Bug#891670: pymssql FTBFS with freetds-dev 1.00.82-2
I've been busy, sorry - feel free to team upload/NMU it but I'm hoping to get to it this weekend. -- Geoffrey Thomas (via mobile) geo...@ldpreload.com > On Jan 28, 2019, at 08:56, Raphael Hertzog wrote: > > Hi, > >> On Sun, 24 Jun 2018, Geoffrey Thomas wrote: >> Upstream is being slow to put out a new release, there's some blocker >> involving the new freetds. I asked if that was resolved yet: >> >> https://github.com/pymssql/pymssql/issues/528 >> >> At some point (probably in a month or two, honestly...) I'll cherry-pick the >> relevant patches if there's no release, but I'd prefer to see upstream be >> confident about the new freetds version. > > There's a new upstream release available. We need this package > updated for Python 3.7 for Kali so we will look into this if nobody > else is quicker. > > Cheers, > -- > Raphaël Hertzog ◈ Debian Developer > > Support Debian LTS: https://www.freexian.com/services/debian-lts.html > Learn to master Debian: https://debian-handbook.info/get/
Bug#891670: pymssql FTBFS with freetds-dev 1.00.82-2
Upstream is being slow to put out a new release, there's some blocker involving the new freetds. I asked if that was resolved yet: https://github.com/pymssql/pymssql/issues/528 At some point (probably in a month or two, honestly...) I'll cherry-pick the relevant patches if there's no release, but I'd prefer to see upstream be confident about the new freetds version. -- Geoffrey Thomas https://ldpreload.com geo...@ldpreload.com
Bug#818318: git security updates for wheezy-backports?
Hi git maintainers, I believe the version of git in wheezy-backports is affected by last week's security issues in #818318 (CVE-2016-2315 and CVE-2016-2324), as well as by CVE-2015-7545, since both of those were applied to the versions in wheezy and jessie. Are you uploading patched versions to backports? Would it be helpful for me to prepare and test an upload to backports? (I'd need sponsorship, since I'm not a DD and also don't have a valid PGP key currently.) Thanks, -- Geoffrey Thomas geo...@hudson-trading.com
Bug#725040: Intention to use NMU to fix the bug
On Tue, 15 Oct 2013, Sergei Golovan wrote: Hi! I've bumped the severity of this bug to serious to ensure Tcl/Tk 8.4 will not go to jessie when it'll become stable. I'm planning to use NMU to fix this bug if there's no objection for that. Hi Sergei, The patch you posted looks good. Please go ahead and NMU -- I don't expect to have time to look at this package in detail for a couple of weeks. Is the USE_INTERP_RESULT something that ought to be fixed in the project's codebase? If so, it would be totally awesome if you filed a bug either here or on the SourceForge project with a link or something to the issue. -- Geoffrey Thomas http://ldpreload.com geo...@ldpreload.com -- To UNSUBSCRIBE, email to debian-bugs-rc-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org
Bug#695361: man-db: Please revert workaround for less 456
On Mon, 6 May 2013, Colin Watson wrote: Thanks for the patch. I'd like to review and upload this myself, but it's a public holiday today so I don't know whether I'll quite get to it before your deadline. I will definitely get to it by tomorrow, if you can wait until then. I'm perfectly happy to wait -- it'll take a bit to get to testing anyway. Speaking of migrations, I'm thinking through how to avoid the same problem (apt threatening to remove less) in saucy. Since man-db is Ubuntu-patched, it won't get automatically imported, so I guess the answer is just to hold off on merging (slash syncing, the patch is just M-A: foreign) until after less gets synced into Ubuntu. The other option would be to remove the Breaks entirely and decide it's not that important. -- Geoffrey Thomas http://ldpreload.com geo...@ldpreload.com -- To UNSUBSCRIBE, email to debian-bugs-rc-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org
Bug#695361: less: buggy backslash handling in prompt string: \ needs to be doubled
On Sun, 5 May 2013, Sven Joachim wrote:
On 2013-03-22 11:01 +0100, Vincent Lefevre wrote:
On 2013-03-22 22:50:40 +1300, Jan Larres wrote:
version 457 of less, released in December, reverts to the old parsing
behaviour and makes the new one available as an option instead. So it
would probably be a better idea to upgrade to that version instead.
I agree. And what's important is that compatible versions of less
and man-db are installed at the same time.
Unfortunately that's currently impossible in Jessie because the version
of man-db there declares a Breaks: less ( 456), and less cannot
transition to testing because of this bug. So it would be good to fix
it or downgrade the severity, since not being able to install both
man-db and less sucks.
Since upstream less has reverted this behavior, it seems to me that the
right approach is to revert man-db's workaround, mark it as breaking less
456 only (since 456 and = 457 are then both okay), and upload the new
less upstream release as closing this bug.
Attached is a debdiff for the change to man-db needed to implement this.
Since you can't have a Breaks field for a range bounded on both sides,
I've just marked it as breaking less 456-1 and 456-1ubuntu1, the only
known packaged versions. I've tested that building this patch works the
way that you'd expect: apt neither installs the new version of less from
unstable, nor attempts to remove it. I've also test-built the new less
upstream release (458; 457 is no longer available) with no other changes
to packaging, and it works fine and `man apt.conf` displays the right
thing.
If the other folks on this bug report think this looks sane, I'll clone
this bug and assign it to man-db.
In addition to uploading the new man-db packaging and the new less
upstream, man-db upstream r1443 should be reverted.
--
Geoffrey Thomas
http://ldpreload.com
geo...@ldpreload.comdiff -Nru man-db-2.6.3/debian/changelog man-db-2.6.3/debian/changelog
--- man-db-2.6.3/debian/changelog 2012-12-16 04:18:24.0 -0800
+++ man-db-2.6.3/debian/changelog 2013-05-05 16:20:30.0 -0700
@@ -1,3 +1,10 @@
+man-db (2.6.3-3geofft1) unstable; urgency=low
+
+ * The incompatible change from less 456 has been reverted, so revert
+our patch and instead Break that version of less.
+
+ -- Geoffrey Thomas geo...@ldpreload.com Sun, 05 May 2013 16:18:17 -0700
+
man-db (2.6.3-3) unstable; urgency=low
* Support parallel builds.
diff -Nru man-db-2.6.3/debian/control man-db-2.6.3/debian/control
--- man-db-2.6.3/debian/control 2012-12-16 04:17:47.0 -0800
+++ man-db-2.6.3/debian/control 2013-05-05 17:11:58.0 -0700
@@ -13,7 +13,7 @@
Suggests: groff, less, www-browser
Provides: man, man-browser
Conflicts: man, suidmanager ( 0.50)
-Breaks: less ( 456)
+Breaks: less (= 456-1), less (= 456-1ubuntu1)
Replaces: man, nlsutils, manpages-de ( 0.5-4)
Multi-Arch: foreign
Description: on-line manual pager
diff -Nru man-db-2.6.3/debian/patches/less-incompatibility.patch
man-db-2.6.3/debian/patches/less-incompatibility.patch
--- man-db-2.6.3/debian/patches/less-incompatibility.patch 2012-12-16
04:05:23.0 -0800
+++ man-db-2.6.3/debian/patches/less-incompatibility.patch 1969-12-31
16:00:00.0 -0800
@@ -1,52 +0,0 @@
-Description: Handle incompatible change to option string escaping in less 456
-Author: Colin Watson cjwat...@debian.org
-Origin: backport,
http://bazaar.launchpad.net/~cjwatson/man-db/trunk/revision/1443
-Bug-Debian: http://bugs.debian.org/695459
-Forwarded: not-needed
-Last-Update: 2012-12-16
-
-Index: b/src/man.c
-===
a/src/man.c
-+++ b/src/man.c
-@@ -814,17 +814,35 @@
- static char *escaped_string;
- char *ptr;
-
-- /* 2*strlen will always be long enough to hold the escaped string */
-+ /* 4*strlen will always be long enough to hold the escaped string */
- ptr = escaped_string = xrealloc (escaped_string,
-- 2 * strlen (string) + 1);
--
-+ 4 * strlen (string) + 1);
-+
- while (*string) {
-+ /* less 456 requires dollar and backslash to be escaped in
-+ * the option string; this means that we need two
-+ * backslashes to effectively escape characters special in
-+ * prompt strings, and that displaying a backslash requires
-+ * two levels of escaping. Note that this appears to be an
-+ * incompatible change, so this will overescape for earlier
-+ * versions of less.
-+ */
- if (*string == '?' ||
- *string == ':' ||
- *string == '.' ||
-- *string == '%' ||
-- *string == '\\')
-+ *string == '%') {
-+ /* Special only in prompt strings
Bug#695361: man-db: Please revert workaround for less 456
clone 695361 -1
reassign -1 man-db
retitle -1 man-db: Please revert workaround for less 456
severity -1 serious
# Justification: 695361 is serious
block 695361 by -1
tags -1 patch
thanks
Hi Colin,
In #695459 you added a workaround for the new backslash-escaping behavior
in less 456-1. Upstream less 457 has since reverted this behavior unless
opted in, so I think the sanest thing to do is to remove this workaround
entirely from man-db, declare a Breaks on just version 456, and put less
= 457 into testing.
Attached is a debdiff for a local test build to do this -- see also my
comment in the less bug, #695361, reproduced below. Please incorporate and
upload this change, and also revert the upstream commit of the workaround;
then less 457 or above can be safely uploaded to close #695361.
I note that the maintainers of both man-db and less are on the
LowThresholdNmu list, so since this is currently blocking sane
dist-upgrades to testing (I'm caring because I tried to upgrade my laptop
and apt is threatening to remove less), I'm happy to NMU either or both
these packages, if that's easier for you. I'll wait for comments until
tomorrow evening my time, about 24h, before doing so.
--
Geoffrey Thomas
http://ldpreload.com
geo...@ldpreload.com
On Mon, 6 May 2013, Vincent Lefevre wrote:
On 2013-05-05 17:55:44 -0700, Geoffrey Thomas wrote:
Since upstream less has reverted this behavior, it seems to me that the
right approach is to revert man-db's workaround, mark it as breaking less
456 only (since 456 and = 457 are then both okay), and upload the new
less upstream release as closing this bug.
Attached is a debdiff for the change to man-db needed to implement this.
Since you can't have a Breaks field for a range bounded on both sides, I've
just marked it as breaking less 456-1 and 456-1ubuntu1, the only known
packaged versions. I've tested that building this patch works the way that
you'd expect: apt neither installs the new version of less from unstable,
nor attempts to remove it. I've also test-built the new less upstream
release (458; 457 is no longer available) with no other changes to
packaging, and it works fine and `man apt.conf` displays the right thing.
I haven't tested, but I think this is the way to do.
--
Vincent Lefèvre vinc...@vinc17.net - Web: http://www.vinc17.net/
100% accessible validated (X)HTML - Blog: http://www.vinc17.net/blog/
Work: CR INRIA - computer arithmetic / AriC project (LIP, ENS-Lyon)
diff -Nru man-db-2.6.3/debian/changelog man-db-2.6.3/debian/changelog
--- man-db-2.6.3/debian/changelog 2012-12-16 04:18:24.0 -0800
+++ man-db-2.6.3/debian/changelog 2013-05-05 16:20:30.0 -0700
@@ -1,3 +1,10 @@
+man-db (2.6.3-3geofft1) unstable; urgency=low
+
+ * The incompatible change from less 456 has been reverted, so revert
+our patch and instead Break that version of less.
+
+ -- Geoffrey Thomas geo...@ldpreload.com Sun, 05 May 2013 16:18:17 -0700
+
man-db (2.6.3-3) unstable; urgency=low
* Support parallel builds.
diff -Nru man-db-2.6.3/debian/control man-db-2.6.3/debian/control
--- man-db-2.6.3/debian/control 2012-12-16 04:17:47.0 -0800
+++ man-db-2.6.3/debian/control 2013-05-05 17:11:58.0 -0700
@@ -13,7 +13,7 @@
Suggests: groff, less, www-browser
Provides: man, man-browser
Conflicts: man, suidmanager ( 0.50)
-Breaks: less ( 456)
+Breaks: less (= 456-1), less (= 456-1ubuntu1)
Replaces: man, nlsutils, manpages-de ( 0.5-4)
Multi-Arch: foreign
Description: on-line manual pager
diff -Nru man-db-2.6.3/debian/patches/less-incompatibility.patch
man-db-2.6.3/debian/patches/less-incompatibility.patch
--- man-db-2.6.3/debian/patches/less-incompatibility.patch 2012-12-16
04:05:23.0 -0800
+++ man-db-2.6.3/debian/patches/less-incompatibility.patch 1969-12-31
16:00:00.0 -0800
@@ -1,52 +0,0 @@
-Description: Handle incompatible change to option string escaping in less 456
-Author: Colin Watson cjwat...@debian.org
-Origin: backport,
http://bazaar.launchpad.net/~cjwatson/man-db/trunk/revision/1443
-Bug-Debian: http://bugs.debian.org/695459
-Forwarded: not-needed
-Last-Update: 2012-12-16
-
-Index: b/src/man.c
-===
a/src/man.c
-+++ b/src/man.c
-@@ -814,17 +814,35 @@
- static char *escaped_string;
- char *ptr;
-
-- /* 2*strlen will always be long enough to hold the escaped string */
-+ /* 4*strlen will always be long enough to hold the escaped string */
- ptr = escaped_string = xrealloc (escaped_string,
-- 2 * strlen (string) + 1);
--
-+ 4 * strlen (string) + 1);
-+
- while (*string) {
-+ /* less 456 requires dollar and backslash to be escaped in
-+ * the option string; this means that we need two
-+ * backslashes to effectively escape characters special
Bug#688713: timidity-daemon: mishandles conffiles
On Thu, 11 Oct 2012, Sébastien Villemot wrote: Thanks for your reply. I have rescheduled the upload to DELAYED/5, so that you can make an upload of your own this weekend. I failed to do this because I got stuck figuring out what was up with the large number of Ubuntu bugs. I think we have more bugs in the maintainer scripts (many of them are undiagnosable because we ignore the return value from adduser)... I'll hopefully get time to do an upload soon, but in the meantime, thanks for the NMU, and Guo, thanks for the patch. -- Geoffrey Thomas http://ldpreload.com geo...@ldpreload.com
Bug#688713: timidity-daemon: mishandles conffiles
On Thu, 11 Oct 2012, Sébastien Villemot wrote: I have uploaded to DELAYED/2 a NMU of timidity versioned 2.13.2-40.1 which fixes that bug. The debdiff is attached. Don't hesitate to tell me if I should delay the upload longer. Thanks, but would you mind delaying until early next week? There are a couple of wrong patches in the current source package (see #649274), and I was hoping to ask for a single freeze exception to deal with that issue too and a couple of Ubuntu ones. I know I've been remiss about doing an upload, but if it's already past time that an NMU is reasonable, I will make a point of attempting to deal with it this weekend -- and if I don't, feel free to let the NMU proceed. -- Geoffrey Thomas http://ldpreload.com geo...@ldpreload.com
Bug#689070: Please take upstream D-Bus patches for CVE-2012-3524
Package: dbus Severity: serious Justification: local privilege escalation Tags: security Hi, CVE-2012-3524 is about setuid binaries linking libdbus being easily trickable to do bad things via a malicious PATH (for finding dbus-launch), or through a DBUS_* address variable using the unixexec address type. Initially the D-Bus developers thought that this should be fixed on the application side (hence the comment in the security-tracker), but decided that it would be better to have a defense-in-depth approach, and change _dbus_getenv to not succeed if the current program is setuid or similar, since that's faster than patching every relevant program. There's a patch in the D-Bus 1.6.6 release that implements this. Many other distros, including RHEL/Fedora, SUSE, and Ubuntu have taken this patch already. There are some other hardening things in the 1.6.6 release that broke gnome-keyring, prompting a 1.6.8 release a few hours later to revert those; you should either take 1.6.8, or just backport the four patches that weren't reverted in 1.6.8: http://cgit.freedesktop.org/dbus/dbus/commit/?id=23fe78ceefb6cefcd58a49c77d1154b68478c8d2 http://cgit.freedesktop.org/dbus/dbus/commit/?id=4b351918b9f70eaedbdb3ab39208bc1f131efae0 http://cgit.freedesktop.org/dbus/dbus/commit/?id=57ae3670508bbf4ec57049de47c9cae727a64802 http://cgit.freedesktop.org/dbus/dbus/commit/?id=f68dbdc3e6f895012ce33939fb524accf31bcca5 I think these are all easily backportable, but I'm happy to supply a debdiff if that'd make it easier for you. More discussion of the issue can be found at https://bugs.freedesktop.org/show_bug.cgi?id=52202 https://bugzilla.novell.com/show_bug.cgi?id=697105 https://bugzilla.redhat.com/show_bug.cgi?id=847402 http://seclists.org/oss-sec/2012/q3/29 -- Geoffrey Thomas gtho...@mokafive.com -- To UNSUBSCRIBE, email to debian-bugs-rc-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org
Bug#649344: timidity: embedded fork of libmikmod needs to be dealt with
Package: timidity Version: 2.13.2-39 Severity: serious Justification: policy 4.13 From libunimod/AUTHORS: Code in this directory was modified by Paolo Bonzini bonz...@gnu.org starting from libmikmod 3.1.8 libmikmod was cleaned up, removing code that was needed by libmikmod's portability (MD_) layer and data that were needed by the player. The player itself now forms the core of TiMidity++'s mod.c file but is detached from libunimod which is nothing more a generic module-UNI converter. In addition, libmikmod's own portable file I/O routines were changed to rely on libarc's URL objects. We do need to deal with this, but upstream mikmod was last merged in in 2000 and I'm a bit worried about what exactly was cleaned up, so I'm not really looking forward to dealing with this before fixing the more pressing bugs like the FTBFS. So I'm filing an RC bug and uploading with a Lintian override, and will take a closer look sometime later. This bug has been presumably present since forever, but the Lintian check is new as of this February. I'm also a bit skeptical of this libarc directory (what it does, whether it's _actually_ free, etc.), incidentally. -- Geoffrey Thomas http://ldpreload.com geo...@ldpreload.com -- To UNSUBSCRIBE, email to debian-bugs-rc-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org
Bug#633501: libjinput-jni: contains undefined symbol EVIOCGUSAGE
tags 633501 + patch
forwarded 633501 http://java.net/jira/browse/JINPUT-44
thanks
Hi all,
I've reported this bug upstream at http://java.net/jira/browse/JINPUT-44 ,
and submitted a patch (attached, and also at
http://ldpreload.com/p/jinput-remove-getDeviceUsageBits.patch ) to
eliminate all code related to this nonexistent ioctl. The package builds
cleanly with this patch applied via debian/patches, and I've verified that
the resulting .so no longer references an EVIOCGUSAGE symbol, but I don't
have enough of an idea of how this package works to properly test that it
addresses the reported crash. (At the very least, I can't reproduce the
crash simply by installing this package, or adding import
net.java.games.input.* to a Hello World class.)
Java team, if you would like me to prepare a debdiff or NMU including this
patch, let me know (and let me know how to test it), and I'll be happy to
do so.
--
Geoffrey Thomas
http://ldpreload.com
geo...@ldpreload.comIndex: trunk/plugins/linux/src/java/net/java/games/input/LinuxEventDevice.java
===
--- trunk/plugins/linux/src/java/net/java/games/input/LinuxEventDevice.java (revision 247)
+++ trunk/plugins/linux/src/java/net/java/games/input/LinuxEventDevice.java (working copy)
@@ -94,28 +94,6 @@
}
private final Controller.Type guessType() throws IOException {
- Controller.Type type_from_usages = guessTypeFromUsages();
- if (type_from_usages == Controller.Type.UNKNOWN)
- return guessTypeFromComponents();
- else
- return type_from_usages;
- }
-
- private final Controller.Type guessTypeFromUsages() throws IOException {
- byte[] usage_bits = getDeviceUsageBits();
- if (isBitSet(usage_bits, NativeDefinitions.USAGE_MOUSE))
- return Controller.Type.MOUSE;
- else if (isBitSet(usage_bits, NativeDefinitions.USAGE_KEYBOARD))
- return Controller.Type.KEYBOARD;
- else if (isBitSet(usage_bits, NativeDefinitions.USAGE_GAMEPAD))
- return Controller.Type.GAMEPAD;
- else if (isBitSet(usage_bits, NativeDefinitions.USAGE_JOYSTICK))
- return Controller.Type.STICK;
- else
- return Controller.Type.UNKNOWN;
- }
-
- private final Controller.Type guessTypeFromComponents() throws IOException {
List components = getComponents();
if (components.size() == 0)
return Controller.Type.UNKNOWN;
@@ -342,15 +320,6 @@
}
private final static native void nGetBits(long fd, int ev_type, byte[] evtype_bits) throws IOException;
- private final byte[] getDeviceUsageBits() throws IOException {
- byte[] bits = new byte[NativeDefinitions.USAGE_MAX/8 + 1];
- if (getVersion() = 0x010001) {
- nGetDeviceUsageBits(fd, bits);
- }
- return bits;
- }
- private final static native void nGetDeviceUsageBits(long fd, byte[] type_bits) throws IOException;
-
public final synchronized void pollKeyStates() throws IOException {
nGetKeyStates(fd, key_states);
}
Index: trunk/plugins/linux/src/native/net_java_games_input_LinuxEventDevice.c
===
--- trunk/plugins/linux/src/native/net_java_games_input_LinuxEventDevice.c (revision 247)
+++ trunk/plugins/linux/src/native/net_java_games_input_LinuxEventDevice.c (working copy)
@@ -112,20 +112,6 @@
return num_effects;
}
-JNIEXPORT void JNICALL Java_net_java_games_input_LinuxEventDevice_nGetDeviceUsageBits(JNIEnv *env, jclass unused, jlong fd_address, jbyteArray usages_array) {
-#if EV_VERSION = 0x010001
- int fd = (int)fd_address;
- jsize len = (*env)-GetArrayLength(env, usages_array);
- jbyte *usages = (*env)-GetByteArrayElements(env, usages_array, NULL);
- if (usages == NULL)
- return;
- int res = ioctl(fd, EVIOCGUSAGE(len), usages);
- (*env)-ReleaseByteArrayElements(env, usages_array, usages, 0);
- if (res == -1)
- throwIOException(env, Failed to get device usages (%d)\n, errno);
-#endif
-}
-
JNIEXPORT void JNICALL Java_net_java_games_input_LinuxEventDevice_nGetBits(JNIEnv *env, jclass unused, jlong fd_address, jint evtype, jbyteArray bits_array) {
int fd = (int)fd_address;
jsize len = (*env)-GetArrayLength(env, bits_array);
Bug#638802: google-authenticator: FTBFS because of build-depends on virtual package
Package: google-authenticator Version: 20110413.68230188bdc7-1 Severity: serious Justification: FTBFS google-authenticator build-depends on libpam-dev, which is a virtual package (provided by libpam0g-dev), which makes both `aptitude build-dep` and `dpkg-checkbuilddeps` sad. (I suppose that you're using sbuild with the aptitude resolver, it may work, since the resolver builds an equivs-style package with the build dependencies.) When building the package with debuild, I got this: dpkg-checkbuilddeps: Unmet build dependencies: libpam-dev dpkg-buildpackage: warning: Build dependencies/conflicts unsatisfied; aborting. dpkg-buildpackage: warning: (Use -d flag to override.) debuild: fatal error at line 1340: dpkg-buildpackage -rfakeroot -D -us -uc failed See also: geofft@leveret:/tmp$ sudo aptitude build-dep google-authenticator [sudo] password for geofft: Unable to satisfy the build-depends: Build-Depends: libpam-dev. Unable to satisfy the build-depends: Build-Depends: libpam-dev. No packages will be installed, upgraded, or removed. 0 packages upgraded, 0 newly installed, 0 to remove and 5 not upgraded. Need to get 0 B of archives. After unpacking 0 B will be used. -- Geoffrey Thomas geo...@mit.edu -- To UNSUBSCRIBE, email to debian-bugs-rc-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org
Bug#629666: patch for this bug
forcemerge 628729 629666 severity 628729 serious thanks This was previously reported by Evan Broder in bug #628799, with essentially the same patch. We don't need to worry about -ldl not existing since the only libc on Debian is glibc. (If this assumption is incorrect, then we'd need to be more clever, e.g., use autoconf.) Attached is a debdiff to incorporate Evan's patch and close all open FTBFSes against this package. -- Geoffrey Thomas geo...@mit.edu On Sun, 21 Aug 2011, Christine Spang wrote: tags 629666 + patch thanks The problem here is that libpam/Makefile is making an erroneous check for /usr/lib/libdl.so, and only if this file exists does it link to libdl, which contains the symbols that the build is erroring out because it can't find. Is there any reason *not* to unconditionally link to libdl? libdl.so is never in /usr/lib these days (perhaps because of multiarch support?). I have attached a patch that does exactly that. cheers, Christine diff -Nru google-authenticator-20110413.68230188bdc7/debian/changelog google-authenticator-20110413.68230188bdc7/debian/changelog --- google-authenticator-20110413.68230188bdc7/debian/changelog 2011-04-13 02:20:20.0 -0400 +++ google-authenticator-20110413.68230188bdc7/debian/changelog 2011-08-21 21:48:21.0 -0400 @@ -1,3 +1,13 @@ +google-authenticator (20110413.68230188bdc7-1.1) unstable; urgency=low + + * Non-maintainer upload. + * Change build dependencies to list a concrete package (Closes: #638802). + * Unconditionally link -ldl (since Debian always uses glibc), to fix a +FTBFS caused by libdl.so moving to multiarch paths (Closes: #628729, +#629666). Patch from Evan Broder e...@ebroder.net. + + -- Geoffrey Thomas geo...@mit.edu Sun, 21 Aug 2011 21:45:15 -0400 + google-authenticator (20110413.68230188bdc7-1) unstable; urgency=low * Upstream update. diff -Nru google-authenticator-20110413.68230188bdc7/debian/control google-authenticator-20110413.68230188bdc7/debian/control --- google-authenticator-20110413.68230188bdc7/debian/control 2011-02-25 15:11:45.0 -0500 +++ google-authenticator-20110413.68230188bdc7/debian/control 2011-08-21 21:31:43.0 -0400 @@ -2,7 +2,7 @@ Section: admin Priority: optional Maintainer: LENART Janos o...@debian.org -Build-Depends: debhelper (= 7.0.5), libpam-dev +Build-Depends: debhelper (= 7.0.5), libpam0g-dev Standards-Version: 3.9.1 Homepage: http://code.google.com/p/google-authenticator/ Vcs-hg: https://google-authenticator.googlecode.com/hg/ diff -Nru google-authenticator-20110413.68230188bdc7/debian/patches/make-multiarch-friendly google-authenticator-20110413.68230188bdc7/debian/patches/make-multiarch-friendly --- google-authenticator-20110413.68230188bdc7/debian/patches/make-multiarch-friendly 1969-12-31 19:00:00.0 -0500 +++ google-authenticator-20110413.68230188bdc7/debian/patches/make-multiarch-friendly 2011-08-21 21:45:02.0 -0400 @@ -0,0 +1,39 @@ +Description: Switch the Makefile to be multiarch-friendly + google-authenticator looks for /usr/lib/libdl.so to determine whether + or not to link -ldl, but since the multiarch transition process moved + libdl to a different location, that test no longer finds it, leading + to a FTBFS on multiarch-enabled systems. + . + Work around that by just always linking libdl, which we definitely + always want to do with glibc. +Author: Evan Broder e...@ebroder.net +Origin: vendor +Bug: http://code.google.com/p/google-authenticator/issues/detail?id=74 + +Index: ubuntu/libpam/Makefile +=== +--- ubuntu.orig/libpam/Makefile2011-05-31 11:40:31.878079519 -0700 ubuntu/libpam/Makefile 2011-05-31 11:52:43.952386500 -0700 +@@ -47,18 +47,17 @@ + pam_google_authenticator_unittest + + google-authenticator: google-authenticator.o base32.o hmac.o sha1.o +- $(CC) -g $(LDFLAGS) $(shell [ -f /usr/lib/libdl.so ] echo -ldl) \ +--o $@ $+ ++ $(CC) -g $(LDFLAGS) \ ++-o $@ $+ -ldl + + demo: demo.o pam_google_authenticator_demo.o base32.o hmac.o sha1.o + $(CC) -g $(LDFLAGS) -rdynamic \ +-$(shell [ -f /usr/lib/libdl.so ] echo -ldl) -o $@ $+ ++-o $@ $+ -ldl + + pam_google_authenticator_unittest: pam_google_authenticator_unittest.o \ +base32.o hmac.o sha1.o + $(CC) -g $(LDFLAGS) -rdynamic -lc \ +- $(shell [ -f /usr/lib/libdl.so ] echo -ldl) \ +- -o $@ $+ ++ -o $@ $+ -ldl + + pam_google_authenticator.so: base32.o hmac.o sha1.o + pam_google_authenticator_testing.so: base32.o hmac.o sha1.o diff -Nru google-authenticator-20110413.68230188bdc7/debian/patches/series google-authenticator-20110413.68230188bdc7/debian
