With latest metropolis 1.2 from https://github.com/matze/mtheme,
my presentation compiles again.
As it would be a lot of work to extract a minimal example from
the horribly complicated beamer code, I suggest we better close
this issue.
Sorry, Joachim
smime.p7s
Description: S/MIME
Package: texlive-fonts-extra
Version: 2023.20231007-2
Beamer theme metropolis uses font Fira.
Worked fine under several Debian releases up to Bookworm.
Broken in Trixie: Throws warnings about
Could not find Fira Sans fonts ...
Could not find Fira Mono fonts ...
then terminates with error
the Debian-patched CMake
will handle them? It will do so by inserting the aformentioned -D options?
Thanks again, Joachim
On 11/01/2023 10:48, Timo Röhling wrote:
Hi Joachim,
On Wed, 30 Nov 2022 17:22:08 +0100 Joachim Wuttke
wrote:
Why does python3-numpy under bookworm, since a few days
> How about this?
>
> #!/bin/sh
> ":" """ "
> if which python3.11 >/dev/null 2>&1; then
> exec python3.11 "$0" "$@"
> else
> echo "Please install python3.11 to use this tool" >&2
> fi
> exit 1
> """
> print("hey, I'm running under
Dear all:
Sandro wrote:
> Please compare the numpy popcon numbers (which we know are not
> detailed, but it's a start) with the handful of people having reported
> this issue. While maintaining numpy, my work caters towards the actual
> numpy users, not people who just want to not have 2 python
Reviewing #945824, I believe that even if we were eventually to overrule
the Python maintainers on this, (i) the impact to users in the meantime
is very low, and (ii) we are about to freeze, so it's unlikely we'd make
any decision applying to bookworm.
So I think we should leave this for the
, Joachim Wuttke
Dear Sandro:
>> Why does python3-numpy under bookworm, since a few days,
>> depend on python3.10 *and* on python3.11?
>
> because it provides compiled extensions for both 3.10. and 3.11.
Sorry, I don't get the logic.
To make good use of this package, we need either 3.10 *or* 3.11.
Why then
Package: python3-numpy
Version: 1:1.21.5-2
Dear Sandro, dear Matthias, dear all:
Why does python3-numpy under bookworm, since a few days, depend on python3.10
*and* on python3.11?
This breaks our software projects that combine C++ with Python.
The CMake command
find_package(Python3
Package: python3-matplotlib
Source: matplotlib (3.5.1-2)
Version: 3.5.1-2+b1
The dependency problems for Matplotlib with LaTex markup
are so notorious that they are already mentioned in the
Matplotlib tutorial [1]:
"On Ubuntu and Gentoo, the base texlive install does not
ship with the type1cm
Package: python3-matplotlib
Source: matplotlib (3.5.1-2)
Version: 3.5.1-2+b1
Run a plot script that contains
from matplotlib import rc
rc('text', usetex=True)
and makes use of LaTeX markup (some labels containing "$...$").
Execution will fail with a long traceback and an error message
solution or workaround?
Set Python_FIND_STRATEGY to LOCATION. Or set policy CMP0094 to NEW. Or require
CMake version >= 3.15.
With this, CMake finds Python 3.9 even if some 3.10 modules are present on the
system.
Amyway, I still think that a Debian package should be able to **support**
The broken code line
find_package(Python3 COMPONENTS Interpreter Development)
comes literally from the CMake docs,
https://cmake.org/cmake/help/v3.22/module/FindPython3.html.
So I reported the inconsistency also at
https://gitlab.kitware.com/cmake/cmake/-/issues/22983.
- Joachim
For the benefit of others who encounter the same problem:
To work around this bug, deinstall the Debian package
python3-numpy for good, and reinstall numpy using pip3.
Very disappointing though that this kind of inconsistencies
is allowed into Debian.
- Joachim
Package: python3-dev
Version: 3.9.7-1
The recent addition of python3.10:any to the dependencies
of python3-numpy under Debian testing (=bookworm) breaks
software that depends on python3-dev.
The numpy maintainer says this is not his problem:
Package: python3-numpy
Version: 1:1.21.4-2
The recent addition of python3.10:any to the dependencies
of python3-numpy under Debian testing (=bookworm) breaks
software that depends on python3-dev.
Specifically, this CMake code
find_package(Python3 COMPONENTS Interpreter Development)
Sorry, Norbert, for what was bogus in my report.
Anyway, having this correspondence in the Debian
data base may help others who will run into the
same problem.
Could you possibly give me a short hint how to
use "eatmydata" to switch off fsync?
And one more question: Isn't texlive-fonts-extra
Package: texlive-fonts-extra
Version: 2020.20210202-3
Severity: serious
Upgrading texlive-fonts-extra takes incredibly long (>20'),
and slows down all other processes.
This issue has been reported before, and has been transferred
to dpkg,
To confirm the issue:
Unpacking texlive-fonts-extra (2020.20210202-3) over (2020.20210202-1) ...
is incredibly slow.
I'm running Debian/bullseye; the SSD is ext4 and far from full.
What is the status of this issue?
What are the conclusions after all the above correspondence?
Is there any
Package: ccache
Version: 4.2-1
Dear Joel or other maintainers:
Package ccache installs a number of symbolic links to /usr/lib/ccache,
among them clang-11 and clang++-11, but not clang and clang++.
May I kindly suggest you add these two?
Thank you for your work on ccache, Joachim
smime.p7s
I now reported the upstream bug
https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=972551,
which may or may not be the root of the reported failure.
smime.p7s
Description: S/MIME Cryptographic Signature
Package: python3-numpy
Version: 1:1.19.2-2+b1
In latest Debian/testing, python3-numpy has added a dependence on
libpython3.9-minimal libpython3.9-stdlib python3.9 python3.9-minimal.
This breaks compatibility with packages like python3, python3-dev,
libpython3-dev that depend exclusively on 3.8.
Markus:
Further investigation shows that the problem is not with NumPy.
CMake not even finds Python.h.
The problem is most likely a mixture of Python 3.8 and 3.9 files on your system.
Try to uninstall libpython3-dev, which still depends on 3.8.
Good luck, Joachim
smime.p7s
Description:
Hi Markus,
I confirm that this is a serious issue.
I got the same error message
Could NOT find Python3 (missing: Python3_INCLUDE_DIRS Python3_LIBRARIES
Development NumPy Development.Module Development.Embed) (found version
"3.9.0")
in a completely different software project after I
Package: libqt5gui5
Version: 5.12.5+dfsg-2
CMake fails with the following message:
CMake Error at /usr/lib/x86_64-linux-gnu/cmake/Qt5Gui/Qt5GuiConfig.cmake:27
(message):
The imported target "Qt5::Gui" references the file
"/usr/lib/x86_64-linux-gnu/libEGL.so"
but this file does not
Package: xfce4-settings
Version: 4.14.1-1
After upgrade from Debian10 (buster) to Debian11 (bullseye = current testing)
and rebooting, xfce4 shows a blueish wallpaper with a "debian10" logo.
I should be able to override this choice from
Menu > Settings > Desktop > Background,
setting Style to
One last ping?
Additionally to what Vincas suggested, if you can reproduce the bug
and use a boot environment where systemd-journald is running, please
provide the complete output of "journalctl -b".
Sorry, colleagues: I have absolutely no time to reinvestigate this
issue and to risk
see https://superuser.com/questions/1276256
for a workaround that worked for me.
- Joachim
Could you check what packages on your system have installed have shipped
AppArmor profile, and so maybe causing the issue, like this:
dpkg -S /etc/apparmor.d/
# dpkg -S /etc/apparmor.d/
apparmor-profiles-extra, apparmor: /etc/apparmor.d
- Joachim
How did you draw the conclusion that this system
hang was caused by deinstalling the apparmor package?
After the Grub chooser, on default boot mode, I got
a black text screen, then one line, saying that /dev/sdb1
was clean, then nothing more happened.
In recovery mode, I got more than one
Package: apparmor
Version: 2.11.0-3
Apparmor can be boot critical.
Therefore, users should be warned from deinstalling.
In more detail:
Cleaning up my Debian/testing installation, it occured
to me that there was an unneeded package apparmor. So I
deinstalled it, without any conflicting
finally I reported it upstream at
https://github.com/trilinos/Trilinos/issues/1655
- Joachim
smime.p7s
Description: S/MIME Cryptographic Signature
Package: trilinos-dev
Version: 12.10.1-3
CMake's find_package takes an option COMPONENTS.
Usage with Trilinos seems to be documented nowhere,
but TrilinosConfig.cmake clearly is intended to
support e.g.
find_package(Trilinos REQUIRED COMPONENTS Epetra Belos Ifpack)
However, at the bottom of
I don't think it is possible for trilinos to depend on both versions.
Of course it should not.
But it does.
$ apt-cache show libtrilinos-amesos-dev
Package: libtrilinos-amesos-dev
Source: trilinos
Version: 12.10.1-3
..
Depends: libmumps-dev, libtrilinos-amesos12 (= 12.10.1-3), trilinos-dev
$
Package: trilinos-all-dev
Version: 12.10.1-3
Since the recent update of Debian/testing from libmumps-4.10.0 to
libmumps-5.1.1,
trilinos-all depends on both of them. This breaks my build:
make[2]: *** No rule to make target '/usr/lib/libsmumps.so', needed by
after providing an ad-hoc link
Package: trilinos-doc
Version: 12.10.1-3
Documentation of the important Trilinos component Epetra
is missing in /usr/share/doc/trilinos.
smime.p7s
Description: S/MIME Cryptographic Signature
Package: clang-format-3.9
Version: 1:3.9-6
The "SEE ALSO" section of the man page says:
"The full documentation for clang-format is maintained as a Texinfo ...
the command info clang-format should give you access to the complete manual."
This is deceptive, since "info clang-format" only prints
I get similar messages when starting evince:
Gtk-WARNING **: Failed to get the GNOME session proxy: The name
org.gnome.SessionManager is not owned
This is not only annoying. It can be badly confusing.
Today, at each attempt to start evince, it printed twice the
above message, then crashed. I
Package: cmake-data
Version: 3.6.2-2
FindBoost.cmake contains a hard-coded choice of supported versions.
In the CMake version currently distributed with Debian/testing
this lists ends with 1.61. The same Debian/testing, however, comes
with libboost-1.62.
Obviously, this is a severe
add a standard /etc/ptpd.conf to the package.
- Joachim Wuttke
smime.p7s
Description: S/MIME Cryptographic Signature
Thank you, Steve, for your detailed comment.
What's certainly different now from 6 years ago is:
Parts of Boost have made it into the standard library of C++11.
Which makes them obsolete qua part of Boost, except transitorily
to support non-compliant compilers. I would see this as one
reason
Package: libboost1.58-dev
Version: 1.58.0+dfsg-5 and others
The way Boost is currently packaged in Debian is non-standard, and inconvenient.
To test dependences and build procedures in my application, I often need to
selectively remove single Boost components from my system. This requires me
to
Package: texlive-math-extra
Version: 2015.20151225-2
The command
\usepackage[<...>]{unicode-math}
causes
! LaTeX Error: File `ucharcat.sty' not found.
because per v0.8b (2015/09/09) LaTeX package unicode-math
has introduced a dependency on LaTex package ucharcat
that is not reflected in the
Given that, can you restate what problem you encountered using the package?
Somehow I discovered that there is a libgtest-dev in Debian, and a
FindGTest in cmake. So I installed libgtest-dev on my system, removed
ThirdParty/gtest from my project, changed CMakeList.txt files, tried
to rebuild
is not and will not be packaged for
Debian.
---
Dr. Joachim Wuttke
Group Leader Scientific Computing
Jülich Centre for Neutron Science (JCNS)
Forschungszentrum Jülich GmbH
Outstation at Heinz Maier-Leibnitz Zentrum (MLZ)
+49 89 289 10715
http://apps.jcns.fz-juelich.de
smime.p7s
Description: S/MIME Cryptographic
Package: cppcheck
Version: 1.69
Running cppcheck on a medium sized project (~40 files,
20kLOC), the number of issues found depends heavily on
whether or not I use multithreading (-j7).
Until this problem is fixed upstream, as a minimal
solution the man page and cppcheck --help should
warn about
Package: texlive-fonts-extra
Version: 2014.20141024-1
Under xelatex,
\usepackage{fontspec}
\usepackage{fourier}
fails while reading fourier.sty with
! LaTeX Error: Command `\hbar' already defined.
whereas the inverse inclusion order works.
smime.p7s
Description: S/MIME Cryptographic
As upstream maintainer of libcerf, I would like to express
a clear preference for solving the conflict by withdrawing
cerf.3 and cerfc.3 from manpages-dev.
I already proposed this solution to upstream,
https://bugzilla.kernel.org/show_bug.cgi?id=80541
where it has been received favorably, but
Package: python-pygccxml
Version: 1.0.0-4
Bug 1:
Homepage: http://www.language-binding.net/pygccxml/pygccxml.html
is dead (Whoops, we can't find that page.)
Bug 2:
When used with boost-python 1.54, this type of error may occur:
The official homepage
http://sourceforge.net/projects/pygccxml/
seems to be dead; the maintainers are unresponsive.
A better solution is the unofficial, but maintained homepage
https://bitbucket.org/ompl/pyplusplus
The decision to point there should be coordinates with
project
Package: python-py++
Version: 1.0.0-1
The metadata of this package (apt-cache show python-py++) include
Homepage: http://www.language-binding.net/pyplusplus/pyplusplus.html
This page no longer exists (Whoops, we can't find that page.)
smime.p7s
Description: S/MIME Cryptographic Signature
It seems the code is currently hosted at
http://sourceforge.net/projects/pygccxml/
This is not impressive as an upstream project homepage
but certainly better than a broken link.
smime.p7s
Description: S/MIME Cryptographic Signature
Package: gnuplot-nox
Version: 4.6.5-1
== observation 1 ==
man gnuplot(1) has
SYNOPSIS
gnuplot [X11 options] [options] [file ...]
and a section X11 OPTIONS that refers to man X(1).
== observation 2 ==
When gnuplot is invoked with X11 options, one gets
$ gnuplot -title bla
unrecognized
Package: gnuplot-nox
Version: 4.6.5-1
== observation 1 ==
man gnuplot(1) has
SYNOPSIS
gnuplot [X11 options] [options] [file ...]
and a section X11 OPTIONS that refers to man X(1).
== observation 2 ==
When gnuplot is invoked with X11 options, one gets
$ gnuplot -title bla
unrecognized
please delete this one; keep #744217
sorry - Joachim
smime.p7s
Description: S/MIME Cryptographic Signature
Package: deborphan
Version: 1.7.28.8
Severity: normal
Tags: upstream
Dear Maintainer,
# apt-get install hdfview
# deborphan
hdfview
in words: after explicitly installing hdfview,
deboprhan incorrectly thinks it is an orphan.
Regards, Joachim
-- System Information:
Debian Release:
Package: emacs23
Version: 23.3+1-1
Severity: important
After migration from squeeze,emacs22 to wheezy,emacs23
emacs consistently shows the following behaviour:
As soon as I start to modify a buffer, I am interrupted by
a question showing up in the bottom line:
file locked by user@hostname (pid
Please re-open the bug:
I experience exactly the same problem as Andrea described:
dragged bookmark bookmarks can't reach the desired place, resulting in
returning at the starting point
it's very annoying - Joachim
my configuration:
Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.8.1.3)
Package: xdm
Version: 1.0.1-6
Inconsistencies in downgrading from sid to etch lead me to completely
reinstalling Debian. So, this report is about Debian etch as of today.
I installed from the CD image
http://cdimage.debian.org/cdimage/etch_di_beta2/i386/iso-cd/debian-testing-i386-netinst.iso.
I
59 matches
Mail list logo