X Strike Force SVN commit: rev 51 - branches/4.3.0/sid/debian
Author: branden
Date: 2003-05-25 18:02:52 -0500 (Sun, 25 May 2003)
New Revision: 51
Modified:
branches/4.3.0/sid/debian/changelog
Log:
merge in mainline changelog entries not included in Daniel's branch; write spurious entry for 4.3.0-1
Modified: branches/4.3.0/sid/debian/changelog
==============================================================================
--- branches/4.3.0/sid/debian/changelog 2003-05-25 22:58:48 UTC (rev 50)
+++ branches/4.3.0/sid/debian/changelog 2003-05-25 23:02:52 UTC (rev 51)
@@ -1,3 +1,959 @@
+xfree86 (4.3.0-0pre1v1) unstable; urgency=low
+
+ * New upstream release.
+ * Some things changed. This release would not have been possible without
+ the assistance of Daniel Stone and ISHIKAWA Mutsumi. A proper changelog
+ entry for this release may not be possible even if God Himself grabbed my
+ keyboard away from me.
+
+ -- Branden Robinson <branden@debian.org> Sun, 25 May 2003 03:51:59 -0500
+
+xfree86 (4.2.1-7) unstable; urgency=medium
+
+ * The "plans change" release; recently it had not been my intention to
+ release another revision of 4.2.1 at all, but LaMont Jones persuaded me to
+ due to problems building KDE on some architectures due to the absence of a
+ libxkbfile_pic.a file. This release also fixes several outstanding
+ issues, of course.
+
+ * RELEASE NOTE: I expect this release to continue to have Int10 problems on
+ the IA-64 architecture, and for the Radeon 3D acceleration to continue to
+ be broken due to GCC 3.2 issues. XFree86 4.3.0 should resolve these
+ issues, and packaging of XFree86 4.3.0 is underway; please see
+ <http://lists.debian.org/debian-x/> for more information.
+
+ * patch #000_stolen_from_HEAD:
+ (xc/lib/Xrender/{Glyph.c,Xrenderint.h}):
+ + Render uses int for 32 bits; Xlib uses long. This matters on Alpha.
+ (Keith Packard)
+ (xc/lib/Xrender/Glyph.c):
+ + XRenderFreeGlyphs wasn't sending the glyphset. (Keith Packard)
+ + CompositeGlyphs functions were re-using overwritten data and trashing
+ the output buffer. (Keith Packard)
+ + CompositeText{8,16,32} weren't switching glyphsets at the right time.
+ (Keith Packard) (Closes: #181815)
+ + Was miscomputing request length for String16 and String32. (Keith
+ Packard)
+ (xc/lib/Xrender/Xrender.c):
+ + Skip unexpected data in QueryFormats. Deal with broken X servers a bit
+ better. (Keith Packard)
+ (xc/programs/Xserver/hw/xfree86/drivers/i810/i830_driver.c):
+ Workaround for BIOSes which force 1MB stolen memory, and agpgart related
+ fix (#5300, Abraham vd Merwe). (Closes: #168469)
+
+ * patch #000_stolen_from_HEAD_ati_driver:
+ (xc/programs/Xserver/hw/xfree86/drivers/ati/r128_dri.c): Second stage of
+ mesa-4 merge. (Not a very useful commit message; fixes hardware scissor
+ handling when VT switching with DRI enabled.) (Closes: #185057)
+
+ * patch #000_stolen_from_HEAD_savage_driver:
+ + (savage_video): Remove whitespace in front of pre-processor directives
+ (Marc La France)
+ + (savage_cursor.c,savage_dga.c,savage_driver.c,savage_driver.h):
+ - Fixed VBE EDID read: due to a missing register setting read ended in
+ endless loop on certain systems (Egbert Eich).
+ - Savage driver: disable HW cursor on stretched LCD displays, let
+ SaveScreen and DGA test if HW cursor is on before disabling/reenabling
+ it (Egbert Eich).
+ + (savage_driver.c):
+ - Initializing clock ranges in several drivers to get rid of the
+ annoying 'scaled from 0.0 MHz' message in xf86PrintModes() (Egbert
+ Eich).
+ - fixed typo in savage driver (Egbert Eich)
+ - Reinstate the VBEInit() call in the savage driver's PreInit(). This
+ appears to fix a problem on some Savage MX platforms (#5649, Mike
+ Harris). (Closes: #182788)
+
+ * patch #000_stolen_from_HEAD_xman:
+ + (buttons.c, handler.c):
+ * Initialize structure field man_globals->search_widget to NULL in xman
+ code, the help message states that pressing Ctrl+S in the help window
+ has no effect, but in some cases it would crash xman. (Paulo C�r
+ Pereira de Andrade)
+ + (man.c):
+ * Fix compiler warning. (Torrey T. Lyons)
+
+ * patch #000_stolen_from_HEAD_xtt:
+ + (xttfuncs.c): removed extraneous hunk of patch that grabbed change to
+ font renderer registration, for which the support code hasn't been
+ backported (Closes: #182687)
+ + (xttcconv.c): Fix an "xtt" font backend module crash that shows up on
+ Linux/PPC, related to an invalid code converter module being loaded
+ because of a broken string comparison in xttconv.c (#A.1647, Chisato
+ Yamauchi, Daisuke MASATSUNA). (Closes: #182835)
+
+ * patch #024,451: resynced with #000_stolen_from_HEAD_ati_driver
+ * patch #032: uh, actually define a ComplexCplusplusProgramTargetNoMan()
+ rule, instead of redefining ComplexCplusplusProgramTarget() (thanks, Colin
+ Watson) (Closes: #186321)
+ * patch #046: updated to create static PIC version of libxkbfile.a as well;
+ fixes FTBFS of certain KDE packages (thanks, LaMont Jones)
+ * patch #049: new; remove duplicate XShmGetEventBase() prototype from XShm.h
+ (Closes: #182727)
+ * patch #052: new; have Xinerama header file #include Xlib header file,
+ since it references the Display structure (thanks, Xavier Bestel)
+ (Closes: #145740)
+ * patch #053: new; patch from Lobos Lunak via Adrian Bunk to fix memory leak
+ in Xlib's _XTextPropertyToTextList() function (Closes: #184817)
+ * patch #078: new; apply part of an xterm patch that was applied and
+ reverted long ago: set window size from xterm instead of child process to
+ avoid race condition with resize. (Keith Packard) (Closes: #39964)
+ * patch #084: new; fix typo in patch to XF86Config.man
+ * patch #454: new; turn off Int10 support for IA-64, apparently when this is
+ mixed with GCC 3.2, bad things happen, per Bdale Garbee
+ * patch #500: resynced with #454
+ * patch #910: new; change X server's "can't find default font" and "can't
+ find cursor font" error messages to whack users with a big fat clue
+
+ * debian/MANIFEST.*: updated to reflect new libxkbfile_pic.a object
+ * debian/*.conffiles: deleted; generation of these files is now handled
+ automatically by dh_installdeb when DH_COMPAT >= 4
+ * debian/*.debugfiles: renamed to *.install
+ * debian/compat: new file, implementing debhelper recommended best practice
+ for setting compatibility level; compat-level bumped from 2 to 4: this
+ means that, among other things, dh_installdeb creates DEBIAN/conffiles
+ files for me, ensuring they stay in sync with what is shipped
+ (Closes: #182505)
+ * debian/control:
+ - bump Standards-Version from 3.5.7 to 3.5.10 (no changes required)
+ - Build-Depend on flex-old instead of flex, since flex recently broke
+ backwards compatibility (fixes FTBFS on any box with a recent flex)
+ - minor tweak to xserver-xfree86's extended description
+ * debian/local/FAQ:
+ - removed reference to XFree86 FAQ; in the words of upstream
+ <URL:http://www.xfree86.org/FAQ/>, "The old XFree86 FAQ has been removed
+ because it is too out of date to be useful."
+ - added new question: "I just upgraded the X server and it doesn't work;
+ also, I'm using Matrox's proprietary "mga_hal" driver module or NVidia's
+ proprietary "nvidia" driver module; what's going on?"
+ - added new question: "I'm using a Macintosh, and while the keyboard works
+ correctly on a virtual terminal, in X it's all messed up. What's the
+ matter?"
+ - added new question, courtesy of Matthew Arnison: "Why do some programs
+ give authorization errors even when they are running as the same user
+ and on the same machine as the X server?"
+ * debian/local/XFree86-FAQ.html: deleted; see above
+ * debian/local/xterm.faq.html: updated from revision 1.75 to 1.77
+ * debian/po/ca.po: updated Catalan debconf template translations (thanks,
+ Ivan Vilata i Balaguer) (Closes: #183317,#183322)
+ * debian/po/{da,de,gl,it,ja,nl,pl,sv}.po: updated debconf template
+ translations for several languages courtesy of Denis Barbier, after a
+ buggy version of po-debconf eviscerated them (Closes: #170591)
+ * debian/po/es.po: updated Spanish debconf template translations (thanks,
+ Javier Fernandez-Sanguino Pe�(Closes: #186147)
+ * debian/po/fr.po: updated French debconf template translations (thanks,
+ Christian Perrier) (Closes: #185708)
+ * debian/po/ru.po: updated Russian debconf template translations (thanks,
+ Serge Winitzki) (Closes: #182701)
+ * debian/rules:
+ - removed DH_COMPAT per debhelper recommended best practice (see above)
+ - remove calls to dh_undocumented
+ - add calls to dh_install --sourcedir=debian/tmp
+ - comment out hand-rolled "idempotency" code, since dh_install and dh_link
+ behave idempotently when DH_COMPAT=4 (thanks, Joey Hess)
+ - restructure, clean up, and add comments to section that defines
+ variables for use in controlling the upstream build environment
+ + use dpkg-parsechangelog to get the version number of the source
+ package
+ + fix (and rewrite) the logic that determines whether this is an
+ official package build
+ + only define the BuilderEMailAddr Imake variable if we're doing an
+ official package build
+ - expand @OFFICIAL_BUILD@ from $(OFFICIAL_BUILD) in maintainer scripts, as
+ we do with @SOURCE_VERSION@ from $(SOURCE_VERSION)
+ - stop generating and shipping the XFree86 FAQ (see above)
+ - remove (commented-out) invocations of dh_suidregister and
+ dh_installmanpages per debhelper recommended best practice
+ * debian/shell-lib.sh:
+ - internal_errormsg(): expand the message that is spewed, but only utter
+ it if the package was part of an official build
+ - check_symlinks_and_bomb(): given the past year's worth of reports to
+ debian-x, it appears that this function only ever generates errors when
+ someone is installing very old or non-policy-compliant packages on the
+ system at the same time XFree86 is being upgraded. Simplified the error
+ messages and called new function find_culprits() when this happens.
+ - find_culprits(): new function that finds Policy-violating overlaps with
+ the XFree86 packages' symlinks, and reports these to the user
+ * debian/xfree86-common.config.in:
+ - change test for official package version to just use the new
+ $OFFICIAL_BUILD variable
+ * debian/xfree86-common.doc-base.xfree86-faq: deleted; see above
+ * debian/xfree86-common.docs: update to stop installing XFree86 FAQ (see
+ above)
+ * debian/xserver-common.README.Debian: updated to remove mention of XFree86
+ FAQ (see above)
+ * debian/xserver-xfree86.config.in:
+ - updates to handling of the use_fbdev question:
+ + treat an empty /proc/fb the same as a nonexistent one, and clarify the
+ corresponding debugging message (Closes: #182704)
+ + parse the contents of /proc/fb (first reported framebuffer device
+ only) so as to treat "VESA" and "OFfb" framebuffer types the same as
+ no kernel framebuffer at all, since these (and possible some other)
+ framebuffer types do not work with the "UseFBDev" option (chipset
+ specific drivers require facilities of a framebuffer device which the
+ generic ones don't provide); also, clarify the corresponding debug
+ message (thanks, Michel D�er)
+ - remove "atimisc", "r128", and "radeon" sub-modules from autogenerated
+ list of video driver modules
+ - migrate existing debconf settings of
+ xserver-xfree86/config/device/driver from "atimisc", "r128", or "radeon"
+ to "ati"
+
+ -- Branden Robinson <branden@debian.org> Sun, 25 May 2003 03:51:59 -0500
+
+xfree86 (4.2.1-6) unstable; urgency=medium
+
+ * RELEASE NOTE: the ELF object loader in the XFree86 X server does not
+ recognize the new symbol relocations used by GCC 3.2 on the IA-64
+ architecture; this means that the "xserver-xfree86" package will not work
+ on IA-64. IA-64 users, put your "xserver-xfree86" package on hold, or use
+ "xserver-xfree86-dbg" instead, until someone is found to patch
+ xc/programs/Xserver/hw/xfree86/loader/elf*.
+
+ * patch #000_post421: resynced with xf-4_2-branch as of 2003-02-18
+ + Fix problems with earlier fix [...] to intermittent XDarwin crash when
+ changing cursors with dual processors (Torrey T. Lyons).
+
+ * patch #000_stolen_from_HEAD:
+ + removed patches to Xlib (xc/lib/X11); see below
+ + (xc/config/cf/Library.tmpl): update and add targets to support proper
+ static compilation of Xlib (Egbert Eich)
+ + (xc/config/util/{gccmdep.cpp,mergelib.cpp,mkdirhier.sh}): add sanity
+ checks (Egbert Eich)
+ + (xc/include/Xos_r.h): warning fixes (Marc Aurele La France)
+ + (xc/programs/Xserver/hw/xfree86/loader/{elfloader.c,loader.c}): Make
+ loader ignore ELF objects that don't define any symbols. (H. J. Lu)
+ + (xc/programs/Xserver/hw/xfree86/loader/elfloader.c: warning fix (Marc
+ Aurele La France)
+ + (xc/programs/Xserver/hw/xfree86/loader/{elf.h,elfloader.c}): Add new
+ relocation entries for Alpha architecture to the loader which are used
+ when compiling with gcc > 3.1 (#A.891, Thorsten Kranzkowski)
+ (Closes: #178496)
+ + (xc/lib/ICE/connect.c): Fix SEGV in ICE library (Petter Reinholdtsen).
+ + (xc/lib/Xmu/EditresCom.c): XEditResCheckMessages trashed XtMalloced
+ memory on some widget hierarchies, due to using incorrect indexes when
+ removing duplicates (A.1409, Rob Arthan).
+ + (xc/programs/Xserver/xkb/ddxBeep.c):
+ - Fix the type/init value for some "Generation" variables. (David Dawes)
+ - Fix the number of beeps when using AccessX and XKB (#A.1302, Alan
+ Coopersmith).
+ - Warning fixes (Marc Aurele La France)
+ - Don't enter an infinite loop setting a new timer when one expires.
+ Problem easy to reproduce, press Alt+Shift+NumLock to enable/disable
+ MouseKeys and it will never stop beeping, even switching to the
+ console does not stop it. The problem also happens when enabling beeps
+ in accessx to notice state changes. (Paulo Cesar Pereira de Andrade)
+ + (xc/programs/Xserver/hw/sun/sunInit.c): Fix SEGV that occurs when Xsun*
+ runs into an unrecognized framebuffer type (Marc La France).
+ + (xc/programs/Xserver/os/auth.c): Don't change the authorization data
+ (and in particular, don't enable local host access) if the X server's
+ authority file is removed or becomes unreadable while the server is
+ running (David Dawes, reported by Dietmar Schr�).
+ + (xc/programs/Xserver/hw/xfree86/drivers/mga/mga_driver.c): Add reporting
+ of G400/G450 in MGA driver (Andrew C. Aitchison).
+ + (xc/programs/Xserver/GL/dri/xf86dri.c,
+ xc/programs/Xserver/GL/glx/glxcmds.c): Add validation for the screen
+ number parameter received over the wire by the X server's DRI extension
+ code, and fix some similar checks in the GLX code. This fixes X server
+ segfaults when an invalid screen value is provided (#A.1434, Felix
+ K�).
+ + (xc/programs/Xserver/fb/fbpict.c): Fix incorrect datatype for the pixmap
+ width in fbCompositeSrcAdd_8000x8000(), and add some small optimisations
+ (#A.1423, Keith Packard).
+ + (xc/programs/Xserver/hw/xfree86/os-support/linux/drm/xf86drm.c): Stop
+ the drm module from creating a world-writable directory in /dev. Since
+ dev nodes can only be created by root, there's no need for anyone else
+ to be able to write to /dev/dri (#5367, Brendan O'Dea).
+ + (xc/programs/Xserver/hw/xfree86/os-support/linux/drm/xf86drm.c,
+ xc/programs/Xserver/hw/xfree86/os-support/xf86drm.h): Use
+ DRM_DEV_DIRMODE again (set to 0755). (David Dawes)
+ + (xc/lib/xtrans/Xtranstli.c): Fix Xtrans TLI code to handle a port scan
+ and not fill the log file with error messages from
+ _XSERVTransTLIAccept() (#A.1333, Fiel Cabral).
+ + (xc/lib/Xt/Create.c): Clear memory allocated in xtWidgetAlloc() to avoid
+ garbage from malloc() being referenced later (#A.1114, Adam J. Richter).
+ + (xc/programs/xkbcomp/symbols.c): Fix an xkbcomp bug that prevents a later
+ definition from specifying actions if the previous definition didn't
+ (#5471, Joe Moss).
+ + (xc/programs/xkbcomp/symbols.c,xc/lib/xkbfile/xkbmisc.c): In XFree86
+ 4.3, Latin base maps can have many three- and four-level keys. Some of
+ the keys should behave as 'alphabetic' keys (should depend on the
+ CapsLock state) and others should be simple 'four level keys'. To
+ separate these two type of keys, people need to carefully specify a type
+ for each key in a symbols map. However, carelessness can can be worked
+ around if xkbcomp itself guesses the type of key. This patch adds
+ autodetection of four-level types in XKB symbols maps.
+ + (xc/programs/xkbcomp/misc.c): Xkbcomp has a check that prevents
+ inclusion of the same partial map in an include statement in different
+ single layout maps. This made sense before the multilayout mechanism
+ was added. But now, it means a complete layout can't be composed from
+ 'single layout' maps if the maps include the same file. Thus, this
+ check has ben removed.
+ + (xc/programs/Xserver/xkb/xkb.c): If some LEDs are lighting when one
+ reloads the XKB keyboard map (using setxkbmap or xf86cfg) the LEDs
+ become 'frozen' and can't be switched off. The patch fixes this bug.
+ (#5544, Ivan Pascal).
+ + (xc/programs/Xserver/xkb/{ddxConfig.c,xkbInit.c}): Fix some memory leaks
+ in xkb initialization code. (Paulo Cesar Pereira de Andrade)
+ + (xc/programs/Xserver/xkb/xkbInit.c);
+ - Fixed some memory leaks. (Egbert Eich)
+ - Reinstate the X server -xkbdir option, but only when the X server's
+ real and effective uids are the same (based on #A.1132, Ivan Popov).
+ - Fix segfault in XkbInitKeyboardDeviceStruct() (David Dawes).
+ - Fix problem in XkbInitKeyboardDeviceStruct(). Thanks to David Dawes
+ for telling me about the problem, caused by a previous patch by me. I
+ also added a comment in the code explaining the reason the strings
+ need to be duplicated. The problem is that
+ XkbInitKeyboardDeviceStruct() is an "entry point" for xkb, and the
+ function arguments are not guaranteed to be allocated, they may point
+ to static strings, command line parameters, etc. In the worst case,
+ there will be a "pseudo leak" of six strings that are used only when
+ the server is initialized; the strings are reachable from
+ xf86Info.xkbxxx, or equivalent data structure for xwin, xnest, etc.
+ (Paulo Cesar Pereira de Andrade)
+ + (xc/programs/xauth/process.c):
+ - Various warning fixes. (Marc La France)
+ - Error out if memory cannot be allocated for authfilename. (Egbert
+ Eich)
+ - Fix a bug where xauth may write an incomplete authority file and
+ delete the old one if there is insufficient disk space (#5612,
+ Harald Hoyer). (Closes: #36184)
+ + (xc/lib/xkbfile/xkbtext.c): Correct problem when the server prints the
+ name of a 'private xkb action' in a format that xkbcomp cannot
+ understand, and was causing problems when calling XkbGetKeyboard (Ivan
+ Pascal).
+
+ * patch #000_stolen_from_HEAD_ati_driver:
+ + (xc/programs/Xserver/hw/xfree86/drivers/ati/atimach64.c): When the mode
+ on server entry is found to be using composite sync on a Mach64 variant,
+ turn on the "compositesync" option (Marc La France).
+ + (xc/programs/Xserver/hw/xfree86/drivers/ati/{atipreinit.c,ativersion.h,
+ ativga.c}): Fix atimisc panel support bug that occurs when the mode on
+ server entry is a VGA mode with large horizontal and/or vertical
+ blanking pulses (Marc La France).
+
+ * patch #000_stolen_from_HEAD_neomagic_driver:
+ + Enabling 2D accel in 24-bit for neomagic 2360 and 2380 (Egbert Eich).
+ + Added support for builtin 1024x480 mode of Sony subnotebooks for
+ Neomagic driver (Egbert Eich).
+ + Fixed support for lowRes (320x240) modes in neomagic driver
+ (Egbert Eich).
+ + When enabling PCI adapters, also enable their bus mastering capability;
+ Consequently, deprecate xf86EnablePciBusMaster() (Marc La France).
+ + Fix 640x480 modes in neomagic driver (Egbert Eich).
+ + Fix an incorrect check for the "StrangeLockups" option in the neomagic
+ driver (#5505, Mike Harris).
+
+ * patch #000_stolen_from_HEAD_newport_driver: new;
+ + Support for 24bpp, fix screen blanking and bug fixes for newport driver
+ (#5329, #5349, Guido Guenther).
+ + Newport driver updates, including:
+ - Add hardware cursor support.
+ - Workaround blank console after VT switch on some newports
+ (based on a patch from Dominik Behr).
+ - Support for the Indigo2 XL (based on a patch from Adrian
+ Schroeter).
+ (#5507, Guido Guenther).
+ + Initializing clock ranges in several drivers to get rid of the
+ annoying 'scaled from 0.0 MHz' message in xf86PrintModes()
+ (Egbert Eich).
+
+ * patch #000_stolen_from_HEAD_savage_driver: (savage_driver.c): No need to
+ softboot twice; reduces startup time. (Alan Hourihane)
+
+ * patch #000_stolen_from_HEAD_xaw: new; moved existing Xaw patches from
+ 000_stolen_from_HEAD here, and added the following:
+ + (AsciiSrc.c,MultiSrc.c): Silently ignore the useStringInPlace resource
+ if the string resource was not specified. This can be easily generated
+ when calling a program using the Xaw text widget. This is a very old
+ problem, and happens with any version of Xaw and Xaw variants. This
+ does not completely fix a problem with the multiSrcObject, if both,
+ string and useStringInPlace are specified, but avoids programs crashing
+ when, for example, calling it as: some-program -xrm '*useStringInPlace:
+ 1' Thanks to Chisato Yamauchi for pointing the problem. (Paulo Cesar
+ Pereira de Andrade)
+ + (MultiSrc.c): improperly checks open() return value (A.1415, Jaromir
+ Dolecek), Fix provided by Hideo Saito in NetBSD. (Paulo Cesar Pereira de
+ Andrade)
+
+ * patch #000_stolen_from_HEAD_xdm: new;
+ (xc/programs/xdm/{Imakefile,access.c.auth.c,dm.c,policy.c,resource.c,
+ sessreg.c.,xdmcp.c}):
+ + xdm fixes from Sebastian Krahmer (#5511):
+ - non exploitable overflow in ReadWord,
+ - use snprintf() everywhere,
+ - try to make DNS spoofing impossible.
+ (xc/programs/xdm/Imakefile): Don't build xdm greeter as a module when
+ HasSharedLibraries is NO. (David Dawes)
+ (xc/programs/xdm/daemon.c): libc5 build fix (Marc La France)
+ (xc/programs/xdm/{error.c,greeter/{Login.c,greet.c}}):
+ + dont't use stdio or exit() in signal handler. (Matthieu Herrb)
+ + On OpenBSD passwords can be more than 8 chars long (Terence M. Rokop).
+ + Warning fixes (Marc La France)
+ (xc/programs/xdm/resource.c):
+ + Remove OS/2 specific stuff and fix a typo (Holger Veit)
+ + define DEF_RANDOM_FILE as "/dev/urandom" if the "linux" symbol is
+ defined (Egbert Eich)
+ (xc/programs/xdm/config/Imakefile):
+ + Fix 'su' usage for FreeBSD in the default xdm config (Eric Anholt).
+ + Warning fix (Matthieu Herrb)
+ (xc/programs/xdm/greeter/verify.c):
+ + libkafs is in not needed for kerberos IV support (Matthieu Herrb)
+ + explicit cast of NULL to void * in va_list (Matthieu Herrb).
+ + Make xdm check the full password string (on Linux) for locked
+ accounts rather than just the first character (#5485, Mike Harris,
+ Nalin Dahyabhai).
+
+ * patch #000_stolen_from_HEAD_xlib: new; moved existing Xlib patches from
+ 000_stolen_from_HEAD here, and added the following:
+ + (GetColor.c,LookupCol.c,ParseCol.c): Don't pass a transformed color name
+ to the Xserver, if Xcms.txt is installed and there is an entry for a
+ given name, the Xcms routines would convert for example "gray70" to
+ "tekhvc:0.0/70.0/0.0", which does not exist in rgb.txt. This patch just
+ remembers the argument string, and if Xcms does not return success, pass
+ the original string to the Xserver request. The string returned by
+ _XcmsResolveColorString() is a pointer to a string in a static table in
+ cmsColNm.c:pairs. (Paulo Cesar Pereira de Andrade)
+ + (GetDflt.c,PutImage.c,ImThaiFlt.c,lcEuc.c,lcGenConv.c,lcJis.c,lcSjis.c):
+ warning fixes (Marc Aurele La France)
+ + (ICWrap.c): Prevent core dumps in Xutf8Reset/Lookup code of Xlib (Masaru
+ Yokoi).
+ + (Imakefile,XlcDL.c,xlibi18n/Xi18nLib.conf): miscellaneous cleanups,
+ including restriction of the _MACH64_NAME weirdness in XlcDL.c to
+ UltraSPARCs only (Egbert Eich) (Closes: #178289)
+ + (Imakefile,XlibInt.c): Remove inconsistencies in the way extension headers are
+ referred to (David Dawes).
+ + (Imakefile): Fixed Xlib build rules so it builds a truly static library
+ again (Egbert Eich).
+ + (XDefaultOMIF.c): pacify GCC 3.2 (Marc Aurele La France)
+ + (XErrorDB): synced with HEAD; while I did not backport the RENDER and RandR
+ extensions themselves from HEAD, I thought it might be useful if Debian's
+ Xlib could recognize the protocol errors from these extensions.
+ - recognize XF86VidModeGetPermissions request
+ - fix incorrect labeling of two vidmode extension error codes
+ (XF86VidModeExtensionDisabled, XF86VidModeClientNotLocal)
+ - recognize XFree86-Misc requests XF86MiscSetGrabKeysState,
+ XF86MiscSetClientVersion, XF86MiscGetFilePaths
+ - recognize RENDER extension requests and errors
+ - recognize RandR extension requests
+ + (XKB.c): The functions XGetXlibControls and XSetXlibControls must work
+ even if the X server doesn't have the XKB extensions (#5513, Ivan
+ Pascal).
+ + (XKBCtrls.c): XkbSetControls should return True when successful, not False
+ (#A.1385, Stephen Montgomery-Smith).
+ + (XKBCvt.c,lcCT.c,lcUTF8.c,lcUniConv/iso8859_11.h): Add support for
+ th_TH.ISO-8859-11 (#5327, Theppitak Karoonboonyanan).
+ + (XKeysymDB): resynced with HEAD; seems harmless to recognize some new
+ keysyms.
+ - Remove XFree86-specific keysyms that already have generic equivalents
+ (Menu, Redo, Print, Undo).
+ - Add private keysyms for the new XKB actions (XFree86 special keys), and
+ move those actions to an XKB compatibility map. This fixes side-effects
+ (loss of auto-repeat and mousekeys functionality) of the way this was
+ implemented previously (Ivan Pascal).
+ - Add extended symbols for the "Logitech Cordless Desktop Navigator"
+ keyboard (#A.1480, Gilbert Fridgen).
+ - Add "Inet" key definitions for the Trust Direct Access keyboard
+ (#A.897, Rapha�Poss).
+ - Added support for special keys found on many ACPI control, Easy Access
+ Keyboards, Internet keyboards, laptops, notebooks and PDA (via
+ xkb/symbols/inet). Added new key symbols for these keyboards (Stanislav
+ Brabec).
+ + (XlcDL.c):
+ - Fix bad assumption that SPARCs are the only architectures that can have
+ the _LP64 symbol in the build environment. (Egbert Eich)
+ - Have fetch_symbol() return NULL if a NULL symbol is requested. (Egbert
+ Eich)
+ + (XlibInt.c):
+ - Protect "cvl" references in recent changes with '#ifdef XTHREADS'
+ (reported by Dan Holm).
+ - Fix a problem in _XEventsQueued() that causes an event reader lock of
+ another thread to be released. This can cause a hang and keyboard
+ lockup in KDE (#5619, Ewald Snel).
+ + (Xutil.h): ISO_Level3_Shift was breaking compose sequences because Xlib
+ wasn't recognising it as a modifier key (#5594, Ivan Pascal).
+ + (imCallbk.c): Only free cbs.data.text when cbs.type is XIMTextType in
+ _XimStatusDrawCallback() (#A.1342, Owen Taylor).
+ + (imDefIm.c): Make XIM locale checking case-independent (based on #A.1422,
+ Leon Ho).
+ + (imLcPrs.c): Reduce the stack usage from 64k to 8k in imLcPrs.c, which
+ fixes a problem with Java for FreeBSD (#5568, Eric Anholt).
+ + (imThaiFlt.c):
+ - Fix a bug in previous Thai XIM changes (#5563, Theppitak Karoonboonyanan).
+ - Thai XIM fixes:
+ - Fix the arguments in calling to StringConversionCallback in Thai XIM
+ filter, according to Hideki Hiura's explanation on the protocol.
+ - Add input sequence correction capability to the Thai XIM by
+ exploiting the XIMStringConversionSubstitution operation.
+ (#5553, Theppitak Karoonboonyanan).
+ - Add support for th_TH.UTF-8. with XIM (#5336, Theppitak Karoonboonyanan).
+ - Fix some bugs in the Thai XIM:
+ - excessive '(' char when switching XKB keyboard group
+ - ineffective Ctrl-<key> and Alt-<key>
+ (#5335, Theppitak Karoonboonyanan).
+ + (imTrX.c): Fix a crash that can happen when some apps are run in CJK
+ locales (#5489, Havoc Pennington).
+ + (lcEuc.c,lcJis.c,lcSjis.c): Allow UTF8 conversion to work for Japanese
+ locales (#A.1527, Etsushi Kato).
+ + (lcFile.c):
+ - clean up _XlcLocaleDirName() (Egbert Eich)
+ - Fix some memory leaks in libX11 i18n code (#A.1314, Olivier Chapius).
+ + (omGeneric.c):
+ - clean up code and improve comments in Oliver Chapuis's fixes (Egbert Eich)
+ - fix memory leaks and speed font loading (Oliver Chapuis)
+ + (omText.c): Fixed obvious typo in OMlib (Egbert Eich).
+ + (xlibi18n/Xi18nLib.conf): reindent preprocessor statements (Egbert Eich)
+
+ * patch #000_stolen_from_HEAD_xtt: new; several fixes nicked from CVS HEAD
+ to fix crashing problems et al.; this does *not* include the
+ FontFilePriorityRegisterRenderer() change (and corresponding revision
+ number bump), which requires changes elsewhere. (Closes: #179731)
+ + Fixed a core dump problem in libXtt (?) (Egbert Eich). [That's the
+ XFree86 CVS commit message / changelog entry, but it sucks. Here's a
+ better one:] "With xtt backend, loading a proportional iso10646 font
+ always results in X server crash. This is also responsible for some of
+ the mysterious mozilla crashing X problems. Because mozilla some times
+ load iso10646 font even without user's explicit request. The problem is
+ caused by the following code inside the "freetype_compute_bounds"
+ function in xc/extras/X-TrueType/xttfuncs.c (between line 1073 and
+ 1114). The [...] code loop through all characters in an encoding to get
+ their metrics. If the very first "c" is 0, there's a problem of
+ referencing null pointer "tmpchar" at where I marked with ^. For most of
+ the encodings in xtt implementation, character does not start from 0.
+ Iso10646 is one of a few that does, thus manifest the problem. The
+ problem exists in all versions of 4.x."
+ + Fix for two problems where the combination of xtt and XAA could result
+ in a crash. The first problem was xtt returning a NULL pointer for the
+ bitmap even though the height was non-zero. The second problem was
+ glyph ascent/descent exceeding the range of maxbounds.ascent/
+ maxbounds.descent (A.1484, Chisato Yamauchi, Masanori Shimada).
+ + Fix an XTT problem where garbage is displayed when there are undefined
+ glyphs for fixed width fonts, and fix some other potential problems
+ (#A.1541, Hidetoshi Yamanouchi, Chisato Yamauchi).
+ + Improve the XTT code that prevents unexpected data to be passed to
+ XAA (#5600, Chisato Yamauchi).
+ + [...] fix a missing portion in X-TT's Big5HKSCS-Unicode table (#5362,
+ Branden Robinson and ThizLinux Laboratory Ltd).
+ + Fix a crash in the "xtt" font module caused by a NULL dereference when
+ the font cache becomes full (#A.1611, Chisato Yamauchi and
+ tokeijikakenoringo).
+
+ * patch #002,011,023,041,046,067,069,072,082,087,091,100,351,905: resynced
+ with above patches from HEAD, and dropped patches merged upstream
+ * patch #025,071,074,600: deleted; superseded by #000_stolen_from_HEAD
+
+ * patch #002:
+ - (xdm.man) Use dot macros instead of backslash roff sequences to mark up
+ literals that use __projectroot__, because cpp won't recognize them
+ otherwise.
+ * patch #009: patch to s3 driver to add support for S3 Trio64V2/DX and /GX
+ chips, courtesy of Santiago Garcia Mantinan and Javier Moran Rua
+ (Closes: #178600)
+ * patch #037: declare drmFreeBufs and drmR128drmR128TextureBlit symbols in
+ the r128 subdriver's symbol lists
+ * patch #046: build static PIC version of Xinerama library (Closes: #175996)
+ * patch #048: update X, XFree86, and xmodmap manpages to use dot macros
+ instead of backslash roff sequences to mark up literals that use
+ __projectroot__, because cpp won't recognize them otherwise.
+ * patch #084,086,091,909: resynced with #048
+ * patch #086: overhaul XFree86 manpage
+ * patch #092: new; patch for ATI Radeon 9100 support, courtesy of Wesley W.
+ Terpstra
+ * patch #103: add missing semicolon to patch for sun keycodes file; d'oh!
+ (thanks, Daniel van Eeden) (Closes: #178465)
+ * patch #905: removed portions of this patch that have been merged upstream
+ * patch #906: re-order xfs's default font catalogue to follow advice given
+ in README.fonts document; move scalable font directories to top of list
+ and remove ":unscaled" lines
+ * patch #909: Debianize XF86Config manpage references in XFree86 manpage to
+ XF86Config-4 (Closes: #180195)
+
+ * debian/MANIFEST.*: updated to reflect new libXinerama_pic.a object
+ * debian/control:
+ - fix lingering problems with xlibmesa-{gl,glu} split; xlibmesa-glu-dev
+ depends on xlibmesa3-glu and xlibmesa-gl-dev | libgl-dev (thanks, Aaron
+ M. Ucko) (Closes: #178310)
+ - dropped xlib6g and xlib6g-dev pseudopackages; these packages are now
+ completely dead
+ * debian/local/dexconf: reorder font path components to follow advice given
+ in README.fonts document; move scalable font directories to top of list,
+ and, for 4.x XF86Config files, remove ":unscaled" lines since they are not
+ needed by XFree86 4.0 and later. Also, add CID font path to 4.x XF86Config
+ files for consistency with 3.x font path configuration.
+ * debian/po/pt_BR.po: updated debconf translations for Brazilian Portuguese
+ (thanks, Andre Luis Lopes) (Closes: #179352)
+ * debian/xlibmesa3-glu.shlibs: fix thinko in shlibs line; "libgl1" to
+ "libglu1" (thanks, Aaron M. Ucko) (Closes: #178374)
+ * debian/xlibs-pic.files*: ship new libXinerama_pic.a object
+ * debian/xserver-xfree86.config.in: updated to work with both Discover 1.x
+ and Discover 2.x
+
+ -- Branden Robinson <branden@debian.org> Tue, 25 Feb 2003 18:01:32 -0500
+
+xfree86 (4.2.1-5) unstable; urgency=low
+
+ * Thanks to ISHIKAWA Mutsumi for authoring, identifying, and otherwise
+ helping out with patches to fix bugs in the ServersOnly build when X
+ library headers aren't installed in the build environment.
+ (Closes: #170508,#170510)
+
+ * patch #000_stolen_from_HEAD:
+ + dropped patch "Fix the number of beeps when using AccessX and XKB
+ (#A.1302, Alan Coopersmith)" since it causes the server to beep nonstop
+ until it's restarted if you press Shift-NumLock to enable PointerKeys.
+ ("xset b off" shuts it up, but it's just beeping silently, as a
+ subsequent "xset b on" reveals) (Closes: #171294)
+ + GLU and GLw aren't extension libraries, so key their default build
+ status to BuildLibraries. (David Dawes)
+ + BuildServersOnly fix when building on a system with no installed X
+ headers or libraries (#5559, ISHIKAWA Mutsumi).
+ + (xc/lib/xkbfile/xkberrs.c): build fixes (Alan Hourihane)
+
+ * patch #000_stolen_from_HEAD_savage_driver: new;
+ + Fixed memory leaks in S3 Savage driver (Egbert Eich).
+ + Savage driver updates (#5203, Tim Roberts).
+ + [...] xf86SetOperatingState() call cleanups in most drivers
+ (Marc La France).
+ + Savage driver updates.
+ - Support for ProSavage DDR (PCI ID 0x8D04)
+ - TV output fixes
+ - support for -pixmap24 flag
+ - XAA fixes to cure lockups
+ (#5412, Tim Roberts).
+ + Fix Mono8x8 code in savage driver, needed ROP_NEEDS_SOURCE, also
+ add NO_PLANEMASK to ScreenToScreen as code doesn't handle that case
+ (Alan Hourihane).
+ + Remove "Option VideoRam" from savage driver, there's no need for
+ it, and utilize the entities given VideoRam size (Alan Hourihane).
+
+ * patch #009: deleted; didn't do what it was supposed to, and caused other
+ Xv problems in i810 (Closes: #166234)
+ * patch #021: new; patch to startx to check for GNU hostname regardless of
+ what uname reports the OS as; first check to see if "hostname --version"
+ works at all, and then if it does check both stdout and stderr of that
+ command for "GNU" (Closes: #167212)
+ * patch #022: new; fix Xaw's StripChart widget to draw scale lines correctly
+ (thanks, Nikita V. Youshchenko) (Closes: #167448)
+ * patch #033: new; fix for ICE message replies when using subprotocols
+ (thanks, Jochen Voss) (Closes: #95951)
+ * patch #035: new; fix for null pointer dereferences in fontfile library,
+ (thanks, Nikita V. Youshchenko) (Closes: #168791)
+ * patch #041: new; patch from ISHIKAWA Mutsumi to fix missing depend target
+ in Xlib's Imakefile if "BuildServersOnly" is YES, "BuildXnestServer" is
+ NO, and "BuildGLXLibrary" is NO.
+ * patch #079: deleted; superseded by #000_stolen_from_HEAD_savage_driver
+ * patch #102: deleted; if Sun Type6 keyboards use the same scancodes as
+ Type5 keyboards to mean different things, then much more of a patch than
+ this is needed. Fixes FTBFS problem on sparc. (Closes: #170509)
+ * patch #203: fix for data type size issues in Xpm library on Alpha (thanks,
+ Kalle Olavi Niemitalo) (Closes: #174282)
+ * patch #909: change XF86Config-4(5x)'s NAME section to refer to itself as
+ XF86Config-4 instead of XF86Config so that "man XF86Config" brings up the
+ section 7 manpage preferentially
+ * patch #004,008,046,085: resynced with other patches
+
+ * debian/control:
+ - add Build-Dependency on libstdc++5-dev (for libGLU in xlibmesa3-glu)
+ - add dependency on debconf to xfree86-common
+ - remove dependency on debconf from xbase-clients, xfs, xlibmesa3, and
+ xterm (see below)
+ - add dependency on xprt-common to xprt, and update its extended
+ description to describe its broken state and available alternative
+ (thanks, Drew Parsons) (Closes: #170678)
+ - C++ ABI 1.02 transition / Mesa 5.0 awareness:
+ + xlibmesa3, xlibmesa3-dbg, xlibmesa-dev split into:
+ xlibmesa3-gl, xlibmesa3-gl-dbg, xlibmesa-gl-dev,
+ xlibmesa3-glu, xlibmesa3-glu-dbg, xlibmesa-glu-dev (Closes: #175157)
+ + xlibmesa3-gl conflicts with and provides libgl1 (pure virtual package)
+ + xlibmesa3-gl replaces mesag3 and xlibmesa3
+ + dropped old replacement of xserver-xfree86 (<< 4.0.2) by xlibmesa3;
+ those versions of xserver-xfree86 are long dead and were never part of
+ a stable Debian release
+ + xlibmesa3-gl-dbg depends on exact same version of xlibmesa3-gl
+ + xlibmesa3-gl-dbg conflicts with and replaces xlibmesa3-dbg
+ + xlibmesa-gl-dev conflicts with and provides libgl-dev (pure virtual
+ package)
+ + xlibmesa3-glu conflicts with and provides libglu1 (mixed virtual
+ package)
+ + xlibmesa3-glu replaces mesag3 (<< 5.0.0-1) and xlibmesa3
+ + xlibmesa3-glu-dbg depends on exact same version of xlibmesa3-glu
+ + xlibmesa3-glu-dbg conflicts with and replaces xlibmesa3-dbg
+ + xlibmesa-glu-dev depends on libstdc++5-dev and libc6-dev | libc-dev
+ + xlibmesa-glu-dev conflicts with xlibmesa-dev and libglu-dev (pure
+ virtual package)
+ + xlibmesa-glu-dev provides libglu-dev (pure virtual package)
+ + xlibmesa-glu-dev replaces mesag-dev (<< 5.0.0-1) and xlibmesa-dev
+ + dropped dependency on xlibmesa3 from xlibosmesa3 (was redundant
+ anyway, shlibs files work fine within the build)
+ + added dependency on xlibs-dev to xlibosmesa-dev, since libOSMesa links
+ against Xlib and Xext
+ + x-window-system-core metapackage depends on xlibmesa3-gl and
+ xlibmesa3-glu instead of xlibmesa3
+ + xlib*mesa* package descriptions tweaked
+ - added xlibmesa3, xlibmesa3-dbg, and xlibmesa-dev pseudopackages
+ - added (<< 4.2.1-5) versioning to Conflicts and Replaces relationships on
+ old xlibmesa3, xlibmesa3-dbg, and xlibmesa-dev packages
+ - added missing versioning (<< 4.0) on xutils's replacement of xlib6g-dev
+ - updated extended descriptions of x-window-system{,-core}
+ - fixed typo in xprt's extended description
+ - bump xdm and xfree86-common's versioned dependencies on debconf to >=
+ 1.2.9 on advice from lintian
+ - uncomment call to dh_installinit in binary-indep rule, since
+ xfree86-common now has an init script
+ * debian/rules: add "-DXnestServer=NO -DXVirtualFramebufferServer=NO
+ -DXprtServer=NO" to SERVERDEBUG_IMAKE_DEFINES (thanks, ISHIKAWA Mutsumi)
+ * debian/shlibs.local: regenerated from debian/*.shlibs
+ * debian/xdm.options: identify correct location of this conffile in comment
+ * debian/xdm.options.5: cosmetic and roff-stylistic fixes
+ * debian/xfree86-common.init: new init script which runs in runlevel S that
+ simply ensures that the X server's Unix domain socket directory exists and
+ has proper ownership and modes (N.B., this does not start any processes)
+ (Closes: #170638)
+ * debian/xfree86-common.{postinst,postrm}.in: invoke update-rc.d and run the
+ package's init script on package configuration; invoke update-rc.d on
+ package purge
+ * debian/xfs.init: when starting the daemon, ensure the font server's socket
+ directory exists and has proper ownership and modes
+ * debian/xfs.options: identify correct location of this conffile in comment
+ * debian/xfs.options.5:
+ - identify correct location of the documented conffile (Closes: #170513)
+ - cosmetic and roff-stylistic fixes
+ * debian/xlibmesa*: debhelper and shlibs files updated to reflect -gl and
+ -glu package split
+ * debian/xlibs.shlibs: update Xrender dependency info to xlibs to (>>
+ 4.2.0), since Xrender added symbols between XFree86 4.1.0 and 4.2.0
+ (thanks, Eric Dorland) (Closes: #174589)
+ * debian/{xbase-clients,xfs,xlibmesa3,xterm}/postinst.in: nuke quaint and
+ curious relics of forgotten debconfage (don't source the confmodule since
+ we're doing anything debconf-related) (also, remove pointless invocation
+ of db_stop in xfs's postinst)
+ * debian/xserver-common.files: the dexconf manpage is in section 1, not 8
+
+ -- Branden Robinson <branden@debian.org> Tue, 21 Jan 2003 12:02:40 -0500
+
+xfree86 (4.2.1-4) unstable; urgency=low
+
+ * patch #000_post421: resynced with xf-4_2-branch as of 2002-11-09
+ + Bump XDarwin version. (Torrey T. Lyons)
+ + Forgot to commit updated RELNOTES at top level for 4.2.1. (David Dawes)
+ + Bump the "snap" version for Mac OS X 10.2.2. (David Dawes)
+ + Fix bad link in upgrade information. (David Dawes)
+
+ * patch #000_stolen_from_HEAD:
+ + Fix overlay bug. ClipNotify wasn't being called in some corner cases.
+ (Alan Hourihane)
+ + Fix pixmap cache corruption on VT switches in mga driver (#A.1279, Ross
+ Mikosh).
+ + Fix application crash when using DPS client libraries (#A.1282, Frank
+ Giessler).
+ + Add missing Increment option in Summa input driver (#A.1276,
+ G.P.Halkes).
+ + Fix the number of beeps when using AccessX and XKB (#A.1302, Alan
+ Coopersmith).
+ + Add symbols that were missing from the registrations lists. This
+ removes a lot of the misleading "unresolved symbols" messages that show
+ up if 'X -configure' crashes. (David Dawes)
+ + [fbdev] The xf24_32bpp module isn't needed when using fb. (David Dawes)
+ + Fix an fbdev driver crash when the VT isn't active (#5169, Michel D�er).
+ + Fix 24bpp displays with the fbdev driver when not using shadowfb (#5351,
+ Michel Lanners).
+ + When parsing XInput XF86Config options, catch NULL option table instead
+ of crashing. (David Dawes)
+ + Fix a double free() that can cause a crash in XCloseIM() (based on
+ #5303, Mo DeJong).
+ + Fix for random crashes when a screen fails to initialise (Marc La
+ France).
+ + Fix server crash when using Xinerama when application closes. (#A.808,
+ Tsukahara Ken).
+ + Fix a SEGV on server reset in the PM3 driver. (Alan Hourihane)
+ + Fixed GetTimeInMillis() to use deltas instead of absolute time returend
+ by gettimeofday(). This ensures time is monotonic in X (Egbert Eich).
+ (Closes: #85365)
+ + Sync with Type1 font rasterizer from HEAD:
+ - Some minor include cleanups. (David Dawes)
+ - Avoid redefining abort(), a well known C library function. This can
+ cause errors when the abort() prototype from stdlib.h is included
+ later by Xos.h. (Matthieu Herrb)
+ - A "hammer" approach at ensuring font properties are freed properly.
+ (Marc La France)
+ - Modify Type1 font RAM allocation heuristics to allow for larger fonts.
+ (#A.1121, Melchior Franz)
+ + Fix a typo in the s3 driver which prevented virtual display sizes.
+ (#A.965, Stef Voltz).
+ + Fix input size of character strings used in lib/X11/lcGetConv.c
+ (#A.828, Yong-Jhen Hong).
+ + Fix a memory leak when calling _XCloseLC and removing the loadable
+ locales (#A.855 Mark Robinson).
+ + Fix wrong check in Xaw when rebuilding the line table. If the width
+ of a line becomes smaller, it should clear to eol. (Paulo Cesar Pereira
+ de Andrade)
+ + In the X server, permit XDM-AUTHENTICATION-1 without also requiring
+ XDMCP support. (Keith Packard)
+ + Fix an initialization problem in the calcomp input driver (#A.1056,
+ Josef Walzer).
+ + Fix mode validatation to deal with video memory sizes 256MB or greater
+ (Marc La France).
+ + Fix missing declaration of XShmGetEventBase() in XShm.h (#5375,
+ Branden Robinson).
+ + Fix a build failure when BuildServersOnly is defined and XnestServer is
+ not defined, and when /usr/X11R6 is not installed. (#A.1145, Valeriy E.
+ Ushakov).
+ + Don't build libGLU when BuildServersOnly set. (Alan Hourihane)
+ + Add a 1152x768 mode suitable for Titanium PowerBooks to the set of
+ build-in modes (#5212, Michel D�er).
+ + Add some high resolution modes to the set of built-in modes (calculated
+ according to the VESA GTF). (David Dawes)
+ + Regenerate xf86DefModes.c with newly added modes. (David Dawes)
+ + Allow Xnest/Xprint to build without traversing lib/X11 subdir. (Alan
+ Hourihane)
+ + Document 'xset dpms force on'. (David Dawes)
+ + Don't assume mmap returns the pointer hint requested. This is in
+ the DENSE map code for Linux/alpha (#5439, Ray Strode).
+ + Fix a segfault when using xkbcomp's -I option, and update the man
+ page to correctly document the way this option works (#5447, Joe Moss).
+
+ * patch #000_stolen_from_HEAD_ati_driver: new; many, many bugfixes to
+ pre-Rage128 ATI chips (sorry, the r128 and radeon sub-drivers have been
+ overhauled far too much since 4.2 for me to steal those); among other
+ things:
+ + Fix IBM-reported bug that could cause PCI bus lockouts with second
+ generation integrated Mach64 controllers (Marc La France).
+ + In the ATI driver, add option "CrtDisplay" to force CRT display in a
+ dual display situation (Marc La France).
+ + In the ATI driver, rename option "CrtScreen" to "NoPanelDisplay",
+ keeping "CrtScreen" as a private option for compatibility (Marc La
+ France).
+ + An attempt at disabling display switches while the server is running on
+ an LT Pro or Mobility M1 (Marc La France).
+ + Fix rounding error in Mach64 DSP calculation (Marc La France).
+ + Add PCI ID for ATI Radeon Mobility M7 'LX' (#5233, Mike Harris). (This
+ patch moved from #000_stolen_from_HEAD.)
+ + Fix long-standing colour map bug in ATI driver (Marc La France).
+ + Fix panel support bug for ATI LTPro, XL and Mobility M1 chips (Marc La
+ France).
+ + Ignore (with messages) any HorizSync and VertRefresh specifications for
+ panels when using ATI LT, LTPro, XL and Mobility M1 (Marc La France).
+
+ * patch #000_stolen_from_HEAD_cyrix_driver: new; updated version of former
+ patch #021; this fixes support for the 5510, 5520 and 5530 -- this driver
+ only works with VSA1 based chips (#A.1260, Alan Cox).
+
+ * patch #002,905: some Debianisms moved to patch #905 where they belong;
+ also, update StorePid() function to use Matthieu Herrb's changes that
+ refuse to open an existing pid file, and take more care with the size of
+ the Pid_t type; lots more changes, including better logging
+ (Closes: #63201)
+ * patch #003,906: some Debianisms moved to patch #906 where they belong;
+ also, update StorePid() function to use Matthieu Herrb's changes that
+ refuse to open an existing pid file, and take more care with the size of
+ the Pid_t type
+ * patch #004: drop part of patch rejected by upstream which implemented a
+ "DefaultGcc2OptimizeOpt" symbol in Imake
+ * patch #021: deleted; now in #000_stolen_from_HEAD_cyrix_driver
+ * patch #022,064,078; deleted; now in #000_stolen_from_HEAD
+ * patch #029: new; include prototypes for functions provided by "Wraphelp.c"
+ in the XDMCP library's Wrap.h header file
+ * patch #034: new; document OverlayMem option in neomagic driver manpage
+ (Diego SANTA CRUZ via Bill Allombert) (Closes: #168139)
+ * patch #041: deleted; now in #000_stolen_from_HEAD
+ * patch #087,100: resynced with #000_stolen_from_HEAD_ati_driver
+ * patch #089,100: resynced with #000_stolen_from_HEAD
+ * patch #090: new; support for ATI Radeon 9000 cards (N.B., not the 9500 or
+ 9700)
+ * patch #091: new; drop the sumFile() function from xdm's genauth.c file,
+ define DEV_RANDOM for Linux and the Hurd, and change the meaning of the
+ randomFile resource to indicate the file that should be opened for the
+ reading of entropy. The default random file on Linux and the Hurd is
+ /dev/random; on NetBSD it remains /dev/urandom per the existing XFree86
+ code. Twinks who don't want xdm "draining their entropy pool" can edit
+ /etc/X11/xdm/xdm-config and set "DisplayManager.randomFile" to "/dev/zero"
+ if they want. Anyway, this fixes a bug with xdm causing machine checks on
+ some architectures when reading a lot of data from /dev/mem.
+ (Closes: #107311)
+ * patch #102,103: new; patch xf86Events.c to recognize Sun Type6 keyboard
+ scancodes, and provide XKB data for type6 keyboards; please test
+ (Closes: #100485)
+ * patch #202: new; patch from Jay Estabrook (via Falk Hueffner) that fixes
+ crashes on Alpha due to datatype confusion in XScreenSaver extension
+ (Closes: #161072)
+ * patch #313: deleted; made obsolete by patch #091
+ * patch #901: updated to fix a few compiler warnings
+ * patch #905: fix typo when declaring authDir resource (fortunately, the
+ default upstream location is a symlink to the location defined, so this
+ doesn't really change anything) (Closes: #167642)
+ * patch #907: in XTerm's color app-defaults file, use instance names, not
+ class names (see #167495)
+
+ * debian/{xfonts-pex.files,xlibs.files.alpha}: removed these obsolete files
+ (thanks, ISHIKAWA Mutsumi) (Closes: #169429)
+ * debian/MANIFEST.*: updated to reflect re-sectioning of dexconf manpage
+ * debian/{*.templates,/po}: migrated to po-debconf
+ * debian/control:
+ - add build dependency on po-debconf, and bump versioned dependency on
+ debhelper to (>= 4.1.16) for support of the New Debconf Template
+ Translation Order
+ - Discuss Xlib's loadable modules for i18n in the extended package
+ descriptions of xlibs and xlibs-dbg. In xlibs-dbg's extended
+ description, give tips on how to debug the i18n modules.
+ - Update URL to driver status document in xserver-xfree86's extended
+ description. (Closes: #168289)
+ * debian/local/dexconf:
+ - improve usage message
+ - rework handling of temporary directories and files; respect $TMPDIR
+ instead of using the more complex method we had before (Closes: #166177)
+ - be sure to quote all expansions of variables that the user can influence
+ - clean up a few diagnostic messages
+ * debian/local/dexconf.8: renamed to debian/local/dexconf.1
+ * debian/local/dexconf.1: updated; added sections on options, evironment,
+ diagnostics, asynchronous events, exit status, and future directions;
+ other minor edits
+ * debian/local/update-fonts-{alias,dir,scale}: reworked diagnostic messages
+ a bit; produce them with shell functions
+ * debian/local/update-fonts-{alias,dir,scale}.8: updated; added sections on
+ operands and diagnostics; other minor edits
+ * debian/local/xserver-wrapper.c: (cosmetic) wrapped source code at 80
+ columns
+ * debian/local/xvfb-run.1: minor cosmetic updates
+ * debian/rules:
+ - try and cut down on the time and space required by the XFree86 build
+ process; when copying the build tree for the debugging server build,
+ omit the doc, fonts, and nls directories
+ - call debconf2po-update after generating and after cleaning the
+ xserver-xfree86-dbg.templates file; I'm not crazy about having to do
+ this, but I don't see any alternative with the current po-debconf tools
+ - do another substitution (this one for $THIS_SERVER) when generating
+ xserver-xfree86-dbg's maintainer scripts, so that the package doesn't
+ have an identity crisis
+ - install dexconf manpage to section 1
+ * debian/scripts/vars.m68k: on m68k, make x-window-system-core depend on
+ xserver-xfree86-dbg instead of xserver-xfree86 since the XFree86 module
+ loader doesn't work on this arch at present; m68k hackers, please help fix
+ this!
+ * debian/xdm.init:
+ - sleep for one second after using start-stop-daemon --stop successfully
+ so that we don't needlessly signal the daemon while it's shutting down
+ - minor change; use shell arithmetic operator ("$(( ))") instead of expr
+ to do simple math
+ * debian/xdm.prerm.in:
+ - migrate the last bit of interactivity in the maintainer scripts to
+ debconf; use debconf to ask the user if he/she wants to stop a running
+ xdm daemon that is managing sessions
+ - stop de-registering an obsolete alternative
+ - stop defining an unused variable
+ * debian/xdm.templates: add template xdm/stop_running_server_with_children
+ * debian/xfs.init:
+ - sleep for one second after using start-stop-daemon --stop successfully
+ so that we don't needlessly signal the daemon while it's shutting down
+ - comsetic change; prevent words from being run together when xfs daemon
+ won't listen to SIGTERM
+ * debian/xlibs-dbg.debugfiles: ship unstripped versions of Xlib's i18n
+ modules in /usr/X11R6/lib/debug/X11/locale/common
+ * debian/xserver-xfree86.templates: add discussion of new "altwin" symbols
+ file and how to use multiple symbols files with xkboptions to
+ xserver-xfree86/config/inputdevice/keyboard/options template description
+
+ -- Branden Robinson <branden@debian.org> Mon, 18 Nov 2002 04:24:48 -0500
+
xfree86 (4.2.1-3) unstable; urgency=low
* patch #000_post421: resynced with xf-4_2-branch as of 2002-10-16
Reply to: