DevHeads.net

Postings by Richard Shaw

BackupPC selinux help

I've got a bug report[1] for BackupPC where the user is having issues with
AVC denials when browsing hosts.

This is actually from my COPR but it's the same SRPM I use for Fedora.
There are almost 50k downloads and this is the only report of a problem so
I don't think there's a fundamental issue with the package but I would
still like to help them out.

They are getting AVC denials when browsing hosts which seems to cause
BackupPC_Admin to write LOCK files in the subdirectories of
/var/lib/BackupPC/.

fedora-review broken?

While I'm trying to use it for a review on RPM Fusion I don't think the
error is related...

Just running fedora-review without any arguments produces this in the log:

Traceback (most recent call last):
File "/usr/lib/python3.7/site-packages/FedoraReview/review_helper.py",
line 236, in run
self._do_run(outfile)
File "/usr/lib/python3.7/site-packages/FedoraReview/review_helper.py",
line 197, in _do_run
Settings.init()
File "/usr/lib/python3.7/site-packages/FedoraReview/settings.py", line
417, in init
args = parser.parse_args()
File "/usr/lib64/python3.7/argparse.py", line 1

Missing arches on EPEL 8 for LibRaw?

I assume this is because LibRaw is available in RHEL but only for x86_64
and ppc64le?

So I'm assuming there is some sort of procedure to build only for s390x and
aarch64 for EPEL?

Thanks,
Richard

Where are f31 packages going?

Now that f32 has branched when I build packages for f31 I can't add them to
a bodhi update nor are they added automatically.

Is f31 acting like pre-gating rawhide or are the packages going into the
nether?

Thanks,
Richard

python-oauth

I need python3-oauth for MythTV on RPM Fusion. I looked into it and there
hasn't been a successful build since F28. I tried a simple update to python
3 BR's and the package built fine so I went ahead and built packages for
Rawhide and F30 based on the "no one seems to care about this package"
status.

Since there hasn't been a release since 2009 it may be best to move to
another "provider" in the long term.

Can't mount bluray disk reliably with 5.1.15/16, worked fine 5.1.11

After a recent update I rebooted and could no longer mount a video bluray
disc. It doesn't have AACS and was working fine before.

The error repeats but here's a snippet:

Jul 11 20:41:52 kernel: sr 3:0:0:0: [sr0] tag#19 FAILED Result:
hostbyte=DID_OK driverbyte=DRIVER_SENSE
Jul 11 20:41:52 kernel: sr 3:0:0:0: [sr0] tag#19 Sense Key : Medium Error
[deferred]
Jul 11 20:41:52 kernel: sr 3:0:0:0: [sr0] tag#19 Add.

Character limit for spec file summary?

Is there an actual limit or guideline on the character length of the
summary tag?

I can't seem to find anything one way or the other.

Thanks,
Richard

Meaning of test gating "waiting" and "ignored"

I've submitted 100s of update but never seen test gating status "waiting"
and "ignored"

What is the meaning?

Thanks,
Richard

Proposal: Drop python2 versions of pyside and shiboken

I did some cursory searching with dnf repoquery and I can't find any
consumers of the python2 side of pyside and shiboken in Rawhide. I
recently moved over the largest consumer (FreeCAD) to the python3 pyside
bindings.

Anyone see a problem with this?

Thanks,
Richard

3DConnection device detected as a trackpoint by X

I'm trying to connect my 3DConnection Space Pilot to Fedora to use in CAD
but X is grabbing the device and treating it like a mouse.

I maintain the spacenavd package and need to find a why to block X from
grabbing the device...

[1152888.005974] usb 1-6: new low-speed USB device number 4 using xhci_hcd
[1152888.307811] usb 1-6: New USB device found, idVendor=046d,
idProduct=c625, bcdDevice= 3.12
[1152888.307815] usb 1-6: New USB device strings: Mfr=3, Product=1,
SerialNumber=0
[1152888.307817] usb 1-6: Product: SpacePilot HP
[1152888.307819] usb 1-6: Manufacturer: 3Dconnexion
[1152888.428039]

How to predict Python 3 SOABI naming scheme?!?

I'm trying to get FreeCAD back in shape on Fedora[1] and what I hope is the
last problem is that it's still building against the Python2 library
(because it's default).

grub2-editenv: error: environment block too small after kernel install

I made the mistake of editing my grubenv while converting my system from
BIOS to UEFI.

I have since manually used grub2-editenv successfully but I still get
"grub2-editenv: error: environment block too small." on kernel upgrades.
I've even tried manually re-padding the file with # to get to 1024 bytes.

I have also filed and issue upstream that grub2-editenv is too fragile. It
should be able to automatically re-pad the file to 1024 bytes.

How do I "fix" this?

Thanks,
Richard

PySide2 Python packaging questions

I'm working on packaging PySide2 for Fedora[1] and have a couple of
questions...

*** Also, I could use an experienced python packager to do the review! ***

1. Should the %python_provide macro be applied to python-<pkgname>-devel
packages? Or just for the main python module package?

2. Do I need to filter requires/provides for C python libraries if their
not private?

Thanks,
Richard

[1] <a href="https://bugzilla.redhat.com/show_bug.cgi?id=1701013" title="https://bugzilla.redhat.com/show_bug.cgi?id=1701013">https://bugzilla.redhat.com/show_bug.cgi?id=1701013</a>

Heads up: OpenColorIO 1.1.1

There's no soname change and fedabidiff output doesn't look to concerning
but I'm no expert there:

[C]'method void
OpenColorIO::v1::Processor::Impl::addColorSpaceConversion(const
OpenColorIO::v1::Config&, const OpenColorIO::v1::ConstContextRcPtr&, const
OpenColorIO::v1::ConstColorSpaceRcPtr&, const
OpenColorIO::v1::ConstColorSpaceRcPtr&)' at Processor.h:96:1 has some
indirect sub-type changes:
parameter 2 of type 'const OpenColorIO::v1::ConstContextRcPtr&' has
sub-type changes:
in referenced type 'const OpenColorIO::v1::ConstContextRcPtr':
in unqualified underlying

clang segmentation fault on armv7hf

I'm working on building PySide2 for Fedora and have a problem with clang
segfaulting only on armv7hf[1]...

The source package has shiboken2, pyside2, and pyside2-tools in one big
archive but I am building shiboken2 and pyside2-tools with GCC and only
pyside2 with clang because if makes use of something clang specific.

[ 1%] Building CXX object
libpyside/CMakeFiles/pyside2.dir/pysidesignal.cpp.o
cd
/builddir/build/BUILD/pyside-setup-everywhere-src-5.12.1/armv7hl-linux/pyside2/libpyside
&& /usr/bin/clang++ -DPYSIDE_EXPORTS -DPYSIDE_QML_PRIVATE_API_SUPPORT=1
-DPYSIDE_QML_SUPPORT=1 -DQT_CORE_L

Fork a 119MB pagure project to updating monitoring?

Am I the only person that finds this silly?

Infra things aside, is it planned to have this functionality in
src.fedoraproject.org?

On a side note, the messages from the-new-hotness still reference pkgdb...
Perhaps it should provide a link to:

<a href="https://pagure.io/releng/fedora-scm-requests/blob/master/f/README.md" title="https://pagure.io/releng/fedora-scm-requests/blob/master/f/README.md">https://pagure.io/releng/fedora-scm-requests/blob/master/f/README.md</a>

or

<a href="https://fedoraproject.org/wiki/Infrastructure/WhatHappenedToPkgdb#How_do_I_change_the_upstream-monitoring.2Fanitya_flag_for_my_packages.3F" title="https://fedoraproject.org/wiki/Infrastructure/WhatHappenedToPkgdb#How_do_I_change_the_upstream-monitoring.2Fanitya_flag_for_my_packages.3F">https://fedoraproject.org/wiki/Infrastructure/WhatHappenedToPkgdb#How_do...</a>

Thanks,
Richard

Multiple Review Requests for one source package?

With PySide2 shiboken2, pyside2, and pyside2-tools are provided in one big
source archive.

I'm working on packaging it[1] but the provided setuptools build system
calls CMake to perform builds and essentially fakeroot installs for build
dependencies between the three projects and then uses setuptools to
manipulate and copy/install the tree.

Problem 1:
Whether a bug in the setuptools script or a bug in setuptools itself
(exposed by their install method) instead of "installing" all three project
in %install it instead installs the same project 3 times when the --root
option is used[2].

I have

PySide2: 64bit builds fail but 32bit builds succeed (Clang issue?)

I'm working on getting PySide2 into Fedora which gives you python bindings
for Qt5.

It uses some code specific to Clang so I can't use gcc.

Scratch build uploads to koji VERY SLOW

Is anyone else experiencing this? I did a speed test on my Xfinity and it
doesn't appear to be on my end.

I'm getting about 100-200kB/s uploads... On larger projects that's very
painful.

Thanks,
Richard

qt4 rebuild

I'm troubleshooting why apiextractor tests segfault during package
building.

apiextractor FTBFS troubleshooting

apiextractor builds but pretty much all the tests fail.

f30 buildroot

How can I tell if a package is available in the buildroot for f30?

Per koji my build has the "f30" tag but it's still not available when I try
to use it in mock.

<a href="https://koji.fedoraproject.org/koji/buildinfo?buildID=1215080" title="https://koji.fedoraproject.org/koji/buildinfo?buildID=1215080">https://koji.fedoraproject.org/koji/buildinfo?buildID=1215080</a>

Thanks,
Richard

Non-responsive maintainer: codeblock

Is anyone able to contact Rick?

RFE: fedpkgdiff?

I'm curious to see if anyone else would find this useful...

What about a fedpkgdiff as a wrapper around pkgdiff (similar to
fedabipkgdiff) but just to see the difference between two packages (either
local or from the repos like fedabipkgdiff), specifically differences in:

- installed files (not contents)
- requires
- provides

Thoughts?

Thanks,
Richard

Retire apiextractor?

Currently apiextractor is FTBFS[1] because all the test fail.

It looking to see if there is a newer version that better supports gcc
9.0.x I found this[2] which states that apiextrator was merged into
shiboken.

So can someone with more familiarity with pyside/shiboken/apiextractor say
if apiextracor should be retired?

Thanks,
Richard

[1] <a href="https://bugzilla.redhat.com/show_bug.cgi?id=1674649" title="https://bugzilla.redhat.com/show_bug.cgi?id=1674649">https://bugzilla.redhat.com/show_bug.cgi?id=1674649</a>
[2] <a href="https://wiki.qt.io/PySide_API_Extractor" title="https://wiki.qt.io/PySide_API_Extractor">https://wiki.qt.io/PySide_API_Extractor</a>

java: Unrecognized option: --add-modules

When did java lose its option for --add-modules?!?!?

# java --add-modules java.activation
Unrecognized option: --add-modules
Error: Could not create the Java Virtual Machine.
Error: A fatal exception has occurred. Program will exit.

# rpm -q java-1.8.0-openjdk
java-1.8.0-openjdk-1.8.0.191.b13-0.fc29.x86_64

I can't start my unifi server now because it needs the java.activation
module that's depreciated.

Thanks,
Richard

%py3_build should accept options

Just want to get others opinions before filing a bug...

Currently %py3_build expands to:

$ rpm -E %py3_build
\
CFLAGS="${CFLAGS:-${RPM_OPT_FLAGS}}"
LDFLAGS="${LDFLAGS:-${RPM_LD_FLAGS}}"\
/usr/bin/python3 setup.py build --executable="/usr/bin/python3 -s"
sleep 1

This does not allow passing of any options that I can tell... For now I'll
just paste this into my spec file.

Thanks,
Richard

install command nit

I noticed a warning on one of my packages that the appdata file was marked
as executable. The source file is not, its 644, but I use the install
command to install it to the metainfo directory.

Does "install" install files as executable by default or is something else
going on here?

Adding -m 0644 fixed it but it shouldn't be necessary.

Thanks,
Richard

Any interest in PySide2 (Qt 5 bindings) for Fedora?

I found the following BZ ticket which explains this situation...

<a href="https://bugzilla.redhat.com/show_bug.cgi?id=1634658" title="https://bugzilla.redhat.com/show_bug.cgi?id=1634658">https://bugzilla.redhat.com/show_bug.cgi?id=1634658</a>

FreeCAD chose to use PySide over PyQt so this (and a couple of other
packages) are preventing me from switching it over to Qt5.

Is there anyone interested in getting PySide2 in Fedora and willing to
submit a Review Request? I'll volunteer to be the reviewer..

Thanks,
Richard