I have briefly tested a patch that largely reverts the 8.2 behaviour,
built on top of the two debian timezone patches.
I don't have a full understanding of how these functions interact, so
there might be corner cases not covered, but it is possible to get the
old behaviour back, especially in
Package: php
Version: 8.2.18-1~deb12u1
PHP isn't parsing the system time zone like it used to be, instead
defaulting to UTC.
Here is a test script:
The patch that is supposed to make this work is
debian/patches/0021-Use-system-timezone.patch, but it apparently no
longer works.
Others
Package: fonts-wine
Version: 8.0.1~repack-2
I just did a dist upgrade and got the change that moves the wine fonts
into the global /usr/share/fonts path where all programs use them. This
appears to be in response to issue #883973.
Please revert this change. These are not the real Microsoft
On 2021-05-05 8:56 p.m., Trent W. Buck wrote:
Debian 10 defaults to nftables, and iptables(8) is a backcompat wrapper:
bash5$ mmdebstrap --quiet buster /dev/null --include=iptables
--customize-hook='chroot $1 readlink -f /usr/sbin/iptables'
/usr/sbin/xtables-nft-multi
sshguard
Have tracked this bug down to an issue with sane-backends, as described
(and worked around) here:
https://gitlab.com/sane-project/backends/-/issues/271
This is not a bug in simple-scan.
Regards,
--Pat
Package: sshguard
Upon upgrading to buster, sshguard in all of my deployments has stopped
working.
I suspect this line in the Debian changelog:
* debian/sshguard.service, Use nft instead iptables.
There doesn't seem to be any obvious way to change this back to iptables.
It is unclear why
Package: acmetool
Version: 0.0.62-2
This package only does ACMEv1. However, beta version 0.2.1 was released
on October 16, 2019:
https://github.com/hlandau/acmetool/releases/tag/v0.2.1
This has ACMEv2 support and I find it works very well. It should be put
into unstable since the ACMEv1
On 2019-09-17 11:00 a.m., Patrick Matthäi wrote:
Could you test the patch below? It should restore the behaviour for
those IPs. then they are (again) listed as "serbia":
That patch did not help. I looked into it further, and the error is
triggered on the other "Unrecognized country code"
Package: geoip-bin
Version: 1.6.12-4
I was trying out the fix for another issue, generating a fresh
geoip-database package using the latest version of the tools (and lib
and -dev).
It fails on interpreting the new (unofficial?) XK country code for Kosovo:
# Building geoip v4 country
From MaxMind web site
(https://dev.maxmind.com/geoip/geoip2/whats-new-in-geoip2/):
"We now distinguish between several types of country data. The country
is the country where the IP address is located. The registered_country
is the country in which the IP is registered. These two may differ
Package: geoip-bin
Version: 1.6.12-3
I believe there is an issue with data conversion in the
geolite2-to-legacy-csv.sh.
Checking the IP 5.62.22.12, which is in Australia, against the generated
data, it comes back as UK.
The source data comes back as Australia:
On 14-01-07 08:29 PM, Andrew Bartlett wrote:
What are the relevant values of in-sync-ness for heimdal?
I would prefer it was exactly the version we have in-tree, but that's
now quite old.
So, it would seem that the present solution is that samba should be
updating the heimdal snapshot,
It is surprisingly difficult to get a stack trace, even with samba-dbg
and heimdal-dbg installed:
#0 0x in ?? ()
#1 0x7f58b4821ff5 in ?? ()
#2 0x7fff2c599618 in ?? ()
#3 0x7f58ba31f260 in ?? ()
#4 0x7f58ba8e57e0 in ?? ()
#5 0x7f58ba2c6280 in ?? ()
#6
Package: libnss-mysql-bg
Version: 1.5-3+b3
Severity: serious
Justification: Policy 4.3
Hello,
I have found that the Debian patch 04_shadow.diff to libnss-mysql-bg causes
the library to hang under completely typical usage.
I am using libnss-mysql-bg with chrooted SFTP as documented here:
Package: mdadm
Version: 3.0-2
dist-upgrading unstable to udev version 146-3 causes mdadm to uninstall,
since udev 'Break's mdadm 3.0-3, and the newest version is 3.0-2.
This needs to be resolved quickly, or there will be many more broken
computers. In my case, I didn't see mdadm was being
Package: gnome-games
Version: 1:2.26.3-1
When gnome-games is installed on a clean squeeze debootstrap, there is a
missing dependency on python-gnome2, which contains the gnomecanvas
Python module. I think some assumptions may have been made that it gets
pulled in via a Recommends.
Author: Pat Suwalski [EMAIL PROTECTED]
This patch fixes the QImage header breaking with -Werror on Lenny:
/usr/include/qt3/qimage.h: In member function
'bool QImageTextKeyLang::operator(const QImageTextKeyLang) const':
/usr/include/qt3/qimage.h:61: error: suggest parentheses around within
Package: update-inetd
Version: 4.27-0.3
Severity: critical
The latest version of update-inetd (4.27-0.3) breaks the ability of
debootstrap to install postfix like so:
debootstrap --include=postfix etch destination
I have verified that this works correctly with 4.27-0.2.
Postfix dies
I should have put more of the configure -x output, as I just noticed
that that is where update-inetd is actually executed from. Here is the
entire execution:
foobar:/# dpkg --configure postfix
Setting up postfix (2.3.6-1) ...
+ . /usr/share/debconf/confmodule
++ '[' '!' '' ']'
++
Tags: patch
Attached is a patch that works around the issue. Apparently, invoke-rc.d
leaves junk in the standard file descriptors (no newline?). Debconf is
then affected when it uses shell's read. By redirecting the output to
null, everything appears to work as before.
To give some feedback
Hello,
I tried the latest package and it does not help, unfortunately. The
compiler is still executed without the -fPIC option.
What *did* fix it is adding -fPIC to the CFLAGS in debian/rules.
Thomas, it would be very good to try to get this into Etch, as the
library is absolutely 100%
Package: libx11
Version: 2:1.0.3-1
Hello,
In the past, a -DLIBXCURSOR was required in debian/rules to set the
Xcursor library to libXcursor.so.1.
This define would break the build in Debian Sarge.
The define is no longer required, since the upstream source appears to
have changed it to
Michael Banck wrote:
This define would break the build in Debian Sarge.
FWIW, you can fix the build with a slightly different define, see what
e.g. Andres did at
http://mouth.squishy.cc/~dilinger/xorg-bpo/libx11_1.0.3-0bpo1.diff.gz
Good thing that horrible -D can go away, though.
Indeed,
This problem still exists in the mesa_6.5.1-0.1 package.
It's an easy fix, please apply it.
--
To UNSUBSCRIBE, email to [EMAIL PROTECTED]
with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Package: twm
Version: 1.0.1-4
The twm package defines CFLAGS in the debian/rules file:
CFLAGS += -DLIBXCURSOR=\\\libXcursor.so.1\\\
These CFLAGS are also in the libX11 rules file. They break when building
on Sarge, but that's not the issue.
Unlike libX11, LIBXCURSOR is not used
Package: mesa
Version: 6.4.2-1
The mesa package doesn't compile on Sarge because gcc-3.3.5 gets confused.
debian/configs defines both -ansi and -std=c99 in the
DEBIAN_COMMON_CFLAGS variable. DEBIAN_COMMON_CXXFLAGS also has -ansi.
From the gcc manpage:
The -ansi option is equivalent
This fix works. Thanks.
--Pat
--
To UNSUBSCRIBE, email to [EMAIL PROTECTED]
with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
RedHat's version of the GL patch solves this problem by adding the
version and not the path to the library, and modifying the function that
puts the library name together very slightly.
It's attached, but I'm not sure that it will show up in the bug. The
source is the Fedora SRPM.
--Pat
---
I have not confirmed this, but I believe this error is because of the
Debian 16_fix_opengl_check.dpatch.
Specifically:
-QLibrary gl(GL);
+QLibrary gl(/usr/lib/libGL.so.1);
This conflicts with the libGL in the LD_LIBRARY_PATH which is the ATI or
nVidia libGL.so.1.
A workaround is to
Indeed, recompiling libqt without the patch and adding xlibmesa-gl-dev
and libxmu-dev makes the OpenGL kscreensavers work with commercial
nVidia and ATI drivers.
--Pat
--
To UNSUBSCRIBE, email to [EMAIL PROTECTED]
with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
This security update really breaks the behaviour of sudo, especially
with regard to the DISPLAY variable used in a lot of projects.
Things like:
sudo xeyes
no longer work.
We have to tell our users to add:
Default: env_reset, env_keep=DISPLAY
to get their functionality back.
This patch definitely fixes the problem over here with pam 0.76.
Sam, I find your logic to confusing as well:
IT is wrong for it to call getpwnam because it cannot change the
password of an account unless it exists in some system that pam_unix.so
understands.
pam_unix understands NIS and that
32 matches
Mail list logo
