Bug#639646: Filtering performance fix
This patch significantly improves performance of message searching/filtering with gtk3+ (or gtk2 with accessibility on). -- Jindrich Makovicka 03_a11y.dpatch Description: application/shellscript
Bug#563590: possible fix
This patch should fix the issue by wrapping the grep and subsequent $? test with set +e and set -e. It also adds a check for VT in screensaver setup, which has also sense only in the console. -- Jindrich Makovicka --- console-screen.sh.orig 2009-12-19 20:27:00.0 +0100 +++ console-screen.sh 2010-01-07 11:24:54.182001990 +0100 @@ -63,6 +63,20 @@ fi +# Detect the console type +VT=no +CONSOLE_TYPE=`fgconsole 2/dev/null` || CONSOLE_TYPE=unknown +if [ ! $CONSOLE_TYPE = serial ] ; then +set +e +readlink /proc/self/fd/0 | grep -q -e /dev/vc -e '/dev/tty[^p]' -e /dev/console +if [ $? -eq 0 ] ; then +VT=yes +fi +set -e +fi + + + reset_vga_palette () { if [ -f /proc/fb ]; then @@ -77,20 +91,10 @@ # be sure the main program is installed [ -x ${SETFONT} ] || return 0 -VT=no -# If we can't access the console, quit -CONSOLE_TYPE=`fgconsole 2/dev/null` || return 0 - -if [ ! $CONSOLE_TYPE = serial ] ; then - readlink /proc/self/fd/0 | grep -q -e /dev/vc -e '/dev/tty[^p]' -e /dev/console - if [ $? -eq 0 ] ; then - VT=yes - reset_vga_palette - fi -fi - [ $VT = no ] return 0 +reset_vga_palette + # start vcstime if [ ${DO_VCSTIME} = yes -a -x ${VCSTIME} ] ; then [ $VERBOSE != no ] log_action_begin_msg Starting clock on text console @@ -181,6 +185,9 @@ } setup_screensaver() { + +[ $VT = no ] return 0 + # screensaver stuff setterm_args= if [ $BLANK_TIME ]; then
Bug#414036: updated patch
As the bug still persists, and one of the hunks in the previous patch was incorrect, here is an update. Rationale for the changes: Hunk 1: This should apply to the words with the -os suffix. - Dionýsos/D = Dionýsech - Eustachos/D = Eustachy the -us version is further down. Hunk 2: This should apply to the words with the -es suffix. - diabetes/Q = diabetech Hunk 3: This should apply to the words with the -ám suffix. - lkám = lkej, lkejme, lkejte there is no word ending on -ém having J suffix in the dictionary Hunk 4: This should apply to the words with the -oupit suffix. - dokoupit/ATN = dokup, dokupme, dokupte the -oužit version is further down -- Jindrich Makovicka --- cs_CZ.aff.orig 2005-11-28 17:16:29.0 +0100 +++ cs_CZ.aff 2007-03-08 20:33:10.0 +0100 @@ -285,8 +285,8 @@ SFX D osové os SFX D osù os SFX D osùm os -SFX D usech [^ighk]os -SFX D usy [^i]os +SFX D osech [^ighk]os +SFX D osy [^i]os SFX D osích ios SFX D osách kos SFX D osi ios @@ -595,7 +595,7 @@ SFX Q esy es SFX Q esù es SFX Q esùm es -SFX Q osech [^ghk]es +SFX Q esech [^ghk]es SFX Q ges zíchges SFX Q ches ¹íchches SFX Q hes zích[^c]hes @@ -900,7 +900,7 @@ SFX M o i [¾¹èøjeiy]o SFX M a ùm a SFX M a ách [ghk]a -SFX M o ech [^ghkei]a +SFX M a ech [^ghkei]a SFX M a y [^¾¹èøcji]a SFX M ona on SFX M onu on @@ -1309,9 +1309,9 @@ SFX J ámaje ám SFX J ámajícám SFX J ámajíce ám -SFX J émej ám -SFX J émejmeám -SFX J émejteám +SFX J ámej ám +SFX J ámejmeám +SFX J ámejteám SFX J m ¹ ím SFX J m 0 ím SFX J 0 e ím @@ -1759,9 +1759,9 @@ SFX A ounit ouniounit SFX A ounit ounìme ounit SFX A ounit ounìte ounit -SFX A ou¾it up oupit -SFX A ou¾it upmeoupit -SFX A ou¾it upteoupit +SFX A oupit up oupit +SFX A oupit upmeoupit +SFX A oupit upteoupit SFX A ouøit uø ouøit SFX A ouøit uømeouøit SFX A ouøit uøteouøit
Bug#534512: regression: pieces have lost borders and become transparent
There are a lot of messages like this: xboard: Can't open bitmap file /usr/share/games/xboard/bitmaps.xchess/.64s.bm...using built-in I guess there is some typo in a printf format string (. instead of %). -- Jindrich Makovicka -- To UNSUBSCRIBE, email to debian-bugs-dist-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org
Bug#483561: 64bit distribution
It would be nice to have a package for x86-64 also. It seems that it suffices to remove dh_strip from the rules file to make the amd64 version build. As the closed source binaries are stripped anyway, it should be ok to remove this line. Regards, -- Jindrich Makovicka -- To UNSUBSCRIBE, email to debian-bugs-dist-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org
Bug#499412: [PATCH] fix slider name corruption
Package: gkrellm-volume Version: 2.1.13-1 When the ADDMIXER configuration command fails (eg. when the corresponding device is removed, subsequent SETDEVNAME commands will modify the name of the last added slider of the previous mixer, which is wrong. Setting the current slider pointer to NULL after adding a mixer fixes the issue. Patch attached. -- Jindrich Makovicka --- gkrellm-volume-2.1.13.orig/volume.c +++ gkrellm-volume-2.1.13/volume.c @@ -533,6 +533,7 @@ if (!strcmp(MUTEALL,command)) SET_FLAG(global_flags,MUTEALL); else if (!strcmp(ADDMIXER,command)) { m = add_mixer_by_id(arg); +s = NULL; } else if (!strcmp(RIGHT_CLICK_CMD,command)) { g_strlcpy(right_click_cmd, arg, sizeof(right_click_cmd)); } else if (!strcmp(ADDDEV,command)) {
Bug#497167: [PATCH] add support for -L pdftex
Package: transfig Version: 1:3.2.5-rel-3.1 Hi, although fig2dev is able to produce a .pdf and a .tex file from a FIG input, transfig is currently cannot create a Makefile for such case. The attached patch add support for -L pdftex, which is handy if you want to include some figures in a document processed by pdflatex. Regards, -- Jindrich Makovicka 19_transfig_pdftex.dpatch Description: Binary data
Bug#492693: impossible to link due to missing export in .so
Package: libpqxx Version: 2.6.9-6 Current libpqxx in Lenny is missing freemem_result_data() in the .so library, and is unusable due to this. It can be fixed by explicitly declaring the function as exportable: --- result.hxx.orig 2007-01-29 20:18:52.0 +0100 +++ result.hxx 2008-07-28 10:45:17.0 +0200 @@ -60,7 +60,7 @@ ~result_data(); }; -void freemem_result_data(result_data *) throw (); +void PQXX_LIBEXPORT freemem_result_data(result_data *) throw (); template inline void PQAllocresult_data::freemem() throw () { freemem_result_data(m_Obj); } } // namespace internal -- Jindrich Makovicka -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Bug#490275: missing pyrhon-setuptools in Depend:
Definitely it's the packaging problem. In addition, trac should depend on python-setuptools as it cannot run without them. -- Jindrich Makovicka -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Bug#468894: wrong arguments to run-mailcap in mc.ext
Package: mc Version: 1:4.6.2~pre1-5 Hi, current version of mc in Sid is missing the colon character between the mime type and filename in all run-mailcap calls in mc.ext - there is a space instead. This causes (at least) lots of bogus error messages like Error: no such file application/pdf to be printed. If this change has been made to fix #442024, better remove the mime type completely. Regards, -- Jindrich Makovicka -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Bug#466199: unable to launch executables
Package: mc Version: 1:4.6.2~pre1-2 Severity: important Due to last changes in mc.ext, mc is now unable to start executables by pressing enter, because run-mailcap cannot find a view mailcap rule. Removing the run-mailcap --action=edit %f from the default target at the end of mc.ext fixes the issue. -- Jindrich Makovicka -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Bug#466019: nfsmount segfaults due to protocol version mismatch
Package: nfs-common Version: 1.1.1-13 Severity: important In current nfsmount, two tests of the mount result version are out of sync. The test in nfs_call_mount tests the mnt_server-pmap.pm_vers to decide what mount function (v2 or v3) to call, but later in nfsmount(), (nfs_pmap-pm_vers == 2) condition is used to decide about the result format, which is wrong. The result is then misinterpreted, causing a segfault. With nfsmount being SUID root, this could be quite serious, although I am currently unsure if it's exploitable. Trivial fix attached. -- Jindrich Makovicka --- nfsmount.c.orig 2007-10-19 05:07:28.0 +0200 +++ nfsmount.c 2008-02-15 23:43:32.0 +0100 @@ -714,7 +714,7 @@ rpc_mount_errors(*nfs_server.hostname, 1, bg); } - if (nfs_pmap-pm_vers == 2) { + if (mnt_pmap-pm_vers = 2) { if (mntres.nfsv2.fhs_status != 0) { nfs_error(_(%s: %s:%s failed, reason given by server: %s), progname, hostname, dirname,
Bug#449178: 21.6 build
If you are interested, I put up a build of 21.6 with minor cleanups like Debian patches split to the debian/patches directory. It also contains the fix for using larger hash tables (24MB) under a normal user I posted here some time ago. You can download it from http://kmlinux.fjfi.cvut.cz/~makovick/crafty/ . The enormous opening book from the Crafty ftp is also available. Regards, -- Jindrich Makovicka -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Bug#460340: exclude is broken in dh_installdocs and dh_installexamples
Package: debhelper Version: 6.0.1 The added code which appends stuff at the beginning of $exclude variable can be executed more than once, being in a loop, and it breaks with more than one -X argument. Patch attached. -- Jindrich Makovicka diff -ur old/dh_installdocs new/dh_installdocs --- old/dh_installdocs 2008-01-12 09:30:11.0 +0100 +++ new/dh_installdocs 2008-01-12 09:38:24.0 +0100 @@ -137,8 +137,8 @@ my $dir = ($basename eq '.') ? $doc : $doc/..; my $pwd=`pwd`; chomp $pwd; -$exclude='\\( -type f -or -type l \\)'.$exclude; -complex_doit(cd '$dir' find '$basename' $exclude -exec cp --parents -dp {} $pwd/$tmp/usr/share/doc/$package \\;); +my $exclude2='\\( -type f -or -type l \\)'.$exclude; +complex_doit(cd '$dir' find '$basename' $exclude2 -exec cp --parents -dp {} $pwd/$tmp/usr/share/doc/$package \\;); } else { doit(cp, -a, $doc, $tmp/usr/share/doc/$package); diff -ur old/dh_installexamples new/dh_installexamples --- old/dh_installexamples 2008-01-12 09:30:34.0 +0100 +++ new/dh_installexamples 2008-01-12 09:38:33.0 +0100 @@ -90,8 +90,8 @@ my $dir = ($basename eq '.') ? $example : $example/..; my $pwd=`pwd`; chomp $pwd; -$exclude = '-type f'.$exclude; -complex_doit(cd '$dir' find '$basename' $exclude -exec cp --parents -dp {} $pwd/$tmp/usr/share/doc/$package/examples \\;); +my $exclude2 = '-type f'.$exclude; +complex_doit(cd '$dir' find '$basename' $exclude2 -exec cp --parents -dp {} $pwd/$tmp/usr/share/doc/$package/examples \\;); } else { doit(cp, -a, $example, $tmp/usr/share/doc/$package/examples);
Bug#459368: applications spawned by GE linked with wrong libraries
Package: googleearth-package Version: 0.4.3 Google Earth starting script currently adjusts and exports LD_LIBRARY_PATH to include Google Earth's own libraries, most notably its own libpng. This means that all programs spawned by Google Earth (eg. browser) try to link the incompatible libraries, which for example break image display in Iceweasel. A possible workaround is to launch Google Earth binary explicitly using ld.so, avoiding setting of any environment variables, which could affect the child processes: exec /lib/ld-2.7.so \ --library-path /usr/lib/googleearth \ /usr/lib/googleearth/googleearth-bin $@ -- Jindrich Makovicka -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Bug#456606: Deprecated function warning causes an error 500 in lighttpd
Package: trac-bzr Version: 0.2+bzr31-3 Hi, Since get_weave has been deprecated in bzr 0.90, the fastcgi version of trac produces a warning about the deprecation on a first run. This causes the first changelog served by trac lighttpd return a 500 Internal server error, which is annoying. The following patch fixes the issue. --- backend.py.orig 2007-03-08 20:24:40.0 +0100 +++ backend.py 2007-12-16 22:17:46.0 +0100 @@ -671,7 +671,7 @@ It may also include a few revisions in which the revision_id did not change, if the modification was subsequently undone. -weave = self.tree.get_weave(self.entry.file_id) +weave = self.tree._get_weave(self.entry.file_id) file_ancestry = weave.get_ancestry(self.entry.revision) # Can't use None here, because it's a legitimate revision id. last_yielded = 'bogus:' Regards, -- Jindrich Makovicka -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Bug#443883: [PATCH] wxgtk2.6 crashes with recent gtk+
Package: wxwidgets2.6 Version: 2.6.3.2.1.5 Severity: grave Hi, with wxWidgets 2.6 and recent GTK+ in Sid, almost all applications using wxWidgets crash due to bugs in GtkBorder handling. I have observed these crashes in aMule (detail dialog), Audacity (about, preferences) and FileZilla (crashes right at startup). The issue can be fixed by replacing g_free with gtk_border_free as in the patch. See also http://mail.gnome.org/archives/desktop-devel-list/2007-August/msg00256.html Regards, -- Jindrich Makovicka diff -ur gtk.orig/button.cpp gtk/button.cpp --- gtk.orig/button.cpp 2006-05-01 06:38:40.0 +0200 +++ gtk/button.cpp 2007-09-24 21:20:37.0 +0200 @@ -89,7 +89,7 @@ right_border += default_border-right; top_border += default_border-top; bottom_border += default_border-bottom; -g_free( default_border ); +gtk_border_free( default_border ); } #else left_border = 6; diff -ur gtk.orig/window.cpp gtk/window.cpp --- gtk.orig/window.cpp 2006-05-01 06:38:40.0 +0200 +++ gtk/window.cpp 2007-09-24 21:47:37.0 +0200 @@ -3276,7 +3276,7 @@ right_border += default_border-right; top_border += default_border-top; bottom_border += default_border-bottom; -g_free( default_border ); +gtk_border_free( default_border ); } #else left_border = 6;
Bug#440528: New upstream 0.6.1 available
Package: yasm Version: 0.5.0-2 Severity: wishlist Hi, there is a new upstream release, featuring SSE3/SSE4 support among others. Given that nasm 0.99.01 in unstable is rather broken (http://www.google.com/search?q=nasm+invalid+operands+in+non-64-bit+mode), it would be nice to have current yasm available as a package. Regards, -- Jindrich Makovicka -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Bug#440102: .bashrc should set HISTCONTROL=ignoreboth when already set to ignorespace
Package: bash Version: 3.1dfsg-8 GNU Midnight Commander uses HISTCONTROL=ignorespace to hide some commands issued to bash like setting of the prompt or xterm title using escape sequences. Due to HISTCONTROL=ignoredups hardcoded in .bashrc, mc polutes .bash_history on each directory change. A possible fix is to check if HISTCONTROL is already set to ignorespace, and in that case, promote it to ignoreboth: if [ $HISTCONTROL = ignorespace ] ; then export HISTCONTROL=ignoreboth else export HISTCONTROL=ignoredups fi Regards, -- Jindrich Makovicka -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Bug#434199: resolvconf script lockup
Package: pdnsd Version: 1.2.5par-2 Severity: important Hi, resolvconf script incorrectly expects the output of /lib/resolvconf/list-records to be non-empty. When an empty list is returned, the script hangs because cat expects data from stdin. On my configuration, this bug disallows the machine to shutdown correctly, and it also needs some Ctrl-C hitting to boot. Patch attached. -- Jindrich Makovicka --- pdnsd.orig 2007-07-22 12:08:05.0 +0200 +++ pdnsd 2007-07-22 12:04:52.0 +0200 @@ -37,9 +37,15 @@ done } -uniquify `cat $(/lib/resolvconf/list-records) \ -| sed -n -e 's/^[[:space:]]*nameserver[[:space:]]\+//p' \ -| grep -v '^127.0.0.1$'` +FILES=$(/lib/resolvconf/list-records) + +if test -n $FILES ; then +uniquify `cat $FILES \ +| sed -n -e 's/^[[:space:]]*nameserver[[:space:]]\+//p' \ +| grep -v '^127.0.0.1$'` +else +RSLT= +fi if [ -n $RSLT ] ; then OUTPUT=$(/usr/sbin/pdnsd-ctl server resolvconf up $RSLT || :)
Bug#431740: [PATCH] claws-mail 2.10.0 compatibility
Package: gkrellm Version: 2.2.10-2 Hi, the attached patch makes gkrellm compatible with Claws 2.10.0, which has renamed the .sylpheed_mark files containing the information on read mails to .claws_mark. With the patch, both variants are recognized. -- Jindrich Makovicka 04_mail.c.dpatch Description: Binary data
Bug#430362: Patched REXX detection is slow with UTF-8 locale
Package: file Version: 4.21-1 Hi, although the CVE-2007-2026 patch avoids the DoS attack, it makes file much slower when a UTF-8 locale is used. Below are the times to identify a ~100K c++ source file observed on 2.8GHz Intel C2D: $ time LC_ALL=cs_CZ.UTF-8 file testfile.cpp testfile.cpp: ASCII C++ program text real0m0.604s user0m0.600s sys 0m0.003s $ time LC_ALL=C file testfile.cpp testfile.cpp: ASCII C++ program text real0m0.060s user0m0.056s sys 0m0.002s This is particularly annoying when using the Midnight Commander, as it uses file command to identify any file before viewing. commenting the lines 100 regex/c =^\\s{0,255}call\\s{1,99}rxfuncadd OS/2 REXX batch file text 100 regex/c =^\\s{0,255}say\ ['] OS/2 REXX batch file text fixes the problem. The slowdown didn't occur with the original vulnerable version either. Regards, -- Jindrich Makovicka -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Bug#427608: openoffice.org-l10n-cs package is empty
Package: openoffice.org Version: 2.2.1~rc3-1 openoffice.org-l10n-cs package contains only documentation, and it is not possible to select the Czech user interface. Only English is available in the Languages/User interface combo box. -- Jindrich Makovicka -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Bug#425298: amuleweb is built without libpng
Package: amule-daemon Version: 2.1.3-2 Severity: grave Current Sid contains a beta version of libpng, which is misdetected by amule's configure, and the latest amule-daemon has been probably built with it. This makes amule web server almost unusable. It also revealed that amuleweb can be trivially crashed by requesting a nonexistent image (in my other report). Build fix attached. -- Jindrich Makovicka diff -u a/acinclude.m4 b/acinclude.m4 --- a/acinclude.m4 2006-03-27 15:20:38.0 +0200 +++ b/acinclude.m4 2007-05-20 20:51:02.0 +0200 @@ -258,11 +258,11 @@ LIBPNG_VERSION=`$LIBPNG_CONFIG_WITH_ARGS --version` libpng_config_major_version=`echo $LIBPNG_VERSION | \ - sed 's/\([[0-9]]*\).\([[0-9]]*\).\([[0-9]]*\)/\1/'` + sed 's/\([[0-9]]*\).\([[0-9]]*\).\([[0-9]]*\).*/\1/'` libpng_config_minor_version=`echo $LIBPNG_VERSION | \ - sed 's/\([[0-9]]*\).\([[0-9]]*\).\([[0-9]]*\)/\2/'` + sed 's/\([[0-9]]*\).\([[0-9]]*\).\([[0-9]]*\).*/\2/'` libpng_config_micro_version=`echo $LIBPNG_VERSION | \ - sed 's/\([[0-9]]*\).\([[0-9]]*\).\([[0-9]]*\)/\3/'` + sed 's/\([[0-9]]*\).\([[0-9]]*\).\([[0-9]]*\).*/\3/'` libpng_requested_major_version=`echo $min_libpng_version | \ sed 's/\([[0-9]]*\).\([[0-9]]*\).\([[0-9]]*\)/\1/'` diff -u a/configure b/configure --- a/configure 2006-06-11 17:08:53.0 +0200 +++ b/configure 2007-05-20 20:53:10.0 +0200 @@ -6917,11 +6917,11 @@ LIBPNG_VERSION=`$LIBPNG_CONFIG_WITH_ARGS --version` libpng_config_major_version=`echo $LIBPNG_VERSION | \ - sed 's/\([0-9]*\).\([0-9]*\).\([0-9]*\)/\1/'` + sed 's/\([0-9]*\).\([0-9]*\).\([0-9]*\).*/\1/'` libpng_config_minor_version=`echo $LIBPNG_VERSION | \ - sed 's/\([0-9]*\).\([0-9]*\).\([0-9]*\)/\2/'` + sed 's/\([0-9]*\).\([0-9]*\).\([0-9]*\).*/\2/'` libpng_config_micro_version=`echo $LIBPNG_VERSION | \ - sed 's/\([0-9]*\).\([0-9]*\).\([0-9]*\)/\3/'` + sed 's/\([0-9]*\).\([0-9]*\).\([0-9]*\).*/\3/'` libpng_requested_major_version=`echo $min_libpng_version | \ sed 's/\([0-9]*\).\([0-9]*\).\([0-9]*\)/\1/'` Only in b: configure~
Bug#425300: remote crash of amuleweb
Package: amule-daemon Version: 2.1.3-2 Severity: grave Amuleweb does not check if a requested image actually exists. This means requests like http://localhost:4711/xyzzy.png (but only if the user is logged in) crash the webserver due to a NULL pointer dereference. Fix attached. -- Jindrich Makovicka --- WebServer.cpp.orig 2006-04-14 07:38:51.0 +0200 +++ WebServer.cpp 2007-05-20 20:23:51.0 +0200 @@ -303,6 +303,12 @@ // To prevent access to non-template images, we disallow use of paths in filenames. wxString imgName = wxFileName::GetPathSeparator() + wxFileName(Data.sURL).GetFullName(); CAnyImage *img = m_ImageLib.GetImage(imgName); + + if (!img) { + webInterface-DebugShow(wxT( imgrequest: failed, not found\n)); + ProcessURL(Data); + return; + } // Only static images are available to visitors, in order to prevent // information leakage, but still allowing images on the login page.
Bug#425298: amuleweb is built without libpng
On Sun, 20 May 2007 21:18:20 +0200 Adeodato Simó [EMAIL PROTECTED] wrote: Version: 2.1.3-3 * Jindrich Makovicka [Sun, 20 May 2007 21:03:29 +0200]: Package: amule-daemon Version: 2.1.3-2 Severity: grave Current Sid contains a beta version of libpng, which is misdetected by amule's configure, and the latest amule-daemon has been probably built with it. This makes amule web server almost unusable. It also revealed that amuleweb can be trivially crashed by requesting a nonexistent image (in my other report). Build fix attached. Dude, I don't know where you got the patch from, but it's identical to the one I wrote this morning to fix this very same bug. http://packages.qa.debian.org/a/amule/news/20070520T120203Z.html http://people.debian.org/~adeodato/code/packages/amule/debian/patches/configure_proper_libpng_detection.diff Well, it's the most obvious way to fix it :) Regards, -- Jindrich Makovicka
Bug#425300: remote crash of amuleweb
On Sun, 20 May 2007 21:22:44 +0200 Adeodato Simó [EMAIL PROTECTED] wrote: severity 425300 normal thanks * Jindrich Makovicka [Sun, 20 May 2007 21:09:04 +0200]: Amuleweb does not check if a requested image actually exists. This means requests like http://localhost:4711/xyzzy.png (but only if the user is logged in) crash the webserver due to a NULL pointer dereference. Okay, but I'm adjusting the severity of this bug, given that (a) user must be logged in, and (b) a wrong URL must be provided by hand, which makes it not happen in the common case. Fix attached. Does this fix come from upstream? No, it's been hand crufted right here, but I didn't check their upstream or bug trackers if it is already fixed or not. Regards, -- Jindrich Makovicka
Bug#421502: build problem: CFLAGS are discarded due to a typo
Package: timidity Version: 2.13.2-11 Because of a typo, EXTRACT_CPPFLAGS destroys the CFLAGS when called. This causes a problem when one wants to rebuild the package with optimizations (on core 2 duo, rebuilding with -march=pentium-m -mtune=pentium-m -msse2 -mfpmath=sse lowers the cpu load about 4-5 times, it is quite a reasonable thing to do). The fix is trivial: --- utils.m4.orig 2007-04-29 17:20:47.0 +0200 +++ utils.m42007-04-29 17:20:54.0 +0200 @@ -242,7 +242,7 @@ [for f in $3; do case .$f in .-I?*|.-D?*)$1=[$]$1 $f ;; - *) $2=[$]$1 $f ;; + *) $2=[$]$2 $f ;; esac done ]) -- Jindrich Makovicka -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Bug#414036: [PATCH] fix typos in Czech affixes
Package: myspell-cs-cz Version: 20040229-4 Hi, newer versions of OpenOffice.org complain about stripping characters incompatible with stripping conditions. The attached patch (hopefully correctly) fixes the typos in the affix file and removes the warnings. -- Jindrich Makovicka --- cs_CZ.aff.orig 2005-11-28 17:16:29.0 +0100 +++ cs_CZ.aff 2007-03-08 20:33:10.0 +0100 @@ -285,8 +285,8 @@ SFX D osové os SFX D osù os SFX D osùm os -SFX D usech [^ighk]os -SFX D usy [^i]os +SFX D osech [^ighk]os +SFX D osy [^i]os SFX D osích ios SFX D osách kos SFX D osi ios @@ -595,7 +595,7 @@ SFX Q esy es SFX Q esù es SFX Q esùm es -SFX Q osech [^ghk]es +SFX Q esech [^ghk]es SFX Q ges zíchges SFX Q ches ¹íchches SFX Q hes zích[^c]hes @@ -900,7 +900,7 @@ SFX M o i [¾¹èøjeiy]o SFX M a ùm a SFX M a ách [ghk]a -SFX M o ech [^ghkei]a +SFX M a ech [^ghkei]a SFX M a y [^¾¹èøcji]a SFX M ona on SFX M onu on @@ -1309,9 +1309,9 @@ SFX J ámaje ám SFX J ámajícám SFX J ámajíce ám -SFX J émej ám -SFX J émejmeám -SFX J émejteám +SFX J émej ém +SFX J émejmeém +SFX J émejteém SFX J m ¹ ím SFX J m 0 ím SFX J 0 e ím @@ -1759,9 +1759,9 @@ SFX A ounit ouniounit SFX A ounit ounìme ounit SFX A ounit ounìte ounit -SFX A ou¾it up oupit -SFX A ou¾it upmeoupit -SFX A ou¾it upteoupit +SFX A oupit up oupit +SFX A oupit upmeoupit +SFX A oupit upteoupit SFX A ouøit uø ouøit SFX A ouøit uømeouøit SFX A ouøit uøteouøit
Bug#411224: [PATCH] get rid of the hash size 24M problem
Package: crafty Version: 20.14-1 Hi, the attached patch replaces the shmget() call with mmap(MAP_SHARED | MAP_ANONYMOUS), available since the 2.4 kernel. The advantage of this approach is that it is available to an ordinary user without any size restrictions, which removes the necessity to modify the kernel parameters by root if one wants to use larger hash hashp values. In addition, it is guaranteed that the mmap()ped memory will be released in the case of application crash. -- Jindrich Makovicka diff -ur orig/crafty-20.14/chess.h crafty-20.14/chess.h --- orig/crafty-20.14/chess.h 2007-02-14 08:08:40.0 +0100 +++ crafty-20.14/chess.h 2007-02-14 07:49:49.0 +0100 @@ -708,7 +708,7 @@ int SetRootAlpha(unsigned char, int); int SetRootBeta(unsigned char, int); void *SharedMalloc(size_t, int); -void SharedFree(void *address); +void SharedFree(void *address, size_t); void SignalInterrupt(int); int StrCnt(char *, char); int Swap(TREE * RESTRICT, int, int, int); diff -ur orig/crafty-20.14/init.c crafty-20.14/init.c --- orig/crafty-20.14/init.c 2006-06-20 15:57:17.0 +0200 +++ crafty-20.14/init.c 2007-02-14 07:48:22.0 +0100 @@ -236,8 +236,8 @@ pawn_hash_table = (PAWN_HASH_ENTRY *) SharedMalloc(cb_pawn_hash_table, 0); if (!trans_ref) { Print(128, malloc() failed, not enough memory.\n); -SharedFree(trans_ref); -SharedFree(pawn_hash_table); +SharedFree(trans_ref, cb_trans_ref); +SharedFree(pawn_hash_table, cb_pawn_hash_table); hash_table_size = 0; pawn_hash_table_size = 0; log_hash = 0; diff -ur orig/crafty-20.14/option.c crafty-20.14/option.c --- orig/crafty-20.14/option.c 2007-02-14 08:08:40.0 +0100 +++ crafty-20.14/option.c 2007-02-14 07:49:22.0 +0100 @@ -1227,7 +1227,7 @@ } if (new_hash_size 0) { if (hash_table_size) { - SharedFree(trans_ref); + SharedFree(trans_ref, cb_trans_ref); } new_hash_size /= 16 * 3; for (log_hash = 0; log_hash (int) (8 * sizeof(int)); log_hash++) @@ -1239,7 +1239,7 @@ trans_ref = (HASH_ENTRY *) SharedMalloc(cb_trans_ref, 0); if (!trans_ref) { printf(malloc() failed, not enough memory.\n); -SharedFree(trans_ref); +SharedFree(trans_ref, cb_trans_ref); hash_table_size = 0; log_hash = 0; trans_ref = 0; @@ -1282,7 +1282,7 @@ return (1); } if (pawn_hash_table) { -SharedFree(pawn_hash_table); +SharedFree(pawn_hash_table, cb_pawn_hash_table); pawn_hash_table_size = 0; log_pawn_hash = 0; pawn_hash_table = 0; @@ -1297,7 +1297,7 @@ pawn_hash_table = (PAWN_HASH_ENTRY *) SharedMalloc(cb_pawn_hash_table, 0); if (!pawn_hash_table) { printf(malloc() failed, not enough memory.\n); -SharedFree(pawn_hash_table); +SharedFree(pawn_hash_table, cb_pawn_hash_table); pawn_hash_table_size = 0; log_pawn_hash = 0; pawn_hash_table = 0; diff -ur orig/crafty-20.14/utility.c crafty-20.14/utility.c --- orig/crafty-20.14/utility.c 2006-06-20 15:57:17.0 +0200 +++ crafty-20.14/utility.c 2007-02-14 08:14:55.0 +0100 @@ -46,6 +46,9 @@ #include sys/ioctl.h # endif # if defined(SMP) +#if defined (LINUX) +# include sys/mman.h +#endif #include signal.h #include sys/wait.h # endif @@ -54,7 +57,7 @@ # if !defined(CLK_TCK) static clock_t clk_tck = 0; # endif -# if defined(SMP) +# if defined(SMP) !defined(LINUX) #include sys/ipc.h #include sys/shm.h # endif @@ -2718,6 +2721,9 @@ { #if defined(UNIX) # if defined(SMP) +#if defined (LINUX) + return (mmap(NULL, size, PROT_READ | PROT_WRITE, MAP_SHARED | MAP_ANONYMOUS, -1, 0)); +#else int shmid; void *shared; @@ -2737,6 +2743,7 @@ shared = shmat(shmid, 0, 0); shmctl(shmid, IPC_RMID, 0); return (shared); +#endif # else return (malloc(size)); # endif @@ -2749,11 +2756,16 @@ #endif } -void SharedFree(void *address) +void SharedFree(void *address, size_t size) { #if defined(SMP) # if defined(UNIX) +#if defined (LINUX) + if (address) +munmap(address, size); +#else shmdt(address); +#endif # else VirtualFree(address, 0, MEM_RELEASE); # endif
Bug#388841: more recent changelog
Newer ChangeLog entries are in the Trac wiki: http://libtorrent.rakshasa.no/wiki/Changelog I just converted them to the ChangeLog format so I am attaching it. Regards, -- Jindrich Makovicka 2006-01-26 Jari Petter Sundell [EMAIL PROTECTED] * libtorrent 0.11.2 released * Added max_open_http option that limits the max number of simultaneous http connections. By default set to 32. * Directory structured view of the file list. The '/' key collapses directories, while the right arrow or ^F enters the selected directory. Changing the priority for a directory changes all contained files. * Replaced old autoconf scripts with pkg-config checks. If the curl or openssl check now fails, make sure pkg-config can find them. * Added create_link and delete_link options which apply to a download. These must be used with the new on_* options, so as to allow you to create symbolic links when starting, stopping, finishing, etc, a torrent. 2006-12-28 Jari Petter Sundell [EMAIL PROTECTED] * libtorrent 0.11.1 released * Added dired view of files and allow priority to be changed for all files within a directory. * Allow initial hash checking to be paused and resumed. * Minor bug fixes. 2006-12-13 Jari Petter Sundell [EMAIL PROTECTED] * libtorrent 0.11.0 released * Added tracker_numwant, encryption, connection_proxy, 'preload_type', 'preload_min_size', 'preload_required_rate', split_file_size, split_suffix, max_file_size, * Encrypted (obfuscated) handshake and connection support. Patch by Josef Drexler, public domain as per earlier agreement. * Append '' instead of '?' to the tracker url when the url already has '?'. * Added support for tunneling through HTTP 1.0 proxies. * Cut the size of the libtorrent binary from 925kb to 617kb by using using GCC's symbol export attributes and changes to how exceptions are constructed. * Moved client info, etc, from rtorrent to libtorrent and cleaned up the API. * Support for the CANCEL piece request message. (Finally) * Allow the user to configure the max file size a torrent is allowed to create, and the option to split large files. 2006-10-27 Jari Petter Sundell [EMAIL PROTECTED] * libtorrent 0.10.4 released * Use the actual completed torrent data in the 'downloaded' field sent to the tracker, rather than the download rate total. This should fix some problems with too much downloaded data being reported to the tracker. 2006-10-05 Jari Petter Sundell [EMAIL PROTECTED] * libtorrent 0.10.3 released * Didn't properly clean up when a torrent with the same info hash was rejected. This caused random crashes and memory leaks. * Properly cancel pieces that are skipped when downloading. 2006-09-14 Jari Petter Sundell [EMAIL PROTECTED] * libtorrent 0.10.2 released * Rebound the old ^P for commands to ^X due to a collision with up key. * The ^F and ^B keys can be used instead of left and right arrows. * Support for qwertz keyboard layout. * Fixed a bug that caused quick hash checking to mark chunks as unfinished if it was unable to map them. * Peer info with the corrupt upload counter is now saved in the session torrent. Uninteresting peers are removed at regular intervals. * Allow only one connection from a single host. A peer's listening port is now kept track of separately from the connection port. * An integer is used in the session torrent instead of a bitfield when either none or all chunks are done. * Added code for preloading chunks that get mapped, currently disabled. * Added kqueue support, which can be enabled with --with-kqueue. Doesn't work on MacOSX 10.4 due to a kernel bug, and has too long polling timeouts. * Added --with-address-space=MB option to change the assumed address space size. Currently set to 1GB on 32bit arch, 4GB on 64bit. * When checking how much free disk-space remains, any file or directory that is a symlink is tested as they may be on a different mount point. * Fixed most of the bugs reported, too numerous to list here. If a bug is still present in this release, let me know. 2006-08-16 Jari Petter Sundell [EMAIL PROTECTED] * libtorrent 0.10.1 released * Max memory usage for mapped chunks is now controlled by ulimit or the max_memory_usage option. This only controls mapped memory, not resident. * Improved syncing so that it tries to sync chunks that are close together. Also the safe_sync option may be used to force MS_SYNC on buggy
Bug#406367: [PATCH] use color instead of dim attribute for the curses driver
Package: aalib Version: 1.4p5-30 Hi, the attached patch makes the aacurses driver use black on black + bold for the dim letters instead of the dim attribute, if the color is available on the terminal. The reason is that A_DIM is ignored by linux console and most of the X terminals I am aware off, and linux console also reports that this attribute is supported, while rendering it the same way as normal color. With this patch, all (3) levels of brightness can be successfully utilized on the console :) Regards, -- Jindrich Makovicka --- aacurses.c.orig 2001-04-26 16:37:31.0 +0200 +++ aacurses.c 2007-01-06 15:50:14.0 +0100 @@ -36,8 +36,14 @@ dest-supported=AA_NORMAL_MASK; #ifdef HAVE_TERMATTRS a = termattrs(); -if (a A_DIM) +if (has_colors()) { + start_color(); + init_pair(1, COLOR_BLACK, COLOR_BLACK); dest-supported |= AA_DIM_MASK; +} else { + if (a A_DIM) + dest-supported |= AA_DIM_MASK; +} if (a A_BOLD) dest-supported |= AA_BOLD_MASK; if (a A_BOLD) @@ -86,7 +92,12 @@ attrset(A_NORMAL); break; case AA_DIM: - attrset(A_DIM); + if (has_colors()) { + attrset(A_BOLD); + attron(COLOR_PAIR(1)); + } else { + attrset(A_DIM); + } break; case AA_BOLD: attrset(A_BOLD);
Bug#402499: initramfs-tools: depmod wastes space and time, please replace it with a modules.dep file
IMHO there should be at least a dummy modules.dep to suppress the FATAL: Cannot load modules.dep messages during booting, when kernel tries to load net-pf-1 etc. depmod -a can later replace the dummy with a correct modules.dep. I currently use the attached hook to do this. -- Jindrich Makovicka modulesdep Description: Binary data
Bug#404817: Cannot build from source
Package: sylpheed-claws-gtk2 Version: 2.6.0-1 Hi, trying to build the package from source obtained via apt-get, I encountered the following problems: manual/xml2pdf, and manual/dtd/manual.xsl were missing and I had to download them from the upstream build dependencies do not contain fop and saxon. when these are not installed, the build process fails when trying to install the documentation. gdk-pixbuf is contained in gtk-2.0. libgdk-pixbuf-dev is a gtk-1.2 library and it is not necessary. moreover, its installation pulls a lot of obsolete gtk-1.2 stuff. Regards, -- Jindrich Makovicka -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Bug#395849: [PATCH] Typo in constellation lines
Package: kstars Version: 3.5.5-1 Hi, there is a typo in constellation line definition for Hercules causing the message kstars: WARNING: No star named alp2Her found. The attached patch fixes the definition so the lines reflect the version at http://en.wikipedia.org/wiki/Hercules_(constellation) . Regards, -- Jindrich Makovicka --- clines.dat.orig 2006-10-28 09:29:42.0 +0200 +++ clines.dat 2006-10-28 09:37:15.0 +0200 @@ -419,7 +419,9 @@ D zet Her M eps Her D del Her -D alp2Her +D mu Her +D xi Her +D omi Her M pi Her D rho Her D the Her
Bug#382300: /etc/hddtemp.db: New entry for Seagate 7200.10 320GB
Package: hddtemp Version: 0.3-beta15-21 Severity: wishlist ST3320620AS 194 C Seagate Barracuda 7200.10 320GB SATA Regards, -- Jindrich Makovicka -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Bug#382296: crash log
I guess I am experiencing the same, here is the log. [2006/08/10 08:02:00, 0] smbd/server.c:main(847) smbd version 3.0.23b started. Copyright Andrew Tridgell and the Samba Team 1992-2006 [2006/08/10 08:02:00, 0] passdb/pdb_interface.c:make_pdb_method_name(174) No builtin nor plugin backend for tdbsam guest found [2006/08/10 08:02:00, 0] lib/util.c:smb_panic(1592) PANIC (pid 8432): pdb_get_methods_reload: failed to get pdb methods for backend tdbsam guest [2006/08/10 08:02:00, 0] lib/util.c:log_stack_trace(1699) BACKTRACE: 7 stack frames: #0 /usr/sbin/smbd(log_stack_trace+0x23) [0x822b923] #1 /usr/sbin/smbd(smb_panic+0x46) [0x822ba16] #2 /usr/sbin/smbd [0x81ea8d1] #3 /usr/sbin/smbd(initialize_password_db+0xe) [0x81ea91e] #4 /usr/sbin/smbd(main+0x591) [0x82c1bf1] #5 /lib/tls/i686/cmov/libc.so.6(__libc_start_main+0xc8) [0xb7be2ea8] #6 /usr/sbin/smbd [0x8082a01] [2006/08/10 08:02:00, 0] lib/util.c:smb_panic(1600) smb_panic(): calling panic action [/usr/share/samba/panic-action 8432] [2006/08/10 08:02:02, 0] lib/util.c:smb_panic(1608) smb_panic(): action returned status 0 [2006/08/10 08:02:02, 0] lib/fault.c:dump_core(173) dumping core in /var/log/samba/cores/smbd Regards, -- Jindrich Makovicka -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Bug#378082: Please allow ttf-dejavu also (or just recommend the font)
Package: googleearth-package Version: 0.0.1 The DejaVu font is a quite widespread internationalized UI font based on the Bitstream Vera, and Google Earth runs fine with it alone, so I'd recommend to depend on ttf-bitstream-vera | ttf-dejavu or to move the font to the Recommended: line. -- Jindrich Makovicka -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Bug#356508: [patch] add a missing struct member initialization
Package: gkrellm-volume Version: 2.1.13-1 the attached patch adds a missing initialization of the changed_state member in the ALSA mixer. -- Jindrich Makovicka --- alsa_mixer.c.orig 2004-08-19 22:07:27.0 +0200 +++ alsa_mixer.c2006-03-12 15:30:05.774491108 +0100 @@ -118,7 +118,7 @@ alsaresult-sids = (snd_mixer_selem_id_t **) malloc(sizeof(snd_mixer_selem_id_t *) * count); alsaresult-ctltype = (int *) malloc(sizeof(int) * count); - + alsaresult-changed_state = 0; for (elem = snd_mixer_first_elem(handle), i = 0; elem; elem = snd_mixer_elem_next(elem)) {
Bug#351955: udev, hdparm and filesystem corruption
snd_util_memhdr_free Feb 5 08:11:36 holly kernel: snd_emu10k1: Unknown symbol __snd_util_memblk_new Feb 5 08:11:36 holly kernel: usb 3-1: new full speed USB device using uhci_hcd and address 2 Feb 5 08:11:36 holly kernel: usb 3-1: new device found, idVendor=04a9, idProduct=220e Feb 5 08:11:36 holly kernel: usb 3-1: new device strings: Mfr=1, Product=2, SerialNumber=0 Regards, -- Jindrich Makovicka -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Bug#347268: defoma hints are not up-to-date
Package: ttf-dejavu Version: 2.1-1 There were some naming changes and the condensed variants have to be added too. The attached patch should bring the hints into sync again. -- Jindrich Makovicka --- ttf-dejavu.hints.old2006-01-08 08:02:05.0 +0100 +++ ttf-dejavu.defoma-hints 2006-01-07 16:43:18.522094900 +0100 @@ -55,7 +55,65 @@ Foundry = Bitstream Priority = 20 end -begin /usr/share/fonts/truetype/ttf-dejavu/DejaVuSerif-Roman.ttf + +begin /usr/share/fonts/truetype/ttf-dejavu/DejaVuSansCondensed.ttf + Family = DejaVu-Sans + FontName = DejaVu-Sans + Encoding = Unicode + Location = English + Charset = ISO8859-1 ISO8859-2 ISO8859-9 ISO8859-15 ISO10646-1 + UniCharset = ISO8859-1 ISO8859-2 ISO8859-9 ISO8859-15 + GeneralFamily = SansSerif + Weight = Medium + Width = Variable + Shape = NoSerif Upright + Foundry = Bitstream + Priority = 20 +end +begin /usr/share/fonts/truetype/ttf-dejavu/DejaVuSansCondensed-Oblique.ttf + Family = DejaVu-Sans + FontName = DejaVu-Sans-Oblique + Encoding = Unicode + Location = English + Charset = ISO8859-1 ISO8859-2 ISO8859-9 ISO8859-15 ISO10646-1 + UniCharset = ISO8859-1 ISO8859-2 ISO8859-9 ISO8859-15 + GeneralFamily = SansSerif + Weight = Medium + Width = Variable + Shape = NoSerif Oblique + Foundry = Bitstream + Priority = 20 +end +begin /usr/share/fonts/truetype/ttf-dejavu/DejaVuSansCondensed-Bold.ttf + Family = DejaVu-Sans + FontName = DejaVu-Sans-Bold + Encoding = Unicode + Location = English + Charset = ISO8859-1 ISO8859-2 ISO8859-9 ISO8859-15 ISO10646-1 + UniCharset = ISO8859-1 ISO8859-2 ISO8859-9 ISO8859-15 + GeneralFamily = SansSerif + Weight = Bold + Width = Variable + Shape = NoSerif Upright + Foundry = Bitstream + Priority = 20 +end +begin /usr/share/fonts/truetype/ttf-dejavu/DejaVuSansCondensed-BoldOblique.ttf + Family = DejaVu-Sans + FontName = DejaVu-Sans-Bold-Oblique + Encoding = Unicode + Location = English + Charset = ISO8859-1 ISO8859-2 ISO8859-9 ISO8859-15 ISO10646-1 + UniCharset = ISO8859-1 ISO8859-2 ISO8859-9 ISO8859-15 + GeneralFamily = SansSerif + Weight = Bold + Width = Variable + Shape = NoSerif Oblique + Foundry = Bitstream + Priority = 20 +end + +begin /usr/share/fonts/truetype/ttf-dejavu/DejaVuSerif.ttf Family = DejaVu-Serif FontName = DejaVu-Serif Encoding = Unicode @@ -83,6 +141,92 @@ Foundry = Bitstream Priority = 20 end +begin /usr/share/fonts/truetype/ttf-dejavu/DejaVuSerif-Oblique.ttf + Family = DejaVu-Serif + FontName = DejaVu-Serif + Encoding = Unicode + Location = English + Charset = ISO8859-1 ISO8859-2 ISO8859-9 ISO8859-15 ISO10646-1 + UniCharset = ISO8859-1 ISO8859-2 ISO8859-9 ISO8859-15 + GeneralFamily = Roman + Weight = Medium + Width = Variable + Shape = Serif Oblique + Foundry = Bitstream + Priority = 20 +end +begin /usr/share/fonts/truetype/ttf-dejavu/DejaVuSerif-BoldOblique.ttf + Family = DejaVu-Serif + FontName = DejaVu-Serif-Bold + Encoding = Unicode + Location = English + Charset = ISO8859-1 ISO8859-2 ISO8859-9 ISO8859-15 ISO10646-1 + UniCharset = ISO8859-1 ISO8859-2 ISO8859-9 ISO8859-15 + GeneralFamily = Roman + Weight = Bold + Width = Variable + Shape = Serif Oblique + Foundry = Bitstream + Priority = 20 +end + +begin /usr/share/fonts/truetype/ttf-dejavu/DejaVuSerifCondensed.ttf + Family = DejaVu-Serif + FontName = DejaVu-Serif + Encoding = Unicode + Location = English + Charset = ISO8859-1 ISO8859-2 ISO8859-9 ISO8859-15 ISO10646-1 + UniCharset = ISO8859-1 ISO8859-2 ISO8859-9 ISO8859-15 + GeneralFamily = Roman + Weight = Medium + Width = Variable + Shape = Serif Upright + Foundry = Bitstream + Priority = 20 +end +begin /usr/share/fonts/truetype/ttf-dejavu/DejaVuSerifCondensed-Bold.ttf + Family = DejaVu-Serif + FontName = DejaVu-Serif-Bold + Encoding = Unicode + Location = English + Charset = ISO8859-1 ISO8859-2 ISO8859-9 ISO8859-15 ISO10646-1 + UniCharset = ISO8859-1 ISO8859-2 ISO8859-9 ISO8859-15 + GeneralFamily = Roman + Weight = Bold + Width = Variable + Shape = Serif Upright + Foundry = Bitstream + Priority = 20 +end +begin /usr/share/fonts/truetype/ttf-dejavu/DejaVuSerifCondensed-Oblique.ttf + Family = DejaVu-Serif + FontName = DejaVu-Serif + Encoding = Unicode + Location = English + Charset = ISO8859-1 ISO8859-2 ISO8859-9 ISO8859-15 ISO10646-1 + UniCharset = ISO8859-1 ISO8859-2 ISO8859-9 ISO8859-15 + GeneralFamily = Roman + Weight = Medium + Width = Variable + Shape = Serif Oblique + Foundry = Bitstream + Priority = 20 +end +begin /usr/share/fonts/truetype/ttf-dejavu/DejaVuSerifCondensed-BoldOblique.ttf + Family = DejaVu-Serif + FontName = DejaVu-Serif-Bold + Encoding = Unicode + Location = English + Charset = ISO8859-1 ISO8859-2 ISO8859-9 ISO8859-15 ISO10646-1 + UniCharset = ISO8859-1 ISO8859-2 ISO8859-9 ISO8859-15 + GeneralFamily = Roman + Weight = Bold + Width = Variable + Shape = Serif Oblique + Foundry = Bitstream + Priority = 20 +end + begin /usr/share
Bug#346150: fixed in upstream
It has been fixed in 6.4 release. Hopefully it will get packaged soon, as this bug really sucks. -- Jindrich Makovicka -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Bug#340001: [patch] please enable the tyuetype font hinting
Package: xserver-xorg Version: 6.8.2.dfsg.1-10 since the migration from xfree86 to x.org, truetype fonts rendered by X in legacy applications using Xaw,Xt or Gtk1 look like crap. could you please enable the truetype hinting so the non-antialiased fonts can look acceptable again? attached patch solves the issue. -- Jindrich Makovicka --- xorg-x11-6.8.2.dfsg.1.orig/xc/lib/font/FreeType/module/ftoption.h +++ xorg-x11-6.8.2.dfsg.1/xc/lib/font/FreeType/module/ftoption.h @@ -436,7 +436,7 @@ /* Do not #undef this macro here, since the build system might */ /* define it for certain configurations only. */ /* */ -/* #define TT_CONFIG_OPTION_BYTECODE_INTERPRETER */ +#define TT_CONFIG_OPTION_BYTECODE_INTERPRETER /*/
Bug#339672: [patch] fix menu display in Java/Swing applications
Package: gtk2-engines-clearlooks Version: 0.6.2-1 When GTK look is used in a Swing application, Java tries to emulate the current theme. However, it doesn't know anything about clearlooks engine and ignores it, only using the color scheme. This means that the selected menu item is invisible, due to a missing PRELIGHT value. The attached patch changes the background color of the selected menu item to a correct one. This change does not affect the look of original GTK applications at all, as PRELIGHT appears to be unused by the engine. Tested with libgtk 2.6.10-2 java 1.5.0-update5. The Swing bug can be seen eg. in JAlbum (http://jalbum.net). The patch is for basic variant of Clearlooks, the others would need a similar modification. -- Jindrich Makovicka --- gtkrc.orig 2005-06-30 12:47:24.0 +0200 +++ gtkrc 2005-09-08 19:55:43.157659021 +0200 @@ -92,6 +92,7 @@ ythickness = 3 fg[PRELIGHT] = #ff text[PRELIGHT] = #ff + bg[PRELIGHT] = #7c99ad } style clearlooks-menu-itembar = clearlooks-default
Bug#304690: fixed in module-init-tools 3.2-pre8-1
What is the purpose of keeping such diversion when there is nothing to divert anymore? There is only one pair of k(all)?syms, the one in modutils. -- Jindrich Makovicka -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Bug#320458: solution
ok, you finally convinced me to rebuild xserver-xorg from source, so i can now confirm that enabling the bytecode interpreter in xc/lib/font/FreeType/module/ftoption.h fixes the issue as expected. -- Jindrich Makovicka -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Bug#315460: jed: Please build against libslang2
Pre-version 0.99.17.111 builds with Slang2. The Debian patch still applies, except UID fixes, which have been included in the upstream. Complete tarball is at http://kmlinux.fjfi.cvut.cz/~makovick/jed_0.99.pre17.111.tar.gz if it helps. Regards, -- Jindrich Makovicka -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Bug#320458: xserver-xorg: gtk1.2 apps have screwed up fonts after migration
I guess the reason is that xorg's internal freetype2 library has the bytecode interpreter disabled. There are patent issues around this, but I hope it can be enabled by removing the comments, as stand-alone freetype2 has it enabled too. The file in question is xc/extras/freetype2/include/freetype/config/ftoption.h , line /* #define TT_CONFIG_OPTION_BYTECODE_INTERPRETER */ -- Jindrich Makovicka -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Bug#299662: [patch] fix log file corruption due to races
Package: privoxy Version: 3.0.3-3 As stated in the FIXME coments in privoxy source, the log writing function is prone to race conditions. As the log output buffer is static, two threads fighting over it can cause weird things being written to the log, maybe even a buffer overflow attack is possible. The attached patch adds two mutexes, guarding the log write and log initialization, and seems to fix the problem - without the patch, I was getting bogus lines in privoxy logs about five times a day. I also changed the format string for thread id to unsigned hex, as signed int was completely inappropriate for my system (looked like it's related to some memory location, rather than process id). I am running Debian Sid on 32bit x86. -- Jindrich Makovicka diff -ur vanilla/privoxy-3.0.3/errlog.c privoxy-3.0.3/errlog.c --- vanilla/privoxy-3.0.3/errlog.c 2003-03-07 04:41:04.0 +0100 +++ privoxy-3.0.3/errlog.c 2005-03-15 13:35:20.0 +0100 @@ -292,6 +292,30 @@ static char *os2_socket_strerr(int errcode, char *tmp_buf); #endif +#ifdef FEATURE_PTHREAD +static inline void lock_logfile() +{ +pthread_mutex_lock(log_mutex); +} +static inline void unlock_logfile() +{ +pthread_mutex_unlock(log_mutex); +} +static inline void lock_loginit() +{ +pthread_mutex_lock(log_init_mutex); +} +static inline void unlock_loginit() +{ +pthread_mutex_unlock(log_init_mutex); +} +#else /* ! FEATURE_PTHREAD */ +static inline void lock_logfile() {} +static inline void unlock_logfile() {} +static inline void lock_loginit() {} +static inline void unlock_loginit() {} +#endif + /* * * Function: fatal_error @@ -346,7 +370,7 @@ { FILE *fp; - /* FIXME RACE HAZARD: should start critical section error_log_use here */ + lock_loginit(); /* set the logging detail level */ debug = debuglevel | LOG_LEVEL_MINIMUM; @@ -354,9 +378,13 @@ if ((logfp != NULL) (logfp != stderr)) { log_error(LOG_LEVEL_INFO, (Re-)Open logfile %s, logfname ? logfname : none); + lock_logfile(); fclose(logfp); + } else { + lock_logfile(); } logfp = stderr; + unlock_logfile(); /* set the designated log file */ if( logfname ) @@ -369,7 +397,9 @@ /* set logging to be completely unbuffered */ setbuf(fp, NULL); + lock_logfile(); logfp = fp; + unlock_logfile(); } log_error(LOG_LEVEL_INFO, Privoxy version VERSION); @@ -378,7 +408,7 @@ log_error(LOG_LEVEL_INFO, Program name: %s, prog_name); } - /* FIXME RACE HAZARD: should end critical section error_log_use here */ + unlock_loginit(); } /* init_error_log */ @@ -428,6 +458,9 @@ return; } + /* protect the whole function because of the static buffer (outbuf) */ + lock_logfile(); + /* FIXME get current thread id */ #ifdef FEATURE_PTHREAD this_thread = (long)pthread_self(); @@ -480,41 +513,41 @@ switch (loglevel) { case LOG_LEVEL_ERROR: - outc = sprintf(outbuf, Privoxy(%05ld) Error: , this_thread); + outc = sprintf(outbuf, Privoxy(%08lx) Error: , this_thread); break; case LOG_LEVEL_FATAL: - outc = sprintf(outbuf, Privoxy(%05ld) Fatal error: , this_thread); + outc = sprintf(outbuf, Privoxy(%08lx) Fatal error: , this_thread); break; case LOG_LEVEL_GPC: - outc = sprintf(outbuf, Privoxy(%05ld) Request: , this_thread); + outc = sprintf(outbuf, Privoxy(%08lx) Request: , this_thread); break; case LOG_LEVEL_CONNECT: - outc = sprintf(outbuf, Privoxy(%05ld) Connect: , this_thread); + outc = sprintf(outbuf, Privoxy(%08lx) Connect: , this_thread); break; case LOG_LEVEL_LOG: - outc = sprintf(outbuf, Privoxy(%05ld) Writing: , this_thread); + outc = sprintf(outbuf, Privoxy(%08lx) Writing: , this_thread); break; case LOG_LEVEL_HEADER: - outc = sprintf(outbuf, Privoxy(%05ld) Header: , this_thread); + outc = sprintf(outbuf, Privoxy(%08lx) Header: , this_thread); break; case LOG_LEVEL_INFO: - outc = sprintf(outbuf, Privoxy(%05ld) Info: , this_thread); + outc = sprintf(outbuf, Privoxy(%08lx) Info: , this_thread); break; case LOG_LEVEL_RE_FILTER: - outc = sprintf(outbuf, Privoxy(%05ld) Re-Filter: , this_thread); + outc = sprintf(outbuf, Privoxy(%08lx) Re-Filter: , this_thread); break; #ifdef FEATURE_FORCE_LOAD case LOG_LEVEL_FORCE: - outc = sprintf(outbuf, Privoxy(%05ld) Force: , this_thread); + outc = sprintf(outbuf, Privoxy(%08lx) Force: , this_thread); break; #endif /* def FEATURE_FORCE_LOAD */ #ifdef FEATURE_FAST_REDIRECTS case LOG_LEVEL_REDIRECTS: - outc = sprintf(outbuf, Privoxy(%05ld) Redirect: , this_thread); + outc = sprintf(outbuf, Privoxy(%08lx
Bug#293302: [patch] use MMX optimized effects on x86
Package: xmame Version: 0.90-1 Current XMAME does not use optimized effect routines on x86 processors, which makes games running with -ef N option _very_ slow (mostly unplayable). The attached patch just adds EFFECT_MMX_ASM=1 to the build options for x86. As MAME uses runtime CPU detection, it should still run even on non-MMX x86 machines, according to the documentation. -- Jindrich Makovicka --- rules.orig 2005-02-02 11:59:58.0 +0100 +++ rules 2005-02-02 12:01:11.0 +0100 @@ -62,6 +62,7 @@ ifeq ($(DEB_HOST_ARCH),i386) DEB_OPTFLAGS=-02 -Wall -Wno-unused -march=i586 -mcpu=i686 -fomit-frame-pointer -funroll-loops -ffast-math -pipe DEB_COMMON_OPTS += X86_MIPS3_DRC=1 +DEB_COMMON_OPTS += EFFECT_MMX_ASM=1 else DEB_OPTFLAGS=-02 -Wall -Wno-unused -fomit-frame-pointer -funroll-loops -ffast-math -fsigned-char -pipe -mlongcall endif