Skip Montanaro for the report. 2002-10-07 Stephen J. Turnbull <stephen(a)xemacs.org&gt; * Makefile.in.in (all-elcs): Depend on autoloads. (autoloads): Depend on lib-src, lwlib, and src. 2002-10-04 Stephen J. Turnbull <stephen(a)xemacs.org&gt; * configure.in (Check for POSIX functions): New section head. getaddrinfo is detected on HP-UX 11.XX, but appears to be non-functional. Disable it. Based on work by Darryl Okahata. lisp/ChangeLog: 2002-10-07 Katsumi Yamaoka <yamaoka(a)jpl.org&gt; * font-lock.el (font-lock-fontify-keywords-region): Fix typo. 2002-10-07 Katsumi Yamaoka <yamaoka(a)jpl.org&gt; * keydefs.el (global-map): Bind C-xrd to delete-rectangle per GNU. 2001-05-10 Daiki Ueno <ueno(a)unixuser.org&gt; * font-lock.el (font-lock-compile-keywords): Rewrite as a side effect free function. (font-lock-fontify-keywords-region): Supply the argument to font-lock-compile-keywords. src/ChangeLog: 2002-10-10 Stephen J. Turnbull <stephen(a)xemacs.org&gt; * realpath.c (win32_readlink): Try to fix truename of wildcard bug. * emacs.c (shut_down_emacs): Improve crash message. 2002-10-09 Stephen J. Turnbull <stephen(a)xemacs.org&gt; * file-coding.c (determine_real_coding_system): Check for coding cookie in first LINES_TO_CHECK lines (currently 2). Cater to shebangs and Python PEP 263. man/ChangeLog: 2002-08-22 Stephen J. Turnbull <stephen(a)xemacs.org&gt; * internals/internals.texi (Regression Testing XEmacs): Document how to skip and warn about tests that depend on packages. nt/ChangeLog: 2002-10-08 Andy Piper <andy(a)xemacs.org&gt; * xemacs.mak: set the stacksize to 8mb which seems to be the magic number for regex. tests/ChangeLog: 2002-08-22 Stephen J. Turnbull <stephen(a)xemacs.org&gt; * automated/test-harness.el (test-harness-from-buffer): Print-Skip: new local function. * automated/syntax-tests.el (forward-comment at buffer boundaries): * automated/test-harness.el (Check-Message): Warn, not barf, if required packages are unavailable. Brief list of changes Candidate 1 -> Candidate 2 ------------------------------------------------ -- Fix: (X11) CPU-eating XEmacs process at gdm logout -- Ben Sigelman -- Fix: Freeze in certain modal dialogs -- Andy Piper -- Fix: (MS Windows) Handle leak -- Mike Alexander -- Fix: (Unix) Subprocess problems -- Mike Sperber, Stephen Turnbull -- Fix: (X11) Crash in external widget -- I. N. Golubev -- Update: Package sites and Texi docs -- Ville Skyttä Change Logs ----------- lisp/ChangeLog: 2002-09-22 Ville Skyttä <ville.skytta(a)xemacs.org&gt; * package-get.el (package-get-download-sites): Bring sites list up to date, prefer xx.xemacs.org addresses. 2002-09-27 Andy Piper <andy(a)xemacs.org&gt; * dialog.el (make-dialog-box): for general dialogs only delete-frame-hook will be run. man/ChangeLog: 2002-09-22 Ville Skyttä <ville.skytta(a)xemacs.org&gt; * xemacs/packages.texi (Available Packages): Add fortran-modes, perl-modes, psgml-dtds, python-modes and ruby-modes. Some consistency tweaks. src/ChangeLog: 2002-08-30 Mike Alexander <mta(a)arbortext.com&gt; * event-msw.c (winsock_closer): Use closesocket, not CloseHandle to close a socket. 2002-10-03 Stephen J. Turnbull <stephen(a)xemacs.org&gt; * callproc.c (child_setup): 21.4 doesn't have retry_close. 2002-09-20 Mike Sperber <mike(a)xemacs.org&gt; (This partially gets back my patch from 2002-07-20 without the bug.) * callproc.c (child_setup): Remove Ben's comment noting confusion over the file descriptors >= 3: close_process_descs() doesn't close those. Fix fd range to go up to MAXDESC-1 instead of 64. * process-unix.c (unix_create_process): Call begin_dont_check_for_quit to inhibit unwanted interaction (and thus breaking of X event synchronicity) in the child. 2002-09-26 Golubev I. N. <gin(a)mo.msk.ru&gt; * frame-x.c (x_delete_frame): do not call XtDisplay on a destroyed widget. 2002-08-20 Ben Sigelman <bhsigelm(a)cs.brown.edu&gt; * event-Xt.c (drain_X_queue): loop over while(XEventsQueued()) instead of while(XtAppPending()&XtIMXEvent). Upon a gdm logout, the xemacs process did not exit, but looped infinitely here. This fixes that problem. Release notes from Candidate 1 ============================== Besides the main issue (syntax optimizations affecting fontlock, indentation, movement, and cperl mode), there are a fair number of dialog fixes. If you're not using the most recent Windows version (which already has the fixes), you may want to check these out. Of special interest: java-mode, cperl-mode, font-lock, indentation ------------------------------------------------------------------- It is important to test RC1 (this candidate) or Release Candidate 2 (there will most likely be an RC2 because there are some important patches pending) because it contains much work on the syntax cache code, which is responsible for seriously degraded performance. This is especially noticable in large (> 2000 line) Java files (on a PIII 650MHz class machine, YMMV). There are three main issues to watch for. o Compare font-lock and indenting performance at the end of large Java files when using RC1 against earlier 21.4 releases. You should see a large speedup in RC1 with `lookup-syntax-properties' set to nil (now the default). There may be speedups in other modes, but java-mode seems to be the worst. If performance still seems unacceptable, please let us know. Also, if some large files seem to have acceptable performance at the end, but others are unacceptable, that is important information. The performance problem depends on the buffer position; for me, it's only perceptible thousands of lines from the beginning of the buffer. o Watch for bugs in font-locking, indentation, and syntax-based movement (forward-comment, forward-sexp, forward-word, etc). To optimize the syntax lookups (ie, `lookup-syntax-properties' set to t), a fair amount of surgery was done on src/syntax.c. I have done my best to avoid changing semantics, but the code is very unclear. Bugs may have been introduced. o cperl users must set `lookup-syntax-properties' to t. Because this performance problem affects everyone, we have changed the default of the Lisp variable `lookup-syntax-properties' to nil. This means that modes that depend on looking up `syntax-table' text properties will not work properly unless you set the variable to t. The only mode I know of that this affects is cperl mode. We will be discussing with the cperl maintainers how best to minimize inconvenience here, but for RC1, cperl users will have to turn on `lookup-syntax-properties' via customize-variable or in their init files. Mode maintainers: on Alan Mackenzie's recommendation, we have aliased parse-sexp-lookup-properties to lookup-syntax-properties. This alias will also be put into the XEmacs 21.5 tree. Other issues: ------------- o Test failures will occur in "make check". Please note that in "make check" 6 regexp tests will fail. This is due to a variance between the ways "\<" and "\b" are implemented; the latter is buggy. This is unrelated to the syntax table changes. It will (I expect) be fixed for RC2. How to get the Release Candidate ================================ ***** Check out from anonymous CVS with cvs -d :pserver:cvs＠cvs.xemacs.org:/pack/xemacscvs checkout -r r21-4-10rc1 _or_ cvs -d :pserver:cvs＠cvs.xemacs.org:/pack/xemacscvs checkout -r candidate-21-4 If you haven't done so already, do a cvs login with password "cvs". Of course, if you already have a checked-out XEmacs tree you can use "cvs update" with the same CVSROOT and -r flags. ***** By FTP from ftp://ftp.xemacs.org/pub/xemacs/xemacs-21.4/pretest/xemacs-21.4.10.tar.gz ***** Patch up from XEmacs 21.4.9 with ftp://ftp.xemacs.org/pub/xemacs/xemacs-21.4/pretest/xemacs-21.4.9-21.4.10... ------------------------------------------------------------------------ Brief summary of Changes to 21.4.10 "Military Intelligence" -- (**) Default lookup-syntax-properties to nil -- Andy Piper, Stephen Turnbull -- Fix: revert-buffer bugs -- Andy Piper -- Fix: Ugly dialog layouts -- Andy Piper -- Fix: "Stale match data" bug -- Stephen Turnbull -- Fix: Code typos -- Stephen Turnbull -- Improve: FontSet defaults -- Ville Skyttä, Stephen Turnbull -- Improve: Optimize update_syntax_cache -- Stephen Turnbull -- Improve: GNU synchs: parse-sexp-lookup-properties -- Alan Mackenzie, Stephen Turnbull -- Docs: New PROBLEMs -- Stephen Turnbull -- Docs: New oxymorons -- Stephen Turnbull -- Docs: New configure.usage -- Stephen Turnbull -- Docs: Update manuals on syntax functions -- Stephen Turnbull -- Docs: Typo fixes -- Jerry James, Ville Skyttä, Stephen Turnbull -- Improve: (Build) New regression tests -- Stephen Turnbull -------- ChangeLog entries from ChangeLog -------- 2002-09-28 Stephen J. Turnbull <stephen(a)xemacs.org&gt; * XEmacs 21.4.10 "Military Intelligence" is released. 2002-09-27 Stephen J. Turnbull <stephen(a)xemacs.org&gt; * PROBLEMS (Running, General): Missing charset in FontSet warnings. * Emacs.ad: Add charsets to *menubar*FontSet and *popup*FontSet. 2002-08-29 Ville Skyttä <ville.skytta(a)xemacs.org&gt; * Emacs.ad: Add *menubar*FontSet and *popup*FontSet entries, (self-)obtained from Red Hat. 2002-09-25 Stephen J. Turnbull <stephen(a)xemacs.org&gt; * etc/OXYMORONS: A couple new ones. 2002-09-03 Stephen J. Turnbull <stephen(a)xemacs.org&gt; * configure.usage: Complete rewrite and reorganization. -------- ChangeLog entries from lib-src/ChangeLog -------- 2002-09-28 Stephen J. Turnbull <stephen(a)xemacs.org&gt; * XEmacs 21.4.10 "Military Intelligence" is released. -------- ChangeLog entries from lisp/ChangeLog -------- 2002-09-28 Stephen J. Turnbull <stephen(a)xemacs.org&gt; * XEmacs 21.4.10 "Military Intelligence" is released. 2002-09-28 Stephen J. Turnbull <stephen(a)xemacs.org&gt; * font-lock.el (font-lock-fontify-syntactically-region): Add comment about lookup-syntax-properties, document LOUDLY as unused. 2002-08-26 Andy Piper <andy(a)xemacs.org&gt; * gutter-items.el (set-progress-feedback-instantiator): Fix bad merge. 2002-09-02 Ville Skyttä <ville.skytta(a)xemacs.org&gt; * bytecomp-runtime.el (make-obsolete): Docstring argument nit. (make-obsolete-variable): Ditto. (make-compatible): Ditto. (make-compatible-variable): Ditto. 2002-09-16 Stephen J. Turnbull <stephen(a)xemacs.org&gt; * obsolete.el (parse-sexp-lookup-properties): new compatible alias for `lookup-syntax-properties'. RFE from Alan Mackenzie. 2002-08-28 Andy Piper <andy(a)xemacs.org&gt; * files.el (revert-buffer): propagate revert-without-query-status. 2002-06-03 Andy Piper <andy(a)xemacs.org&gt; * dialog-items.el (make-search-dialog): align various subcontrols in visually pleasing ways, allow frame to autosize to the dialog size. * dialog.el (make-dialog-box): add a border inside the frame. Add :autosize property, if t then reset the frame size appropriately before the frame is mapped. * gutter-items.el (set-progress-feedback-instantiator): align various subcontrols. * gutter-items.el (set-progress-abort-instantiator): ditto. -------- ChangeLog entries from lwlib/ChangeLog -------- 2002-09-28 Stephen J. Turnbull <stephen(a)xemacs.org&gt; * XEmacs 21.4.10 "Military Intelligence" is released. -------- ChangeLog entries from man/ChangeLog -------- 2002-09-28 Stephen J. Turnbull <stephen(a)xemacs.org&gt; * XEmacs 21.4.10 "Military Intelligence" is released. 2002-09-20 Stephen J. Turnbull <stephen(a)xemacs.org&gt; * internals/internals.texi (Techniques for XEmacs Developers): More performance optimization hints. (Modules for Other Aspects of the Lisp Interpreter and Object System): Describe syntax code internals. * lispref/syntax.texi (Syntax Basics): XEmacs "20" -> "20 and later". (Syntax Class Table): Deprecate SPC as whitespace designator. (Syntax Flags): Rewrite for `8-bit' comment syntax flags. -------- ChangeLog entries from netinstall/ChangeLog -------- 2002-09-28 Stephen J. Turnbull <stephen(a)xemacs.org&gt; * XEmacs 21.4.10 "Military Intelligence" is released. -------- ChangeLog entries from nt/ChangeLog -------- 2002-09-28 Stephen J. Turnbull <stephen(a)xemacs.org&gt; * XEmacs 21.4.10 "Military Intelligence" is released. -------- ChangeLog entries from src/ChangeLog -------- 2002-09-28 Stephen J. Turnbull <stephen(a)xemacs.org&gt; * XEmacs 21.4.10 "Military Intelligence" is released. 2002-09-28 Stephen J. Turnbull <stephen(a)xemacs.org&gt; * font-lock.c (SYNTAX_START_STYLE, SYNTAX_END_STYLE): Quiet GCC. (find_context): Improve comments. * regex.c (re_compile_fastmap): Gratuitous #ifdef comment fixes. (re_match_2): An XCHAR_TABLE operation was missing. Many of the following changes were inspired by Andy Piper's (now obsolete) patch to make syntax-table text property lookup optional. [Documentation] * syntax.h (top level): Document semantics of syntax flags. * syntax.c (comstr_start, Fforward_comment): Improve comments. * syntax.c, syntax.h (top level): Document syntax cache better in comments. [Instrument syntax caching (temporary change)] * syntax.h (enum syntax_cache_statistics_functions, struct syntax_cache_statistics, SCS_STATISTICS_SET_FUNCTION, SYNTAX_CACHE_STATISTICS_COUNT_INIT): New types and macros. * syntax.c (scan_words, Fforward_comment, scan_lists, Fbackward_prefix_characters): * font-lock.c (find_context): Use them to instrument syntax caching. * syntax.c (struct syntax_cache_statistics, scs_statistics, syntax_cache_statistics_function_names): New type and variables. * syntax.c (update_syntax_cache): Update and report statistics. [Major syntax cache optimizations] syntax.c (lookup_syntax_properties): Default off. * syntax.h (UPDATE_SYNTAX_INTERNAL): New, implements UPDATE_SYNTAX*. * syntax.c (update_syntax_cache): Move init test from update_syntax_cache to UPDATE_SYNTAX_INTERNAL. [Minor syntax cache optimizations and code clarifications] * syntax.h (SYNTAX_COMMENT_BITS, SYNTAX_START_P, SYNTAX_END_P, SYNTAX_STYLES_MATCH_START_P, SYNTAX_STYLES_MATCH_1CHAR_P, SYNTAX_STYLES_MATCH_END_P, STYLE_FOUND_1CHAR_P, STYLE_FOUND_P, SYNTAX_COMMENT_MASK_END, SYNTAX_COMMENT_MASK_START, SYNTAX_COMMENT_1CHAR_MASK): Remove redundant argument `table'. (Code clarification.) * syntax.c (scan_words, find_end_of_comment, scan_sexps_forward, scan_lists): Compute code from syncode, use Boolean macros instead of redundant masking operations. (Code clarification; any decent compiler can do this optimization.) * syntax.c (scan_lists): Rearrange switches (case Sstring_fence) for clarity. * syntax.h, syntax.c (NEXT_SINGLE_PROPERTY_CHANGE, PREVIOUS_SINGLE_PROPERTY_CHANGE, update_syntax_cache): Conditionally use F*_single_property_change instead of F*_extent_change. Currently F*_single_property_change is _much_ less efficient. * dired.c (Fdirectory_files): Qt not valid in syntax_cache.object. * syntax.h (SETUP_SYNTAX_CACHE_FOR_OBJECT): Initialize syntax_cache.object properly, don't leave Qt or Qnil. (SYNTAX_CACHE_BYTE_TO_CHAR): Use SETUP_SYNTAX_CACHE_FOR_OBJECT. * syntax.c (update_syntax_cache): Use BUFFERP and STRINGP, not tests for Qt or Qnil. 2002-09-21 Stephen J. Turnbull <stephen(a)xemacs.org&gt; * redisplay.c (point_in_line_start_cache): * extents.c (print_extent_1): Fix warnings. * extents.c (Fset_extent_property): Document `syntax-table' property. * extents.h (Fnext_single_property_change): EXFUN it. 2002-09-09 Stephen J. Turnbull <stephen(a)xemacs.org&gt; * search.c (clear_unused_search_regs): New static function. (search_buffer): (simple_search): (boyer_moore): Use it. Fixes "stale match data" bug reported by Martin Stjernholm. Minor clarifications in comments. * regex.c (re_match_2_internal): Ensure no stale submatches. 2002-09-09 Stephen J. Turnbull <stephen(a)xemacs.org&gt; * event-Xt.c (enqueue_focus_event): Add missing prototype. 2002-08-29 Jerry James <james(a)xemacs.org&gt; * lisp.h: structrures -> structures. 2002-06-03 Andy Piper <andy(a)xemacs.org&gt; * frame.c (frame_conversion_internal): add conversion appropriate for setting frame dimensions in pixels. * console.h (struct console_methods): add device-specific widget layout functions. * emacs.c (main_1): initialize glyph vars. * event-Xt.c (handle_focus_event_1): call handle_focus_event_2 (handle_focus_event_2): new function. Allow focus to be set to a particular window and widget. (enqueue_focus_event): new function. Create a synthetic focus event. (emacs_Xt_handle_magic_event): call handle_focus_event_2. * frame.c (Fset_frame_pixel_height): new function. Allow pixel-based sizing. Does the right thing on character-oriented displays. (Fset_frame_pixel_width): ditto. (Fset_frame_pixel_size): ditto. (syms_of_frame): declare. * general-slots.h: add :vertically-justify and :horizontally-justify. * glyphs-msw.c (xbm_create_bitmap_from_data): warning fix. (mswindows_widget_border_width): new function. (mswindows_combo_box_instantiate): set border widths. (console_type_create_glyphs_mswindows): declare. * glyphs-widget.c: (check_valid_justification): allow vertical and horizontal justification. (widget_border_width): new function. Call device methods. (widget_instance_border_width): ditto. (logical_unit_height): new function. Determine the height of a logical-unit. Logical units are intended to provide grid-like layout for subcontrols in a layout. If the layout is :vertically-justify top then subcontrols will be positioned on logical-unit boundaries, thus allowing adjacent subcontrols to line up. (widget_logical_unit_height): ditto for widgets. (redisplay_widget): misc cleanup. (widget_spacing): new function. Call device specific methods. Intended to be the default gap between adjacent widgets. (widget_query_geometry): use new spacing functions to determine appropriate geometry. (initialize_widget_image_instance): initialize horizontal and vertical justification. (widget_instantiate): use logica-units where appropriate for sizing large widgets. (button_query_geometry): use widget_instance_border_width. (tree_view_query_geometry): ditto. (tab_control_query_geometry): ditto. (edit_field_query_geometry): new function. (layout_update): pick up :justify, :horizontally-justify and :vertically-justify none of which worked previously. :justify is used as a fallback for both horizontal and vertical justification. (layout_query_geometry): use logical units to size appropriately. (layout_layout): ditto. (Fwidget_logical_to_character_width): new function. Allow users to convert from logical units to characters so that frame sizes can be set appropriately. (Fwidget_logical_to_character_height): ditto. (syms_of_glyphs_widget): declare. (image_instantiator_edit_fields): add edit_field_query_geometry. (VALID_LAYOUT_KEYWORDS): declare :vertically-justify and :horizontally-justify. (specifier_vars_of_glyphs_widget): new function. Declare Vwidget_border_width which is not yet used. * glyphs-x.c (x_map_subwindow): enqueue appropriate focus events. (x_widget_border_width): new function. (x_widget_instantiate): create a sensible name for widgets. (console_type_create_glyphs_x): declare x_widget_border_width. * glyphs.h (DEFAULT_WIDGET_BORDER_WIDTH): new macro. (struct Lisp_Image_Instance): add :vertically-justify and :horizontally-justify. (LAYOUT_JUSTIFY_TOP): new. (IMAGE_INSTANCE_SUBWINDOW_H_JUSTIFY): new. (ROUND_UP): new macro. Round argument up to next integral interval. * redisplay-output.c (redisplay_output_layout): misc formatting. * symsinit.h: declare specifier_vars_of_glyphs_widget. -------- ChangeLog entries from tests/ChangeLog -------- 2002-09-28 Stephen J. Turnbull <stephen(a)xemacs.org&gt; * XEmacs 21.4.10 "Military Intelligence" is released. 2002-09-12 Stephen J. Turnbull <stephen(a)xemacs.org&gt; * automated/regexp-tests.el: Add word-boundary regexp tests. 2002-09-09 Stephen J. Turnbull <stephen(a)xemacs.org&gt; * automated/regexp-tests.el: Add test for stale subexpr match-data. Thanks to Martin Stjernholm for the report. * automated/syntax-tests.el: Conditionalize syntax-table property tests on feature. Enable feature if present. -- 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 -- 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 -- 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

"Stephen J. Turnbull" <stephen(a)xemacs.org&gt; writes: Good news! With this candidate I am able to use the newest version of Oort Gnus, ie something you've checked in today has fixed the problem reported in http://list-archive.xemacs.org/xemacs-beta/200209/msg00468.htm Yeah! norbert.

"Stephen J. Turnbull" <stephen(a)xemacs.org&gt; writes: Hmm, thought so from the ChangeLog entries. The changes to the Gnus code-base are pretty high at the moment, so the bug's obfuscation must have happened there. Wait and see, norbert.

rendhalver(a)xemacs.org (Rendhalver [Peter Brown]) writes: Hi! Yup, 4.7 as of yesterday. I don't think that's associated, though. norbert.