X Strike Force X.Org X11 SVN commit: r145 - trunk/debian
Author: branden
Date: 2005-06-05 19:22:56 -0500 (Sun, 05 Jun 2005)
New Revision: 145
Modified:
trunk/debian/changelog
Log:
Add xfree86 package changelog entries from 4.3.0.dfsg.1-7 to -14.
Modified: trunk/debian/changelog
===================================================================
--- trunk/debian/changelog 2005-06-05 23:46:35 UTC (rev 144)
+++ trunk/debian/changelog 2005-06-06 00:22:56 UTC (rev 145)
@@ -134,8 +134,1373 @@
actually reflect package release history as a Debian user would see it).
Ship this changelog in /usr/share/doc/xorg-common.
- -- Branden Robinson <branden@debian.org> Thu, 19 May 2005 23:46:32 -0500
+ -- Branden Robinson <branden@debian.org> Sun, 5 Jun 2005 19:21:53 -0500
+xfree86 (4.3.0.dfsg.1-14) unstable; urgency=high
+
+ Urgency set to high due to fix for security flaw (see below).
+
+ Changes by David Nusinow and Branden Robinson:
+
+ * Replace s_popen()-based fix for CAN-2004-0914 with a better fix from
+ freedesktop.org xorg CVS. There were several problems with s_popen(),
+ some merely functional, and some themselves security-flawed. There does
+ not appear to be a MITRE CVE candidate ID for this problem. The only
+ differences between the Xpm library code in this package release and that
+ in freedesktop.org xorg CVS HEAD as of 2005-05-26 are 1) whitespace
+ differences; 2) RCS keyword differences; 3) preprocessor directives to
+ support autotoolization of the library; 4) support for pre-ANSI C
+ compilers in function declarations; 5) removal of some dead code in
+ create.c; and 6) preprocessor-enforced ignorance of PutPixel32() on 64-bit
+ systems (whose implementation was already disabled on 64-bit systems).
+ (Closes: #308783)
+
+ Changes by Branden Robinson:
+
+ * Fix grammar error in Spanish debconf template translations (thanks, Steve
+ Langasek). (Closes: #309450)
+
+ * Grab fixes from freedesktop.org xorg CVS that make the DPMS extension
+ header (dpms.h) and Xpm library header file (xpm.h) usable with C++ code
+ by using the _XFUNCPROTO{BEGIN,END} macros instead of nothing at all (in
+ the DPMS case) or a literal 'extern "C" {' construct (in the XPM case).
+ C++ applications using the DPMS headers were likely to crash; see
+ <URL: https://bugs.freedesktop.org/show_bug.cgi?id=830 >. The DPMS fix
+ also adds a header self-inclusion guard to dpms.h.
+
+ * Invert sense of tests invoking has_multiplexed_mouse; the comments were
+ right, but the logic was exactly backwards. This should fix warp-speed
+ and left-handed configuration mouse device problems introduced (or simply
+ not fixed for USB and PS/2 mouse users of Linux 2.6) in -13. Thanks to
+ Steve Langasek for eyeballing the code and catching my thinko.
+ (Closes: #311184)
+
+ -- Fabio M. Di Nitto <fabbione@fabbione.net> Wed, 01 Jun 2005 07:01:50 +0200
+
+xfree86 (4.3.0.dfsg.1-13) unstable; urgency=high
+
+ Urgency set to high due to fix for security flaw CAN-2005-0605 (see below).
+
+ Changes by Denis Barbier:
+
+ * Update Spanish debconf template translations (thanks, Javier
+ Fernández-Sanguino Peña). (Closes: #298538)
+
+ * Revert XLC_LOCALE/zh_TW.UTF-8 to the version shipped in 4.3.0.dfsg.1-6.
+ Tetralet explained that Chinese users are likely to have the unifont
+ package installed because it is part of Chinese tasks, and this font
+ covers all needed glyphs. With this font, having ISO10646 listed first
+ in this file is harmless, whereas GTK1.2 apps are unreadable when it is
+ at the bottom of this file. (Closes: #296719)
+
+ Changes by Branden Robinson:
+
+ * Fix CAN-2005-0605: libxpm4's scan.c file may allow attackers to execute
+ arbitrary code via a negative bitmap_unit value that leads to a buffer
+ overflow. (Closes: #299272)
+
+ * Update FAQ entry "What is the story with XFree86's license?" to clarify
+ wording and add a bit of future-proofing, per request from Daniel Stone.
+
+ * Update FAQ to remove hyperlinks to the ViewCVS installation on
+ necrotic.deadbeast.net, which no longer exists, and replace hyperlinks to
+ snapshots of the xfree86 repository with hyperlinks to the Apache-served
+ xfree86 repository itself.
+
+ * Update FAQ entry "How can I help debug the XFree86 X server?" to include
+ instructions on attaching GDB to a running X server process, and advice on
+ how to recover the console while the X server is wedged.
+
+ * Update FAQ entry "What is the story with XFree86 3.x?" to remove
+ CAN-2004-0914 as a potential security threat -- XFree86 3.x did not ship
+ the Xpm library, which that advisory covered.
+
+ * In lxterm, redirect the standard error output of "locale charmap" to
+ /dev/null so that we don't attempt to parse it. Thanks to Denis Barbier
+ for pointing this out.
+
+ * Update xfonts-100dpi-transcoded and xfonts-75dpi-transcoded's postinst and
+ postrm scripts to stop invoking update-fonts-alias; these packages ship no
+ font aliases, and can provoke warnings from that command if their
+ non-transcoded counterpart packages are not installed. Thanks to Gregory
+ Colpart for catching this. (Closes: #287221)
+
+ * Stop using non-POSIX "local" keyword in shell-lib.sh.
+
+ * Fix xdm's prerm script to remove /etc/X11/default-display-manager if the
+ corresponding debconf question has disappeared (meaning that the last
+ package which provided that shared template is being removed); this
+ enables a clean purge of the /etc/X11 directory (among the xfree86
+ packages, at least).
+
+ * Add dependencies on xfree86-common to every package that ships files in
+ /usr/X11R6 and doesn't already have such a dependency; this ensures that
+ the files located there are always locatable via the
+ /usr/{bin,include,lib}/X11 symbolic links provided by xfree86-common.
+
+ * Enhance xdm's logrotate configuration per Andrew Suffield.
+ + Rotate log file weekly.
+ + Keep logs around for one year.
+ + Compress log files.
+ + Delay compression of log files until after they've been rotated.
+ (Closes: #285871)
+
+ * Make xdm support log rotation properly.
+ + Write patch #099x.
+ - Implement new ReopenLogFile() and ReopenLogFileNotify() functions.
+ ReopenLogFileNotify() is a signal handler for SIGUSR2.
+ - Move the definition of the WRITES() macro from error.c to dm_error.h
+ so that dm.c can use it as well.
+ - Document xdm's signal handling in its manpage.
+ + Update xdm's logrotate configuration to add a postrotate script which
+ signals xdm with SIGUSR2.
+ (Closes: #303688)
+
+ * Grab patches by Thomas Winischhofer from freedesktop.org's xorg CVS which
+ fix the busted MiscPassMessage() extension implementation in xfree86 which
+ 1) copies from an invalid memory location, and 2) has memory leaks --
+ neither is a good trait in a setuid root program like the X server.
+ (Closes: #285807)
+
+ * Grab patches by Chisato Yamauchi, David Dawes, and Marc Aurele La France
+ from xfree86 CVS (none later than 2004-01-12), to perform bounds checking
+ on font metrics and avoid SEGVs in xfs and the X server. Resync patch
+ #099v. (Closes: #284448)
+
+ * Apply patches from Alex Williamson (also in freedesktop.org xorg CVS) to
+ make the XFree86 X server's ELF object loader ensure cache coherency on
+ IA64. Resync patch #071. (Closes: #303461)
+
+ * Update dexconf to use a new function, has_multiplexed_mouse(), to check
+ the Linux kernel version and only write a "Generic Mouse" InputDevice
+ stanza (and reference to it in the ServerLayout section) if the
+ "Configured Mouse" is not a device type multiplexed through
+ /dev/input/mice. In Linux 2.4, only /dev/input/mouse* devices were
+ treated this way, but now PS/2 and serial mouse devices are as well. This
+ resolves problems with button remapping and doubled mouse events which
+ caused "warp speed" mouse motion. (Closes: #269407)
+
+ * Replace mouse configuration code in xserver-xfree86.config.in with a
+ function, configure_mouse(), which performs the same functions but adds
+ knowledge of many more mouse types and attempts to guess a reasonable
+ default mouse device (dependent on architecture and even sub-architecture
+ in some cases) in the event detection does not work. Add copious amounts
+ of debugging statements (suppressed by default). Among other things, this
+ adds support for the /dev/adbmouse device node used on Apple Desktop
+ Bus-equipped Macintoshes. (Closes: #293527)
+
+ * Fix translation error in Brazilian Portguese debconf template
+ translations. (Thanks, André Luís Lopes!) (Closes: #308506)
+
+ * Add Vietnamese debconf template translations. (Thanks, Clytie Siddall!)
+ (Closes: #308422)
+
+ -- Fabio M. Di Nitto <fabbione@fabbione.net> Fri, 13 May 2005 11:12:39 +0200
+
+xfree86 (4.3.0.dfsg.1-12) unstable; urgency=medium
+
+ * Urgency set to medium due to fix for release-critical bug #295175
+ (xfree86-common fails to purge; see below).
+
+ Changes by Branden Robinson:
+
+ * Fix xfree86-common and xdm's postrm scripts to shunt update-rc.d's
+ standard output to /dev/null; the #DEBHELPER# expando causes the postrm to
+ be re-executed with the debconf frontend, which is notorious for mistaking
+ anything that appears on standard output as food, and is confused by
+ update-rc.d's unconditional chattery spew. Thanks to Adam Conrad and
+ Steve Lanagasek for walking me through the issue on IRC.
+ (Closes: #295175)
+
+ * Fix missing .HP tag in XGetWindowProperty() (et al.) manpage. Thanks to
+ Tollef Fog Heen for bringing this to my attention.
+
+ * Update FAQ entry "How can I help debug 3D acceleration problems?" per
+ feedback from Michel Dänzer.
+
+ -- Branden Robinson <branden@debian.org> Fri, 18 Feb 2005 14:45:15 -0500
+
+xfree86 (4.3.0.dfsg.1-11) unstable; urgency=medium
+
+ * Urgency set to medium due to fix for release-critical bug #286181
+ (xvfb-run broken; see below).
+
+ Changes by Denis Barbier:
+
+ * The loading of srvr_ctrl(xfree86) symbol definitions for Sun keyboards
+ was broken and made such keyboards unusable. (Closes: #286268)
+
+ * Make keypad delete key <KPDL> generate KP_Separator, not comma, in
+ Brazilian XKB layouts (thanks, Rafael Ávila de Espíndola). Update patch
+ #099z_xkb_level3_ralt_switch. (Closes: #286667)
+
+ * The basic variant of 'us' layout was not displayed by kxkb because its
+ 'hidden' attribute was wrongly set. Thanks, Frans Pop! (Closes: #290935)
+
+ Changes by Branden Robinson:
+
+ * Overhaul Xnest(1x) manpage. Among other cleanups:
+ + Fix documented syntax of the "-geometry" option's argument.
+ (Closes: #282760)
+
+ * Drop run_parts() function in Xsession scripts in favor of "run-parts
+ --list". Update Xsession(5) manual page accordingly. Increment
+ xfree86-common's versioned dependency on debianutils from (>= 1.13) to
+ (>= 1.16.9). (Closes: #283764)
+
+ * Expand and update uxterm(1x) manual page.
+
+ * Implement koi8rxterm, a uxterm-like wrapper for KOI8-R locales. Ship
+ script, manual page, and app-defaults file. Update MANIFEST files.
+ Update xterm's package description and add a Suggests: on xfonts-cyrillic.
+ Register koi8rxterm as x-terminal-emulator alternative (priority of 20,
+ same as xterm itself). Deregister koi8rxterm as alternative when the
+ package is removed. (Closes: #173565)
+
+ * Implement lxterm, a simple wrapper that chooses whether to launch xterm,
+ koi8rxterm, or uxterm based on the output of "locale charmap". Ship
+ script and manual page. Update MANIFEST files. Update xterm's package
+ description. Register lxterm as an x-terminal emulator alternative,
+ giving it a priority of 30. Lower uxterm's alternative priority to 20.
+ Deregister lxterm as an alternative when the package is removed.
+ (Closes: #279252)
+
+ * Apply patch from Bjorn Helgaas to fix SEGV in ati/radeon driver that
+ occurs when RADEONQueryConnectedDisplays() calls vbeDoEDID() but the
+ latter function returns NULL. Thanks, Bjorn! (Closes: #284025)
+
+ * Tighten the OSMesa library's dependencies on Mesa, because OSMesa depends
+ on private _glapi* functions exported only by the Mesa version of the GL
+ library. Add exact versioned dependency of xlibosmesa4 on "xlibmesa-gl (=
+ ${Source-Version)", and replace xlibosmesa-dev's dependency on "libgl-dev"
+ with "xlibmesa-gl-dev (= ${Source-Version)". Thanks to Marcelo Magallon
+ for pointing this out. (Closes: #287612)
+
+ * Apply Jan Wilhelm Stumpel's fix for the en_US.UTF-8 Compose map in
+ 4.3.0.dfsg.1-9 to pt_BR.UTF-8 as well.
+
+ * Add FAQ entry: How can I set up my X server or X terminal to use XKB
+ without any X libraries or XKB data installed?
+
+ * Remove broken hyperlink to freedesktop.org in the FAQ. Thanks to Loïc
+ Minier for pointing this out. (Closes: #285222)
+
+ * Update information in the FAQ regarding problems GNU Emacs, XEmacs,
+ Openbox, and Sawfish have with XKB's "fake keys". Retitle entry from "My
+ keyboard configuration worked with XFree86 4.2; why is it messed up now?"
+ to "My keyboard configuration worked with previous versions of XFree86;
+ why is it messed up now? For example, my Windows keys don't work right in
+ Emacs or my window manager anymore."
+
+ * Remove spurious quotes from variable expansions of $XVFBARGS and
+ $LISTENTCP in xvfb-run. Thanks to Jeff Lessem for the patch!
+ (Closes: #286181)
+
+ * Add patch by Timothy Baldwin to evdev support to fix broken backslash
+ keys. (Closes: #266274)
+
+ * Fix xserver-xfree86's configure script to not die if Discover 1.x's
+ discover command segfaults. Thanks to Kyle McMartin for reporting this
+ problem and testing the fix. (Closes: #291722)
+
+ * Add more cleanups and enchancements to xdm in patch #002:
+ + Actually attempt to write data to the auth file in
+ SaveServerAuthorizations(); that way we detect problems like ENOSPC
+ (full filesystem, user over quota) early enough to do something about
+ it. (Closes: #217505, #253480)
+ + Add comments documenting race-based symlink attack prevention
+ techniques when creating files.
+ + Add more detailed debugging and logging of errors when replacing
+ authorization files in SetUserAuthorization() and
+ RemoveUserAuthorization().
+ + Add RemovePid() function (which removes the pidFile), and register it
+ with atexit(). (Closes: #213076)
+ + Add more debugging to execute().
+ + Use the term "process ID" instead of "process-id".
+
+ * Add more clearnups and enhancements to xfs in patch #003b:
+ + Add RemovePid() function which removes the process ID file, and register
+ it with atexit() so that it is automatically invoked when xfs exits.
+ + Refer to "user ID" and "process ID" in diagnostic messages, not "userid"
+ and "process-id".
+ + Add "FUTURE DIRECTIONS" section to manpage.
+ + Perform massive cleanup and reformatting of manpage.
+
+ * Update the stillrunning() function in xdm's and xfs's init scripts to
+ remove a known-stale /var/run/{xdm,xfs}.pid file if it is present. Stop
+ removing the daemon-not-stopped file, as code removing it is run
+ elsewhere. Stop forcing the script to exit if given the "stop" argument
+ and the daemon is assumed to not be running due to no process ID file
+ being found (this would cause "restart" to fail to start the daemon in
+ this case). Remove the "daemon-not-stopped" flag file if asked to stop
+ the daemon and it either appears to not be running, or was successfully
+ stopped. Update and clarify comments in xdm's and xfs's init scripts.
+
+ * Fix typo in German debconf template translations. Thanks, Georg Neis!
+ (Closes: #289508)
+
+ * Permit lower pixel clocks (less than 17.75 MHz, but greater than 12 MHz)
+ on older (Millenium, Mystique) Matrox cards. Thanks to Jan Gorski for
+ supplying information (updates fix for #261993).
+
+ * Fix the tdfx driver to stop failing to explain why some modes are
+ rejected. If TDFXValidMode() would return MODE_BAD, use xf86DrvMsg() to
+ document the reason.
+
+ * Fix xlsfonts, xprop, xwd, and xwininfo to not bleat a usage message when
+ XOpenDisplay() fails. Being unable to connect to the X server is an
+ operational error, not a user-input error.
+
+ * Add patch from Manoj Srivastava that implements support for SELinux in
+ imake and xdm. (Note that this patch only adds source-level support, and
+ does not actually enable it.) Thanks, Manoj! (Closes: #233551)
+
+ * Update patch #007 to fix the problem with XFree86-authored manpages
+ reporting "Version Version ..." in the footer.
+
+ * Update XTerm to patch #200 (2005-02-06) from Thomas Dickey's website.
+ + Make active-icon work properly when TrueType fonts are used.
+ (Closes: #286068)
+ + Fix a regression from patch #197 which disowned the selection if it was
+ scrolled, e.g., by the user pressing return at the bottom of the screen.
+ (Closes: #277832, #291787)
+ Resync patches #059, #803, and #907.
+
+ * Move the debhelper expando to immediately precede exit in maintainer
+ scripts, except where (as in the case of xdm and xfs), we use flag files
+ that indicate when the installation is over. In those cases, place the
+ expando immediately before removal of such files.
+
+ * Add copyright notice and GPL license boilerplate to init scripts.
+
+ * Enhance xserver-common's preinst script to support upgrades from package
+ versions between 4.0.1-6 and 4.2.1-10 (thanks, Brian Sammon).
+ (Closes: #267503)
+
+ * Update patch #087 (XPM library security fixes) to revert regressions in
+ functionality caused by overly aggressive validation of filespec strings
+ in OpenReadFile() and OpenWriteFile(). (Closes: #286164)
+
+ * Add __attribute__ ((packed)) to declaration of XChar2b structure in
+ Xlib.h, for GCC/ARM systems only. This is a kludge to work around a bad
+ design decision in the GCC ARM ABI, and has been discussed with Jim
+ Gettys, Keith Packard, and Phil Blundell. Thanks to Nicolas George and
+ Gaëtan LEURENT for bringing this issue to my attention. (Closes: #285396)
+
+ * Fix the "Advanced" monitor selection method in xserver-xfree86's config
+ script to use the monitor sync ranges autodetected by read-edid, if
+ available. Thanks to Jay Birkenbilt for the patch. (Closes: #229850)
+
+ * Reimplement mkdirhier and rewrite its manual page.
+ + No longer adds extraneous leading slashes to filespecs.
+ (Closes: #141347)
+ + No longer creates spurious levels of directory hierarchy.
+ (Closes: #232357)
+
+ * Grab patch from X.Org CVS (via Jan Kreuzer) to increase DDC retry delay a
+ little bit, which permits DDC probing to succeed on some hardware
+ configurations where it would otherwise fail. Thanks, Jan!
+ (Closes: #291243)
+
+ * Update FAQ entry "What is the story with XFree86 3.x?" to mention the Xsun
+ X server packages.
+
+ * Stop using the "local" keyword in /bin/sh scripts -- it turns out "local"
+ is not POSIX, but posh -n won't tell you that as it doesn't do a syntax
+ parse on function bodies, which is the only place that keyword makes
+ sense. Comment out this usage instead of removing it as a means of
+ stoking my likely-futile hopes that the Debian Policy Manual will carve
+ out another exception to POSIX compliance for /bin/sh scripts in this
+ area. A language with only global scope for variables scarcely bears
+ thinking about. Thanks to Clint Adams for pointing out this unfiled RC
+ bug, and to Steve Langasek for hounding me incessantly about it.
+
+ * Replace post-processed comments with pre-processed ones in patch #096;
+ this quiets build-time warnings. Also document why this patch is not yet
+ suitable for submission to upstream.
+
+ * In xserver-xfree86's config script, if the discover, mdetect, or get-edid
+ (from the read-edid package) commands are not available, set the
+ corresponding autodetection question to false and mark it as seen. That
+ way, if one of the hardware detection packages is subsequently installed,
+ the user is not confusingly shown a "do autodection?" debconf question,
+ and then nothing done about the answer.
+
+ * Fix validate-posix-sh script to stop pretending that ash and dash are
+ different shells -- in Debian, they're not.
+
+ * Apply patch from Steve Langasek to drop the xprt package, and stop
+ building the Xprt X server, since this functionality did not work in any
+ version of XFree86 up to 4.3.0, and has moved to the xprt-xprintorg
+ package. (The client-side Xp library is still built.) Thanks, Steve!
+ + Update patch #003 to #define XprtServer NO and BuildXprintLib YES on
+ Debian systems. Resync patches #004, #099c, #099s, and #600.
+ + Update rules file to stop turning off the build of Xprt in the server
+ debug target, since that's now redundant.
+ + Update MANIFEST files to reflect the fact that Xprt is no longer built.
+ + Remove xprt package's stanza from debian/control.
+ + Remove debian/xprt.install debhelper file.
+
+ * Remove declaration of non-DFSG-freeness of the Compugraphic and Adobe
+ fonts (shipped as part of the Xprt server) from debian/copyright. Update
+ the prune-non-free script to stop removing these files from the upstream
+ archive. Roland Mainz has argued fairly persuasively on the debian-x
+ mailing list that the fonts were in fact freely licensed. In the event
+ Mr. Mainz is wrong, the point is pretty close to void, since 1) Xprt is
+ not being built or shipped anymore; 2) this change does not restore the
+ files in question to the .orig.tar.gz; 3) the X Strike Force doesn't
+ anticipate shipping these files in the future X.Org packages, either.
+
+ * Update patch #071 to be sure to pass the mprotect() in
+ ElfCollectSections() a page-aligned request (as the other use of
+ mprotect(), in ELFCreateGot(), already does). This reportedly prevents
+ SEGVs in the X server's ELF object loader on AMD64 systems. Thanks to
+ Piotr Kaczuba for this information. (Closes: #277699 (again))
+
+ * Add patch #099v by Guillem Jover to fix a SEGV in font server code
+ shared between xfs and the X server. Thanks, Guillem! (Closes: #294320)
+
+ * Migrate the bug scripts to the more sophisticated world order now in use.
+ + Move current scripts from /usr/share/bug/$PACKAGE to
+ /usr/share/bug/$PACKAGE/script.
+
+ * Write and ship a "presubj" bug message for the xserver-xfree86 package
+ (also used by xserver-xfree86-dbg).
+
+ * Add two entries to the FAQ:
+ + How can I help debug the XFree86 X server?
+ + How can I help debug 3D acceleration problems?
+
+ -- Branden Robinson <branden@debian.org> Fri, 11 Feb 2005 02:14:27 -0500
+
+xfree86 (4.3.0.dfsg.1-10) unstable; urgency=medium
+
+ * Upload urgency set to medium due to fix for stable-release-critical bugs
+ (#285133 and #285150; see below).
+
+ Changes by Branden Robinson:
+
+ * Make factual updates, clarifications, and wording corrections to the FAQ:
+ + Point out that the X.Org relicensing debacle was back in 1998, not
+ recent.
+ + Be more clear about why OS distributors stuck with XFree86 after it was
+ forked, but before it was relicensed.
+ + Identify which license is referred to as the "XFree86 1.1 license".
+ + Clarify the origin of the contradictory statements regarding the
+ GPL-compatibility of XFree86.
+ + Clarify the discussion of the "relicensing pilot project" that was
+ performed on the XFree86 "auto-config" code. The code was checked in,
+ *then* relicensed without other changes.
+ + Point out that most distributors have settled on X.Org's X11R6.8.1 for
+ their X Window System implementation, at least for the time being.
+ + Stress that XFree86 3.x is no longer supported.
+ + Remove language that discusses future directions of XFree86 3.x support
+ in dexconf, since it is unlikely that that work is going to happen.
+
+ * Increase the minimum pixel clock for Matrox cards based on feedback from
+ Teemu Ikonen. (Closes: #261993)
+
+ * Tidy up and improve fbdev driver messages, correcting spelling, adding
+ information, and otherwise enhancing them. (Closes: #275318)
+
+ * Make corrections to the "How does the keyboard work in the X Window
+ System?" FAQ entry based on feedback from Frank Murphy. Thanks, Frank!
+ (Closes: #279055)
+
+ * Expand discussion of configuring the mouse for left-handed use in the FAQ.
+ Thanks to Marc-Aurèle DARCHE for the information! (Closes: #281504)
+
+ * Fix syntax error in xfs package's init script. (Closes: #285133)
+
+ * Fix syntax error in xdm's Xstartup script. (Closes: #285150)
+
+ -- Fabio M. Di Nitto <fabbione@fabbione.net> Wed, 15 Dec 2004 18:40:17 +0100
+
+xfree86 (4.3.0.dfsg.1-9) unstable; urgency=high
+
+ * Security update release. Resolves CAN-2004-0914 (several Xpm library
+ vulnerabilities).
+
+ Changes by Branden Robinson:
+
+ * Update Danish debconf template translations (thanks, Claus Hindsgaul).
+ (Closes: #274101)
+
+ * Add FAQ entry: Why does composing characters work in some applications but
+ not others?
+
+ * Apply patch from Jan Wilhelm Stumpel to correct miscoded Unicode Plane 1
+ characters in en_US.UTF-8 compose map. (Closes: #267231)
+
+ * Install xlibs's bug script in the binary-indep rule, not binary-arch,
+ since xlibs is an architecture "all" package. Now the script will
+ actually be shipped.
+
+ * If the restorecon command is available, restore SELinux file security
+ contexts after setting up X server and ICE socket directories
+ (xfree86-common) and font server socket directories (xfs) after creating
+ them in the init scripts. (Closes: #268812, #271030)
+
+ * Fix missing-word typo in xnest's package description. Thanks to Roland
+ Stigge for catching this. (Closes: #268997)
+
+ * Remove build-dependency on libstdc++5-dev | libstdc++-dev. A package
+ providing libstdc++-dev is build-essential. Thanks to Steve Langasek for
+ his counsel on this issue. (Closes: #268461)
+
+ * Make xserver-common conflict with xserver-xfree86 and xserver-xfree86-dbg
+ 4.3.0.dfsg.1-[56] using exact versioned conflicts. This prevents versions
+ of the xserver-xfree86{,-dbg} packages that depend on a now-vanished
+ interface in dexconf from being installed in conjunction with it. See the
+ changelog entry for 4.3.0.dfsg.1-7 for more information. Thanks to Adrian
+ Bunk for this suggestion.
+
+ * Grab from XFree86 CVS (no later than 2003-12-31) several updates to the
+ Video4Linux (v4l) driver:
+ + [#include regionstr.h, not miscstruct.h.] (#A.1859, Alexander Stohr)
+ [David Dawes]
+ + [Update format string when reporting memPhysBase.] (David Dawes)
+ + Fix a problem with the v4l driver not seeing all the tv-norms available
+ under the bttv driver with the Linux 2.6.0-test11 kernel (#5976, Marco
+ Antonio Alvarez). [David Dawes]
+ + Change v4l "norm" names to the bttv versions (#5993, Marco Antonio
+ Alvarez). [David Dawes]
+ (Closes: #270496)
+
+ * Update "Further Information" section of FAQ.
+
+ * Add FAQ entry: What are Debian's plans with respect to X.Org and
+ XFree86?
+
+ * Add FAQ entry: Sometimes I get garbage characters like 1;2c in my xterm
+ windows; what's happening?
+
+ * Update FAQ entry: My keyboard configuration worked with XFree86 4.2; why
+ is it messed up now?
+
+ * Update FAQ entry: How does the keyboard work in the X Window System?
+
+ * Apply patch (#036) submitted to XFree86 Bugzilla on 2003-04-03 by the DRI
+ Developers to fix OpenGL application crashes with "Assertion `vb.context
+ == ctx' failed." for ATI R200 DRI users (thanks, Martin Dickopp).
+ (Closes: #241534)
+
+ * Add copyright and (MIT/X11 style) license for Bitstream Type1 fonts to
+ copyright file. Thanks to Ben Harris for pointing this out.
+ (Closes: #274018)
+
+ * Clean up and expand xclock(1x) manpage to document the effects the RENDER
+ extension has on command-line option and X resource handling. Thanks to
+ Vincent McIntyre for prompting this. (Closes: #241566)
+
+ * Use run() to unregister obsolete debconf templates in xserver-xfree86,
+ since failure to do so is not fatal, and will occur if we are upgrading
+ from a version of the package so old (pre-woody unstable) that the
+ templates don't exist to be unregistered. Thanks to Florian Laws for
+ diagnosing this bug. (Closes: #275710)
+
+ * Stop manually compressing the plain text version of the Debian X FAQ;
+ dh_compress will take care of that. Tell dh_compress to avoid compressing
+ the XHTML version of the FAQ.
+
+ * Restore xdm's Xreset script, which was removed in 4.2.1-12. If the
+ use-sessreg option is set and the sessreg command is available, user X
+ sessions will be de-registered from the utmp and wtmp files again.
+ Update maintainer scripts to stop removing the conffile. Update
+ xdm.install and MANIFEST.* files to ship Xreset script. The old
+ xconsole-related logic is not restored. (Closes: #275329)
+
+ * Replace XFree86 X server pre-release banner message with long boilerplate
+ that declaims the release as heavily modified by Debian, and directs users
+ to the Debian Bug Tracking System and the Debian X FAQ instead of XFree86
+ mailing lists; also replace an occurence of an XFree86 mailing list
+ address in a tseng driver message (patch #914).
+
+ * Update xlibs's package description to stop implying that *no* packages
+ have up-to-date dependencies on it (xbase-clients does, due to xkbcomp,
+ which is statically linked against libxkbfile).
+
+ * Apply patch from Jeff Licquia to port the X configlet to python-configlet
+ version 2 so that xserver-xfree86 package configuration
+ doesn't crash for people who have the configlet infrastructure installed.
+ Make xserver-common conflict with python-configlet (<< 2.0) accordingly.
+ (Closes: #263076)
+
+ * Apply patch from Jason Lunz for silly thinko that caused arguments to
+ xvfb-run to be misparsed if they contained whitespace. Thanks, Jason!
+ (Closes: #276415)
+
+ * Tidy up and make consistent all Debian-specific manpages.
+
+ * Update package descriptions to replace the term "pseudopackage" (which
+ apparently only the xfree86 source package uses) with "dummy package" in
+ the short description, and "transitional package" in the extended
+ description, per section 6.7.7 of the Debian Developers' Reference.
+
+ * Update patch #009 to improve the fix for the Rage128-specific problem with
+ engine lockups; because CCE idle commands may never finish, eventually
+ stop the engine altogether after a long enough period of inactivity.
+ Thanks to Daniel Jacobowitz for the original patch, this improved version,
+ and the real-world testing. (Closes: #278654)
+
+ * Add FAQ entry: What is the story with XFree86 3.x?
+
+ * Apply some fixes to the XFree86 X server's ELF object loader:
+ + Stop conditionally #including <sys/mman.h> since it is unconditionally
+ imported earlier in the file.
+ + Perform an mprotect() for AMD64 Linux systems as well as OpenBSD to
+ prevent SEGVs on NX-enabled kernels. (Closes: #277699)
+ [2005-02-09: Didn't really fix this bug. BR]
+ + Fix a problem with parentheses when using the preprocessor (doesn't
+ affect Debian, but would cause unexpected behavior on OpenBSD).
+
+ * Perform some cleanup work on the shared library maintainer scripts of
+ libdps1, libxaw6, libxaw7, xlibmesa-gl, xlibmesa-glu, and xlibosmesa4;
+ thanks to Fabio for noticing the inconsistencies.
+ + Call register_x_lib_dir_with_ld_so() before invoking ldconfig in the
+ post-installation scripts.
+ + Perform configuration actions regardless of the value of "$1" in the
+ post-installation scripts.
+ + Call deregister_x_lib_dir_with_ld_so() before invoking ldconfig in the
+ post-removal scripts.
+ + Move the #DEBHELPER# expando later in the scripts where applicable.
+
+ * Perform some cleanup work on the update-fonts-{alias,dir,scale} utilities
+ and their manual pages.
+ + Eliminate XSIisms (-a, -o) in uses of test(1)/[(1).
+ + Re-indent at four spaces.
+ + Query terminal for its width if $COLUMNS is not set, as shell-lib.sh
+ does. Fall back to a width of 80 if stty returns garbage or nothing.
+ + Rename error() to die().
+ + Add usage() function and -h, --help option.
+ + Use exit status 2 to indicate usage errors as opposed to operational
+ failures.
+ + Be more careful with quoting shell variables.
+ + Add and expand comments.
+ + Add SVN Id keyword.
+ + Update Vim modeline.
+ + Update manual pages to document changes in interface and behavior (there
+ are no backwards-incomaptible changes).
+
+ * Perform some cleanup work on the xvfb-run utility and its manual page.
+ + Query terminal for its width if $COLUMNS is not set, as shell-lib.sh
+ does. Fall back to a width of 80 if stty returns garbage or nothing.
+ + Add message() and error() functions.
+ + Enhance usage() function to (additionally) display an error message
+ consisting of the function arguments, if any.
+ + Use error() and usage() instead of echo for error messages.
+ + Cosmetically tidy up usage message, and correct a typo.
+ + Declare locally-scoped variable in find_free_servernum().
+ + Store getopt's exit status and report it in an error message if it is
+ nonzero.
+ + Clarify the error message if getopt rearranges the command line in a way
+ we don't expect.
+ + Tidy up comment, shell function, and redirection style. Wrap long
+ lines. Quote variables that may have whitespace in them. Update Vim
+ modeline.
+ + Describe effect of $COLUMNS environment variable in manpage.
+ + Remove erroneous description of exit status 1, and explain why it is not
+ used.
+ + Add description of exit statuses 0, 4, 5, and 6.
+ + Fix markup error.
+
+ * Replace many occurences of XSI connectives in test(1) with POSIX shell
+ connectives in scripts.
+
+ * Apply patch from David Mosberger that replaces the fix for #225526 with
+ one that works on systems that do not have a PCI bus numbered 0. Thanks,
+ David! (Closes: #279436)
+
+ * Apply patch from Richard Mortimer to fix the XFree86 X server's ELF object
+ loader to set the PROT_EXEC flag on mmap()ed modules regardless of machine
+ architecture. (It was already trying to do this, but there are three
+ preprocessor statements involved, and we were only patching one.)
+ (Closes: #280384)
+
+ * Update XTerm to patch #197 (2004-09-30) from Thomas Dickey's website.
+ + Change default translations so a BtnDown which is not recognized is
+ simply ignored rather than emitting a bell. That makes it less
+ obtrusive when the user tries to use a mouse which provides more
+ capabilities than the X mouse driver supports, e.g., one with a
+ horizontal scroll wheel. (Closes: #265133)
+ + Note in xterm's manpage that "translations" is not specific to xterm.
+ (Closes: #278897)
+ + Modify uxterm script to use locale program to verify if the derived
+ locale is installed. (Closes: #246398)
+ + Modify the criteria for disowning primary selection. Previously, this
+ happened anytime the cursor was moved before the end of the selection.
+ That would ensure that any insert/delete of char or line, as well as
+ scrolling, would disown the selection. The new criteria change this to
+ checking if the operations would modify the data which is highlighted.
+ (Closes: #277832)
+ + Correct handling selection of tabs over cleared space in UTF-8 mode. In
+ this special case, xterm represents the whitespace with a null rather
+ than a space character. (Closes: #276447)
+ + Amend fix for infinite loop from patch #192 to check if there is wrapped
+ text to output in a following iteration. (Closes: #273202)
+ + Amend solution for Debian #252873, #260471 from patch #194 by making the
+ cursor not explicitly colored if only the foreground color is set, and
+ the cursor is on a blank space. (Closes: #275473)
+
+ * Add magic register poke to neomagic driver which apparently speeds up DVD
+ playback operations. Thanks to Tony Hill for finding this patch.
+ Unfortunately, despite efforts to locate some, there appears to be no
+ documentation of why this register poke works. (Closes: #277038)
+
+ Changes by Denis Barbier and Fabio M. Di Nitto:
+
+ * Edit xc/programs/xkbcomp/symbols/pc/Imakefile so that the new pc/us_intl
+ layout file gets installed. Update MANIFEST files and xlibs.install to
+ ensure the file get shipped.
+ + Ship a multi-layout aware us_intl layout. (Closes: #271326)
+ + Make us_intl work again. (Closes: #274513)
+
+ * Update Dutch debconf template translations (thanks, Bart Cornelis).
+ (Closes: #278714)
+
+ Changes by Denis Barbier:
+
+ * Make Sun keyboards load srvr_ctrl(xfree86) symbol definitions to have
+ access to standard Ctrl+Alt key sequences. (Closes: #236086)
+
+ * Add several multi-layout aware layouts:
+ + ca (Canada) contains several variants: "fr" is meant as a replacement
+ for ca_enhanced, "fr-legacy" is another variant, and "multi" is an
+ implementation of CAN/CSA-Z243.200-92. This standard requires two
+ groups, so it is in fact divided into "multi" and "multi-2gr" variants.
+ By default, "ca" loads both "multi" and "multi-2gr" to provide a
+ full-featured keyboard. When it is combined with other layouts, the
+ simpler variant "fr" is used, but others can be explicitly specified.
+ (Closes: #245371, #252068, #258575)
+ + The new ca layout also fixes ISO_Level3_Shift, which was mapped to
+ <RCTL> instead of <RALT>. (Closes: #274457)
+ + jp (Japan) also requires a combination of two variants. When this is
+ the only loaded layout, the full combination is used. When this layout
+ is combined with others, only the "jp106" variant is loaded.
+ + hu (Hungary) layout has been simplified; some variants have been removed.
+ (Closes: #263561)
+ + ch (Switzerland) replaces de_CH and fr_CH.
+ + mn (Mongolia) and vn (Vietnam) are simple layouts.
+
+ * In UTF-8 compose files, replace U00xx Unicode codepoints by the
+ corresponding keysyms. Otherwise the compose character is in
+ the range 0x01000000-0x010000FF which is not valid according to
+ appendix A of the X11 protocol. Even if most applications are able
+ to recognize these broken characters, few of them fail, most notably
+ xemacs and users are then unable to use composition when in UTF-8
+ locales. (Closes: #235574)
+
+ * Replace level3(ralt_switch_multikey) by level3(ralt_switch) in all
+ layouts so that symbols on the 4th column may be accessed with
+ Shift+Alt_R or Alt_R+Shift. (Closes: #270235)
+
+ + In /etc/X11/xkb/rules/xfree86, move lv3 definitions after altwin so
+ that the former can override the latter. See Bug#276143.
+
+ Changes by Fabio M. Di Nitto:
+
+ * Update wacom input driver from http://linuxwacom.sourceforge.net/ up to
+ version 0.6.4:
+ + Remove now obsolete patch 036_wacom_usb_tablet_update.diff
+ + Add patch 000_stolen_from_sourceforge_wacom_driver.diff
+ + Fix input timeout problems if the device is not connected.
+ No references are available in upstream changelog, but tests do not
+ show this behavior anymore. (Closes: #231837)
+ + Fix compatibility with kernel 2.6 wacom driver. (Closes: #250331)
+ + Add support for Cintiq boards. According to #172526 logs the patch
+ has been included, reworked and cleaned by upstream a long time ago.
+ No references are available in upstream changelog. (Closes: #172526)
+ + A combination of new kernel, xfree86 and wacom driver appears to have
+ fixed the APM resume issue as described in #80140. Tests done
+ on a powerpc laptop do not show this problem anymore.
+ No references are available in upstream changelog. (Closes: #80140)
+
+ * Add workaround for the UseBIOS default setting in the savage driver:
+ + Add patch 099i_pro_savage_ddr_set_use_bios_to_false.diff
+ + Set UseBios default to "no" for PROSAVAGE_DDR and PROSAVAGE_DDRK, as
+ described at <URL: http://www.probo.com/timr/savage40.html >.
+
+ * Fix nl_BE locale alias information.
+
+ * Use validate-posix-sh at build time to perform syntax check on generated
+ maintainer scripts.
+
+ * Increase verbosity of validate-posix-sh warnings. Stop immediately
+ exiting upon encountering the first shell interpeter that fails on the
+ script; instead, attempt all the interpreters and report all that fail.
+
+ * Sync debian/rules install-server target with install and make
+ binary-server work again.
+
+ Changes by Fabio M. Di Nitto and Branden Robinson:
+
+ * Update patch #087 to include fruits of Xpm source code security audit by
+ Thomas Beige. Resolves CAN-2004-0914: memory leak, improper use of signed
+ integers, and overflows in the Xpm library. Resync offset in patch #200.
+
+ -- Fabio M. Di Nitto <fabbione@fabbione.net> Thu, 9 Dec 2004 17:14:45 +0100
+
+xfree86 (4.3.0.dfsg.1-8) unstable; urgency=high
+
+ Changes by Denis Barbier:
+
+ * Update Danish debconf template translations (thanks, Claus Hindsgaul).
+ (Closes: #270503)
+
+ * Grab /etc/X11/xkb/symbols/pc/us_intl from XFree86 CVS (2003-10-04). This
+ makes us_intl multi-layout aware, and binds the right Meta key to
+ Mode_Switch. (Closes: #271326, #234085)
+
+ * Revert upstream change to altwin behavior applied in 4.3.0.dfsg.1-1.
+ The Meta_* keysyms must be removed from the shift level of Alt keys when
+ altwin:meta_win or altwin:left_meta_win are used; otherwise, XKB-unaware
+ applications (like GNU Emacs) may believe that mod1 and mod4 are bound to
+ Meta keys.
+
+ * Fix inconsistencies in rules/xfree86* files:
+ + rules/xfree86: List in $oldlayouts all layouts which are currently not
+ multi-layout aware in the Debian xlibs package, including hu_US (thanks,
+ Veres Lajos). (Closes: #247463, #270810)
+ + rules/xfree86: Support the lv3:ralt_switch option, which exists in
+ xfree86.lst but was forgotten in this file.
+ + rules/xfree86: Remove special case that causes PC keyboard models with
+ the us_intl layout to load the us layout and add the us_intl layout to
+ it. (Closes: #234085)
+ + rules/xfree86.{lst,xml}: Replace the options altwin:meta_super and
+ altwin:meta_hyper with the correct names altwin:super_win and
+ altwin:hyper_win. (Closes: #271259)
+ + rules/xfree86.{lst,xml}: Synchronize these files.
+
+ Changes by Denis Barbier and Branden Robinson:
+
+ * Add FAQ entry: My keyboard configuration worked with XFree86 4.2; why is
+ it messed up now? (Closes: #259740)
+
+ Changes by Branden Robinson:
+
+ * In the Xsession script, attempt to create a file of non-zero length in
+ /tmp; a full filesystem can cause mysterious X session failures.
+
+ * Add FAQ entry: How does the keyboard work in the X Window System?
+
+ * Update XTerm to patch #196 (2004-08-15) from Thomas Dickey's website.
+ + Change clearing operations so foreground color attribute is not set.
+ Usually this is benign, but in some cases when the cursor color is not
+ set explicitly, the cursor would show this color. (Closes: #252873)
+ + Modify initialization and cleanup of utmp data to also compare the
+ ut_line member. (Closes: #256468)
+
+ * Update XTerm FAQ to revision 1.86 (2004-07-28) from Thomas Dickey's
+ website.
+
+ Changes by Fabio M. Di Nitto:
+
+ * Create debian/tmp/usr/X11R6/lib/X11/doc when NOT_BUILDING_XFREE86_X_SERVER
+ is defined. Fixes FTBFS on s390.
+
+ * Security update (patch #087). Resolves the following issues:
+ + CAN-2004-0687: stack overflows in libXpm
+ + CAN-2004-0688: integer overflows in libXpm
+ (Closes: #272493)
+
+ * Update nv driver and some support code, and grab an XVideo bugfix from
+ X.Org CVS as of 2004-09-15:
+ + Update 000_post430.diff:
+ - Drop nv_setup.c bits.
+ + Update 000_stolen_from_HEAD.diff:
+ - Include xf86xv.c Xv fix for big-endian machines.
+ + Rename 000_stolen_from_HEAD_nv_driver.diff to
+ 000_stolen_from_xorg_nv_driver and update:
+ - Full nv driver from X.org.
+ - Removed 2 references to XORG_VERSION_CURRENT since it is not defined.
+ + Rename 000_stolen_from_freedesktop.org.diff to 000_stolen_from_xorg.diff
+ and update:
+ - Add xaa bits required to compile the new nv driver.
+ - Add fbdevhw bits required to compile the new nv driver.
+ + Update 030_Xserver_and_driver_region_primitive_fixups.diff:
+ - REGION_EQUAL backport from X.org.
+ - Fix call to REGION_EQUAL in Xv autopaint initialization error
+ introduced in 4.3.0.dfsg.1-7.
+ + Drop 098_nv_xvideo_fullscreen_fix.diff:
+ - Now included upstream.
+ + Rediff 911_debian_XF86Config_to_XF86Config-4.diff.
+ + Update MANIFEST and xserver-xfree86.install files to reflect the new
+ riva128.o submodule.
+ (Closes: #269025, #268759, #271235, #270228, #271071, #270714)
+
+ * Update Czech debconf template translations (thanks, Miroslav Kure).
+ (Closes: #273516)
+
+ Changes by ISHIKAWA Mutsumi:
+
+ * Update Japanese debconf template translations (thanks, Kenshi Muto).
+
+ -- Fabio M. Di Nitto <fabbione@fabbione.net> Tue, 28 Sep 2004 13:14:12 +0200
+
+xfree86 (4.3.0.dfsg.1-7) unstable; urgency=high
+
+ Changes by Branden Robinson:
+
+ * Update Turkish debconf template translations (thanks, Recai Oktas).
+ (Closes: #256360)
+
+ * Grab latest version of XTerm (#191) from Thomas Dickey's website.
+
+ * Add FAQ entry: Why is xterm using the "fixed" font for everything,
+ disregarding the font configuration from the X resource database?
+ (Closes: #254650)
+
+ * Update hyperlink to Debian X FAQ in NEWS file. Thanks to Calum Mackay for
+ pointing this out.
+
+ * Make libxpm-dev rather than xlibs-static-dev provide libxpm4-dev.
+ Thanks to Daniel Schelper for pointing this out. (Closes: #256420)
+
+ * Move atobm and its manpage from xutils to xbase-clients; this shuts up
+ dangling symlink warning mails from cron on systems with xutils installed
+ but not xbase-clients. Make xbase-clients declare that it replaces xutils
+ (<< 4.3.0.dfsg.1-7). (Closes: #200061)
+
+ * Invert the sense of a test in debian/rules; this results in xlibmesa-dri's
+ bug script actually being shipped. D'oh!
+
+ * Make xbase-clients depend on xlibs and explain why in the package's
+ extended description. Thanks to Andreas Metzler for catching this.
+ (Closes: #257023)
+
+ * Update FAQ entry "How do I run an X client as root when the X session
+ is run by a user?" to document existence of sux.
+
+ * Update Glide3 library support for amd64 and ia64 thanks to Guillem Jover.
+ Increment versioned build-dependency on libglide3 to >= 2002.04.10-7, and
+ require this library for building on amd64 and ia64. Edit patch #003 to
+ enable compilation of tdfx DRI Mesa module on amd64 and ia64. Shut up cpp
+ warnings when generating Makefiles by defining a default value for
+ HasGlide3 in xfree86.cf ("NO"). Resync patches #004, #048, #103, #150,
+ #250, #500, and #600. Ship tdfx DRI Mesa module in xlibmesa-dri for amd64
+ and ia64, and unstripped version of same in xlibmesa-dri-dbg.
+ (Closes: #255270)
+
+ * Fix up Glide3 library support in various respects. Stop shipping tdfx DRI
+ Mesa module on powerpc until it stops being useless (i.e., until
+ libglide3-dev is available on that architecture). Move xlibmesa-gl's
+ suggestion of the libglide3 package (and explanation thereof) to the
+ xlibmesa-dri package.
+
+ * Add FAQ entry: I'm having trouble getting dual-head support to work on my
+ ATI Radeon card. Can you help? (Closes: #236197)
+
+ * Update FAQ to include hyperlink to the upstream source for the XTerm FAQ.
+
+ * Grab latest version of XTerm (#192) from Thomas Dickey's website.
+ According to the upstream changelog, this release addresses several Debian
+ bugs.
+ + Change resource settings for color4 and color12; add some discussion in
+ XTerm-col.ad. (Closes: #255070)
+ + Add a note in xterm manpage regarding possible conflict between resource
+ settings for xterm.vt100.font and xterm.vt100.utf8Fonts.font.
+ (See #254650.)
+ + Add case to uxterm to accommodate locales ending with "@euro", e.g.,
+ fr_FR.UTF-8@euro (report/analysis by Matthieu Lagouge).
+ (Closes: #255197)
+ + Save the fontnames for bold fonts that are derived from normal fonts, or
+ from the boldFont resource, so the same value is restored when switching
+ with the VT Fonts menu. (Closes: #256086)
+ + Fix typo in manpage's description of character classes.
+ (Closes: #257073)
+ + Modify terminfo to accommodate luit, which relies on G1 being used via
+ an ISO-2022 escape sequence (analysis by Juliusz Chrobocek).
+ (See #254316; in Debian, this is actually an ncurses-base bug.)
+
+ * Update XTerm FAQ to latest version (1.84) from Thomas Dickey's website.
+
+ * Grab from XFree86 CVS (no later than 2003-11-10) several fixes to the
+ RandR extension manual page and server-side implementation.
+ + Remove reference to non-existent Xrender(3) man page (reported to
+ webmaster). [David Dawes]
+ + Fix manual pages typos (Bugzilla #316, Jens Schweikhardt). [Matthieu
+ Herrb]
+ + Removed Copyright sign from man pages as this causes some formatters to
+ choke (Egbert Eich).
+ + Fixed rounding of refresh rate in RandR, updated Xrandr man page
+ (BugzillaR #527, Egbert Eich).
+ + Fix RandR initialization problem when default screen resolution is
+ different from the virtual resolution. (Bugzilla #853, Bob Paauwe).
+ [Alan Hourihane]
+ + The call to xf86RandRSetMode() from xf86RandRCreateScreenResources()
+ doesn't appear to be necessary. Removing it restores the documented
+ ViewPort behaviour when RandR is enabled (David Dawes).
+ + Warning fix [Marc Aurele La France]
+ + C-style comments were being stripped by preprocessing. [David Dawes]
+ (Closes: #256297)
+
+ * Fix bug in COMPOUND_TEXT translation of the GBK character encoding
+ (thanks, Su Yong). (Closes: #255224)
+
+ * Fix missing element of Debian X FAQ conversion to XHTML; correct
+ s390-specific version of xfree86-common dh_installdocs file to recognize
+ that the FAQ is already gzipped.
+
+ * Move BDF font character set conversion files from xfonts-base package to
+ xutils (these files are used and needed by the ucs2any tool in the latter
+ package, and not by anything else). Update MANIFEST and .install files
+ accordingly. Make xutils declare that it replaces xfonts-base (<<
+ 4.3.0.dfsg.1-7).
+
+ * Rewrite and greatly enhance manifest-install-reconcile tool to be able to
+ check against lists of known false positives and false negatives on a
+ per-architecture basis. Ship these lists, add a usage message to
+ manifest-install-reconcile, and document use of the script in the README
+ file.
+
+ * Grab from XFree86 CVS (no later than 2003-12-18) fixes to the ja_JP.UTF-8,
+ ko_KR.UTF-8, and zh_TW.UTF-8 XLC_LOCALE files to prefer ISO
+ 10646-1-encoded fonts less strongly. Add zh_CN.UTF-8 XLC_LOCALE file.
+ Update MANIFEST and .install files accordingly. (Closes: #255701)
+
+ * Provide local, hacked version of debconf-updatepo (in debian/scripts) that
+ doesn't destroy the comment and header block of debian/po/templates.pot.
+ Add "pristine" comment and header file in debian/po, add debian/rules
+ targets to handle PO and POT file updates using our local script, and
+ document the recommended procedure in debian/README. Hopefully the real
+ debconf-updatepo will be fixed soon.
+
+ * Update debian/po/*.{po,pot} files to update comment header and get effects
+ of a debconf-updatepo bugfix which misidentified template types in some
+ situations.
+
+ * Update Czech debconf template translations (thanks, Miroslav Kure).
+ (Closes: #260868)
+
+ * Update German debconf template translations (thanks, Florian Ernst and
+ Alwin Meschede). (Closes: #261239)
+
+ * Update Dutch debconf template translations (thanks, Bart Cornelis).
+ (Closes: #261337)
+
+ * Update Brazilian Portuguese debconf template translations (thanks, André
+ Luís Lopes). (Closes: #261389, #269507)
+
+ * Apply patch from Thomas Dickey to fix Imake's CppRawFileTarget() so that
+ it no longer tempts cpp into corrupting manpages. Thanks, Thomas!
+ (Closes: #259996)
+
+ * Update hurd-i386 support.
+ + Update gnu.cf to declare platform support for groff, libpng, and PAM
+ libraries in parity with Linux; also specify correct path to cpp
+ executable.
+ + Tell XTerm's Imakefile that openpty() needs -lutil on GNUArchitecture
+ (patch #803).
+ + Disable special hostname registration code for __GNU__ in xdm (patch
+ #804).
+ + Update MANIFEST.hurd-i386.in in light of updates to build configuration.
+ + Remove MANFIEST.hurd-i386.all; it is no longer needed.
+ + Update mir-false-negatives and mir-false-positives to reflect changes in
+ MANIFEST files.
+ (Thanks, Michael Banck!)
+
+ * Grab from XFree86 CVS (2003-05-29) a fix for a SEGV in libX11's lcPrTxt()
+ function, which caused application crashes when, e.g., pasting text into
+ XTerm. Also remove patch #051 in favor of identical fix committed to
+ XFree86 CVS on 2003-04-03. Thanks to Thomas Dickey for tracking this
+ bug, and fix, down. (Closes: #250655)
+
+ * Expand, correct, and tidy up the setxkbmap(1x), xkbcomp(1x), xkbevd(1x),
+ and xkbprint(1x) manpages; also correct typos in xkbcomp's README.config
+ file (patch #051). (Closes: #255371)
+
+ * Perform major stylistic fixes to the XF86Config-4(5x) manpage, greatly
+ expand its cross-reference section to mention all shipped section 4x
+ manpages (e.g. radeon(4x)), and fix some typos (patch #099e).
+ (Closes: #257440)
+
+ * Update manpage cross-references to refer to XF86Config-4(5x) instead of
+ XF86Config(5x), and add brief explanatory note to top of XF86Config-4(5x)
+ (patch #911).
+
+ * Update FAQ to split off some of existing entry "Why are some fonts (or
+ graphical rulers) sized incorrectly?" into new question "How do I set the
+ DPI (dots-per-inch) value used by the X server?".
+
+ * Grab fix from XFree86 CVS (2003-05-07) to bounds problems in *scanf(),
+ snprintf() and strncpy() calls in lbxproxy, makepsres, and xedit.
+
+ * Tighten regex in patch-audit rule so messages like "patching file
+ xc/doc/man/Xt/XtOffset.man" are not false positives.
+
+ * Grab from XFree86 CVS (2003-05-29) a boatload of typographical fixes to
+ manual pages. Resync and make further corrections to manpages in patches
+ #026 (appres, ico) and #051 (setxkbmap, xkbcomp, xkbevd, xkbprint).
+ Resync patches #032, #099e, #911.
+
+ * Grab from XFree86 CVS (2003-07-16) more updates to the nv driver. Fixes
+ several bugs as well as adding support for several chipsets, including
+ GeForce FX 5600SE, GeForce FX Go5600, GeForce FX Go5650, Quadro FX Go700,
+ GeForce FX Go5100, GeForce FX 5900 Ultra, GeForce FX 5900, Quadro FX 3000,
+ and the "mystery" TNT2 card with a PCI model ID of 0x002a. Extend
+ reversion of REGION_EQUAL macro implementation to nv driver in patch #030.
+ Resync patch #911.
+
+ * Grab from XFree86 CVS (2003-07-07) more updates to the trident driver,
+ placing them in their own patch file. Fixes several bugs and adds support
+ for TV chipsets VT1621 and CH7005. Drop trident driver patches in
+ #000_stolen_from_HEAD in favor of new patch file. Extend reversion of
+ REGION_EQUAL macro implementation to trident driver in patch #030. Resync
+ patch #911.
+
+ * Change Catalan debconf template translations to use the characters "LATIN
+ SMALL LETTER L" and "MIDDLE DOT" instead of "LATIN SMALL LETTER L WITH
+ MIDDLE DOT" so that the file can be successfully transcoded to ISO
+ 8859-15, shutting up hateful noise from one of the programs called by
+ dh_installdebconf (thanks, Guillem Jover).
+
+ * Grab from XFree86 CVS (2003-10-02) several updates to the mouse input
+ driver, fixing bugs and adding support for the Kensington
+ "ValuMouseScroll" protocol. Grab an additional fix from XFree86 CVS
+ (2003-12-08) for mouse wheel misbehavior on PowerPC. Resync patches #040
+ and #097. Add "ValuMouseScroll" to the list of supported serial mouse
+ protocols in xserver-xfree86's configuration script.
+
+ * Grab from XFree86 CVS (2003-10-02) several bugfixes to xedit, including a
+ fix for a SEGV in Lisp_New(). Move existing xedit patches from XFree86
+ CVS to their own patch file. (Closes: #257835)
+
+ * Major housecleaning of X11's locale support (patch #011a).
+ + Add all locales supported by version 2.3.2 of the GNU C Library except
+ zh_CN.GB18030, sq_AL.ISO-8859-1, and tg_TJ.KOI8-T. X has no support for
+ the GB18030 and KOI8-T character sets, and glibc appears to disagree
+ with most of the world about which character set Albanians use (see, for
+ example, iso_8859-2(7)). This baka gaijin has learned his lesson about
+ trying to report l10n bugs in glibc, so I'll let someone else do the
+ jousting. :-P
+ + Synchronize compose.dir with locale.dir wherever possible (i.e.,
+ wherever compose tables exist) so that as many supported locales as
+ possible get a Compose map. Comment out references to compose tables
+ that don't actually exist.
+ + Disable support for iu_CA.NUNACOM-8. X11 doesn't have any support for
+ the NUNACOM-8 character set. (iu_CA.UTF-8 is still supported.)
+ + Disable support for locales that are neither supported by the GNU C
+ Library, nor use standard ISO 639 language or 3166 territory codes.
+ This includes Esperanto (eo_EO, eo_XX), and what I presume is
+ Pennsylvania "Dutch" (pd_DE, pd_US).
+ + Turn off intrusion of Nynorsk into Nyanja's ISO 639 space. (Nynorsk is
+ nn, not ny.)
+ + Retain support for a3 as an alias for az (Azerbaijani), though I'm
+ dubious about it.
+ + Turn off support for the bogus (and ancient) FreeBSD-ism of using
+ "lt_LN.ISO8859-1" as a "generic" Latin-1 locale.
+ + Remove bogus "ee" locales for Estonian (et).
+ + Remove ISO 8859-15 support for languages that don't use that character
+ set according to glibc.
+ + Alias "fi" and "fi_FI" to use ISO 8859-1, not ISO 8859-15, for
+ compatiblity with glibc (thanks, Jukka Suomela). (Closes: #266834)
+ + Remove @euro modifier support from locales that don't use that currency
+ (i.e., countries not in the Euro zone).
+ + Ensure we support the @euro modifier for all locales that use it
+ (according to glibc), even if the codeset is explicitly specified. See
+ Debian #208137.
+ + Rewrite some comments to be in grammatical English.
+ + Add locale-munger script, which I used to gather the information that
+ drove these changes.
+ + Resync patch #096.
+ + Fix a few Debian bugs in the process:
+ - en_GB.ISO-8859-15 is now supported. (Closes: #122039)
+ - et_EE.ISO-8859-15 is now supported. (Closes: #258223)
+ - sv_SE.ISO-8859-15 is now supported. (Closes: #243237)
+
+ * Apply patch to savage driver to turn off RandR extension when driver's
+ "Rotate" option is used, preventing screen garbling (patch #099f; thanks,
+ Jorg Schuler). Update manpage to document this fact. Resync patch #911.
+ (Closes: #258986)
+
+ * Fix wrong comment character in Debian customizations to XTerm application
+ defaults file. (Closes: #252274)
+
+ * Grab from X.Org CVS a fix for SEGVs in the chips driver when the current
+ mode wasn't valid. Thanks to Ryan Underwood for identifying and testing
+ this fix. Resync patches #044 and #045. (Closes: #260753)
+
+ * Make keypad delete key <KPDL> generate KP_Separator, not KP_Decimal, in
+ Polish XKB layouts (thanks, Emil Novak). (Closes: #261163)
+
+ * Grab latest version of Thomas Winischhofer's SiS driver for the XFree86 X
+ server from his website as of 2004-08-20. Resync patch #911.
+ + Circumvent unresolved symbol complaints by explicitly loading the "glx"
+ module when DRI is used. (Closes: #254601)
+
+ * Fix up the Mesa DRI tdfx module so that it really does use only libglide3,
+ not libglide3-dev (#099h). Open the actual libglide shared library we
+ need, not the symbolic link used at compile time, which may not be
+ installed on non-development systems. When opening chipset-specific glide
+ libraries, use the filenames actually in use by the current version of the
+ libglide3 package. Thanks to Guillem Jover for this patch.
+ (Closes: #260099)
+
+ * Update "Why does the X server take up so much memory?" FAQ entry with
+ information from Albert Cahalan about how I/O is charged to processes in
+ the procps utilities (he says it actually isn't). Also add description of
+ xrestop package and utility.
+
+ * Overhaul dexconf.
+ + Drop support for XFree86 3.x X server format.
+ + Drop support for -f, --format and -s, --stdout options.
+ + Use exec and open file descriptor 4 for output, instead of repetitiously
+ redirecting standard output to a file explicitly all the time. We can't
+ use fd 3 since debconf uses it.
+ + Use here documents a little more, and printf a little less.
+ + Put blank lines between XF86Config-4 sections so that the file is easier
+ to read.
+ + Overwrite the destination file instead of "mv"ing or "cp"ing it; this
+ way it is never unlinked. (The maintainer scripts implement enough
+ paranoia already.)
+ + Tidy up shell style.
+ + Modernize dexconf(1) manual page to undocument obsolete options, replace
+ the "FUTURE DIRECTIONS" section with "HISTORY", correct and update
+ "DIAGNOSTICS" section, provide more cross-references, and put a newline
+ at the end of every sentence in the document source (per the GNU Roff
+ texinfo documentation).
+ + Use -o instead of -s and redirection when invoking dexconf from
+ xserver-xfree86's postinst script, since the -s option is no longer
+ supported.
+ + WARNING: xserver-xfree86{,-dbg} 4.3.0.dfsg.1-[56] are not compatible with
+ this version of xserver-common, or planned future versions, due to the
+ above-mentioned removal of dexconf's -s option. At the same time,
+ saying xserver-common 4.3.0.dfsg.1-7 conflicts: xserver-xfree86
+ (<< 4.3.0.dfsg.1-7) and xserver-xfree86-dbg (<< 4.3.0.dfsg.1-7) is
+ excessive because 4.3.0.dfsg.1-4 and earlier will work. I have heard
+ that stating such a conflict can make things difficult for apt's
+ problem resolver (or even impossible, if in conjunction with enough
+ other versioned dependency or conflicts), so the risk seems greater than
+ the reward. It's too bad we don't have a way to conflict on ranges that
+ aren't bounded on either zero or infinity. Given that 4.3.0.dfsg.1-[56]
+ have never appeared even in sarge, I'm making this judgement call. Any
+ bugs about this will be closed with this changelog entry included,
+ unless you can rebut my premises; i.e., "I am Jason Gunthorpe and I
+ declare that apt can eat a stack of versioned conflicts like this for
+ breakfast and still see the One True Upgrade Path!", or "You silly
+ person, of *course* dpkg and apt accept 'Conflicts: (4.3.0.dfsg.1-4 <<)
+ xserver-xfree86 (<< 4.3.0.dfsg.1-7)'!"
+
+ Changes by Fabio M. Di Nitto and Branden Robinson:
+
+ * Support building only the parts of the source tree needed by
+ architecture-specific packages.
+ + Modify Debian Imake configuration to support building with fonts and
+ specs docs turned off (patch #003). Resynchronize GNU/Hurd,
+ GNU/FreeBSD, and GNU/NetBSD Imake configurations with Debian
+ configuration (patches #800, #820, #842, #900). Resync diffs in patches
+ #004, #099c, #600.
+ + Modify Imakefiles to build and install the font utilities even if the
+ fonts themselves are not being shipped (patch #095).
+ + Make the "build" target a dummy target, replacing it with
+ "build-arch-only" and "build-all" targets. Target-specific variable
+ values are used to invoke a "real-build" target, turning off the
+ building of fonts and "specs docs" as needed.
+ + Add comments to rules file, update $(stampdir_targets) and .PHONY
+ target.
+ + Split MANIFEST files into arch-specific and arch-independent parts.
+ Update the README to explain how these files work now and how to update
+ them.
+ + Ship the XTerm control sequences document only in the xspecs package.
+ Consequences:
+ + Reduce the disk space required to build the package by ~520MB when the
+ "binary-arch" target is used (e.g., by "dpkg-buildpackage -B".)
+ + Reduce the build time by 20 to 30 percent when the "binary-arch" target
+ is used (ccache users might see even more improvement).
+ + Make the buildd admins hate us less.
+
+ * Rewrite migration of app-defaults (in libxt6) and xkb (in libx11-6)
+ directories.
+ + Enhance the shell libary with new functions:
+ - Implement check_symlink() and change check_symlinks_and_warn() and
+ check_symlinks_and_bomb() to use it.
+ - Implement make_symlink_sane(). Ensure that the specified symbolic
+ link symlink exists, and points to the specified target. If symlink
+ does not exist, create it and point it at target. If symlink exists
+ but is not a symbolic link, back it up. If symlink exists, is a
+ symbolic link, but points to the wrong location, fix it. If symlink
+ exists, is a symbolic link, and already points to target, do nothing.
+ This function wouldn't be needed if ln had an -I, --idempotent option.
+ - Implement migrate_dir_to_symlink(). Per Debian Policy section 6.5.4,
+ "A directory will never be replaced by a symbolic link to a directory
+ or vice versa; instead, the existing state (symlink or not) will be
+ left alone and dpkg will follow the symlink if there is one." We have
+ to do it ourselves. This function moves the contents of old_location,
+ a directory, into new_location, a directory, then makes old_location a
+ symbolic link to new_location. old_location need not exist, but if it
+ does, it must be a directory (or a symlink to a directory). If it is
+ not, it is backed up. If new_location exists already and is not a
+ directory, it is backed up.
+ + Add config scripts for libx11-6 and libxt6, which detect if
+ /usr/X11R6/lib/X11/{app-defaults,xkb} is a directory but not a symlink,
+ and use debconf to prompt the user (at low priority) whether the
+ directories should be moved. If the user says no, abort the script. If
+ the user says yes (default), start the migration with
+ migrate_dir_to_symlink().
+ + Update preinst scripts to use debconf as the configs script do; this is
+ unusual but necessary because if we're going to do the migration we need
+ to start before any other packages (which may use the old directories)
+ are unpacked.
+ + In the preinst scripts, call check_symlink(), and if that fails, call
+ make_symlink_sane(). Stop calling check_symlinks_and_warn().
+ + Update postinst scripts to indicate that migration is completed.
+ + Update postrm scripts to run migration in reverse if it had been started
+ and the package installation or upgrade is aborted.
+ + We keep shipping /etc/X11/{app-defaults,xkb} symlinks as part of the
+ package payloads, because manipulating them in the preinst isn't enough
+ -- dpkg will remove "disappearing" files from the system *after* the
+ preinst runs but before the postinst does, leaving a window for packages
+ using the legacy paths to unpack to the old location with no symlink to
+ redirect them, which is precisely what we don't want (see section 6.5 of
+ the Debian Policy Manual).
+
+ * Make xserver-xfree86 config script's validate_monitor_frequency_db_input()
+ function more strict, so it doesn't regard open-ended ranges like "40-" as
+ valid.
+
+ * Fix some lingering references to an unreleased version of xserver-common
+ in the debian/control file; bump the versioning up to the subsequent
+ package release.
+
+ Changes by Robert Millan:
+
+ * Get glxinfo to link properly with g++ (instead of adding -lstdc++).
+
+ Changes by Fabio M. Di Nitto:
+
+ * Fix default drivers lists in xserver-xfree86.config.in. The via driver is
+ only available for hurd-i386 and i386.
+
+ * Fix bug in xserver-xfree86 configuration script where failure of
+ autodetection would cause the script to think it had detected one X server
+ or XFree86 video driver instead ($(echo "" | wc -l) is "1", not "0").
+
+ * Add French and Danish translation of "Generic Monitor" to
+ xserver-xfree86.config.in. Thanks to Christian Perrier and Anders Dybdal.
+
+ * Add Danish translation of "Generic Video Card" to
+ xserver-xfree86.config.in. Thanks to Anders Dybdal.
+
+ Changes by Denis Barbier:
+
+ * Grab from XFree86 CVS (no later than 2003-12-18) several fixes to XKB:
+ + Fix for XKB map 'altwin' to avoid one keysym to more than one modifier
+ mapping (Ivan Pascal). This really fixes the Alt-Tab issue.
+ (Closes: #256706)
+ + Fix xmodmap's output of the modifiers map to not display BadKey when
+ the first column keysym is empty (Ivan Pascal).
+ + The "modifier_add none" construct was introduced by Ivan Pascal some
+ times ago, but it was not documented in the changelog and a piece was
+ missing to enable this feature. No keymaps currently use this feature,
+ but users may find it useful with broken keymaps.
+
+ * Update Italian debconf template translations (thanks, Danilo Piazzalunga).
+ (Closes: #270126)
+
+ * Update French debconf template translations (thanks, Clement Stenac).
+
+ -- Fabio M. Di Nitto <fabbione@fabbione.net> Mon, 06 Sep 2004 09:54:18 +0200
+
xfree86 (4.3.0.dfsg.1-6) unstable; urgency=low
Changes by Branden Robinson:
Reply to: