1.2.4:
Alan Coopersmith (3):
Make sure to leave room for trailing nil byte in yyGetNumber
Delete redundant redeclarations of extern Atoms in symbols.c
xkbcomp 1.2.4
Andreas Wettstein (1):
include resets group compatibility modifiers #43091
Jeremy Huddleston (6):
Include strings.h for strcasecmp and string.h for strdup
Add missing _X_NORETURN to uFatalError
Fix -Wformat warnings
Correct our zeroing out of the row in InitRowInfo
Silence -Wshadow warnings
Silence -Wparentheses-equality
Peter Hutterer (2):
Print version number on -version.
Add .pc file for version export
1.2.3:
There was a pretty bad problem with the 1.2.2 release, which quite
effectively crushed Any+AnyOfOrNone(All) interp mappings, in that
xkeyboard-config actually relied on an explicit Any+AnyOfOrNone(All)
mapping to function. 1.2.3 builds on the previous 1.2.2 fix to allow
explicit Any+AnyOfOrNone(All) mappings, while still ignoring interp
mappings whose keysyms could not be found.
I've regression-tested this against xkbcomp 1.1.1 by comparing the
results of compiling every layout (primary variants only) in
xkeyboard-config, and there were no differences at all.
Distributions are strongly recommended to push 1.2.3 instead of 1.2.2,
whose tarballs have been removed from the archive.
1.2.2:
This is an semi-urgent bugfix release for the two commits of mine, which
fix a particularly bad failure in the compat code. If you had a
SymInterpret definition (i.e. to map a keysym to an action, as used for
VT switch and others) with an unknown keycode, every key that wasn't
already mapped to an action would get mapped to your new action.
So, for example, if you had XF86LogWinTree mapped to the PrWins action,
and a libX11 that was unaware of the new symbol, every key aside from VT
switching, zapping, pointer keys and modifiers, would dump the window
tree to your X log, and do nothing else.
Distributions are strongly, strongly recommended to upgrade, including
in stable series.
1.2.1:
This release allows the uploading of keymaps per-keyboard using the -i flag
(which wasn't hooked up to anything before).
Example usage:
xkbcomp -i 12 qwerty.xkb :0
xkbcomp -i 13 azerty.xkb :0
The above will load a qwerty and azerty map in to the keyboards 12 and 13,
respectively. Watch out for bug 21859 though.
http://bugs.freedesktop.org/show_bug.cgi?id=21859
Aside from this new feature, we have the usual cleanups in the build system.
1.2.0:
This release provides performance improvements and bug fixes, mixed in
with the usual recent collection of build configuration improvements and
janitorial cleanups.
xkbcomp wanted a release. This version does not have a lot of new features
but enough code was modified to justify the 1.1.0 bump.
The most interesting feature is the ability to configure the xkb data dir at
configure time rather than being forced to the hardcoded default.
Other changes include a few bug fixes, cleanups, comments and a run of
'indent' over the repository. The latter may require distribution patches
to be adjusted.
Cheers,
Peter
Alan Coopersmith (5):
Check for strdup & strcasecmp before assuming we need to provide our own
Remove X_NOT_POSIX #ifdefs
Remove #if 0'd typedef that appears nowhere else.
sprintf -> snprintf conversions
Use temporary buffer for generating Uxxx names to avoid overflow
Dan Nicholson (1):
Allow the default XKB config root to be set from configure
Matthieu Herrb (1):
Minimal man page addition for new -i option.
Paulo Cesar Pereira de Andrade (2):
Correct make distcheck and most gcc and sparse warnings.
Correct a remaining of a wrong change to make some DEBUG_VAR's static.
Peter Hutterer (13):
Remove RCS tags.
Add a few explanatory comments.
Indent fixes.
If we're not using XkbDesc, don't require it as a parameter.
Add some explanatory comments
Silence valgrind warnings.
Remove useless longestPath variable.
More comments.
Don't pass useless handlers around in HandleIncludeKeycodes.
Remove unused merge parameter from HandleKeyNameVar.
Don't overwrite previously assigned type indices with useless ones.
Check for NULL strings before calling strdup.
Bump to 1.1.0.
commit 1aecdffaa0db7bbf85bc0aae9043e9437b25c30d
Author: Daniel Stone <daniel@fooishbar.org>
Date: Thu Apr 17 00:52:29 2008 +0300
Don't scan paths which make NO SENSE WHATSOEVER TO SCAN
Hey, I wonder if we have XKB files in our directory! I wonder if we
haven't bothered with a structure, and let's try to open a file called
'misc' in someone's home directory! What a surprise, it's not a valid
XKB file! Let's fail miserably!
SURPRISINGLY, THIS IS NOT USEFUL BEHAVIOUR.
commit a3a05fd489a9a35942a4419c537634e796dbebd1
Author: Daniel Stone <daniel@fooishbar.org>
Date: Thu Apr 17 00:50:07 2008 +0300
xkbcomp: Take a device ID argument
Allows the user to set maps (or whatever) on arbitrary devices.
Adam Jackson (2):
Bug #7645: Fix a conditional that always evaluates to FALSE.
xkbcomp 1.0.4
Alan Coopersmith (1):
Bug 14185: MAINTAINERCLEANFILES multiply defined in Makefile.am
Daniel Drake (1):
Bug #11025: xkbcomp COPYING file
James Cloos (3):
Rename .cvsignore to .gitignore
Add *~ to .gitignore to skip patch/emacs droppings
Replace static ChangeLog with dist-hook to generate from git log
Matthieu Herrb (2):
Also mark xkbparse.h as a generated file.
Revert "Also mark xkbparse.h as a generated file."
Tilman Sauerbeck (3):
Fixed an invalid memory access.
Fixed a bunch of const correctness bugs.
Ignore *.o and ylwrap.
