DevHeads.net

Postings by Richard Shaw

Unhelpful message about noarch packages differing between arches...

Ok, I was just doing a simple rebuilt of root as OCE was updated to a new
version in rawhide...

The F26 rebuild worked fine but on rawhide I'm getting:

BuildError: The following noarch package built differently on
different architectures: root-doc-6.08.06-4.fc27.noarch.rpm
rpmdiff output was:
error: cannot open Packages index using db5 - Permission denied (13)
error: cannot open Packages database in /var/lib/rpm
error: cannot open Packages database in /var/lib/rpm
removed /usr/share/doc/root/html/pict1_rs701_BayesianCalculator.C.png
removed /usr/share/doc/root/html/pict1_zdemo.py.png

s390x builders down?

Subject says it all...

Thanks,
Richard

F26 update ejected?

I have two packages which I got notifications that they were "ejected" from
the push, an example:

BackupPC-4.1.2-1.fc26 ejected from the push because u"Cannot find relevant
tag for BackupPC-4.1.2-1.fc26.

fedpkg new-sources still uploading twice?

I know this was reported as far back as December...

Am I the only one still seeing sources getting uploaded twice?

Thanks,
Richard

Best practice to allow access in packaging scrips

I'm working on improving the packaging for BackupPC, one of the issues is
that it uses a lot of non-standard permission to prevent user access to
potentially sensitive content.

Currently, for the web interface to work the apache user needs to be added
to the backuppc group so that /etc/BackupPC/apache.users can be read for
the password protected login.

Would it be a good idea to do this in %pre/post?

usermod -a -G backuppc apache

Or perhaps use setfacl?

There's several steps the end user has to do but I'd like to automate as
much as possible.

Thanks,
Richard

Upcoming BackupPC 4.0

BackupPC 4.0 has been released but there major changes that prevent a
seamless upgrade and some new dependencies that are not in Fedora.

With version 4.0 hard links are no longer used for deduplication and
instead attribute files are stored in each directory.

While 4.0 is backwards compatible with 3.X backups there are some client
side configuration changes that may cause problems.

Additionally there are two new requirements:

rsync-bpc - A fork of rsync to speed up the previous perl based version in
3.X that has added capabilities needed for BackupPC

<a href="https://bugzilla.redhat.com/show_bug.c" title="https://bugzilla.redhat.com/show_bug.c">https://bugzilla.redhat.com/show_bug.c</a>

GRIPE: A package is not FTBFS if the dependencies can't be installed

I received a FTBFS[1] on one of my packages for the F26 rebuild only to
find out the only reason it failed (that I know of thus far) is that not
all the dependencies could be installed.

The root cause seems to be that not all dependent packages were rebuilt
when boost was updated to 1.63.

%pre is a lie?

I'm working on creating a package for the UniFi controller software for
Fedora/EL7.

Hamlib 3.1 coming to rawhide

Probably not of interest to most but I plan on building hamlib 3.1 and
rebuilding its dependencies over the next couple of days.

# dnf repoquery --alldeps --whatrequires hamlib --queryformat '%{name}'
cqrlog
fldigi
freedv
grig
hamlib
hamlib-c++
hamlib-devel
klog
perl-hamlib
python-hamlib
qle
qsstv
soundmodem
tcl-hamlib
tucnak2
xlog

Thanks,
Richard

OpenImageIO: i686 builds failing

I could use some help figuring out the best fix for the following problem
where only the i686 builds fail:

cd /builddir/build/BUILD/oiio-Release-1.7.7/build/linux/src/libutil &&
/usr/bin/cmake -E cmake_link_script CMakeFiles/fmath_test.dir/link.txt
--verbose=1
/usr/bin/c++ -O2 -g -pipe -Wall -Werror=format-security
-Wp,-D_FORTIFY_SOURCE=2 -fexceptions -fstack-protector-strong
--param=ssp-buffer-size=4 -grecord-gcc-switches
-specs=/usr/lib/rpm/redhat/redhat-hardened-cc1 -m32 -march=i686 -mtune=atom
-fasynchronous-unwind-tables -O2 -g -DNDEBUG -Wl,-z,relro
-specs=/usr/lib/rpm/redhat/redhat

Non-responsive maintainer: corsepiu

Unfortunately I am in the position that I must initiate the non-responsive
maintainer policy for Ralf.

The Coin3 package has a serious bug[1] that was only recently uncovered by
GCC 6. During the process of figuring out the problem (and solution) I
asked the devel list for help. Thanks to those that quickly came to my aid.

In my efforts to get the problem fixed quickly I went ahead pushed a fix
into rawhide.

rebase-helper broken?

I keep seeing attempted rebuilds of new versions of my packages fail but in
almost every case I update the version, spectool -g the source down, and
rebuild without issues. The rebase-helper logs are not particularly human
friendly, in this case it looks like it tried to build the old version?!?

<a href="https://bugzilla.redhat.com/show_bug.cgi?id=1329024#c5" title="https://bugzilla.redhat.com/show_bug.cgi?id=1329024#c5">https://bugzilla.redhat.com/show_bug.cgi?id=1329024#c5</a>

Thanks,
Richard

HEADS-UP: Updated yaml-cpp coming to rawhide

I plan to build the latest version of yaml-cpp, 0.5.3, for rawhide soon.
There's no announced soversion bump but abi-compliance-checker doesn't
think they're compatible[1] so probably best to play it safe and rebuild
dependencies.

# dnf repoquery --alldeps --whatrequires "libyaml-cpp.so.0.5()(64bit)"
--source
Local Packages for Fedora 23 57 MB/s | 172 kB 00:00
calamares-1.1.3-1.fc23.src.rpm
calamares-1.1.4.2-1.fc23.src.rpm
librime-1.2-6.fc23.src.rpm
mongodb-3.0.6-1.fc23.src.rpm
mongodb-3.0.9-1.fc23.src.rpm
pdns-3.4.6-1.fc23.src.rpm
pdns-3.4.8-1.fc23.src.rpm
yaml-cpp-0.

freecad: Need help with segfault problem on F24

I'm pretty much stumped on a problem where freecad (really python-pivy)
segfaults on F24 but seems to run fine on F23 and I can't find any
substantive difference between any of the packages.

Strangely installing Coin3 from f23 on a f24 system seems to fix the
problem...

<a href="https://bugzilla.redhat.com/show_bug.cgi?id=1323159" title="https://bugzilla.redhat.com/show_bug.cgi?id=1323159">https://bugzilla.redhat.com/show_bug.cgi?id=1323159</a>

Any pointers/help would be appreciated.

Thanks,
Richard

Koschei eating Koji?

I have finally gotten enough packages that I'm starting to run into more
that 2 level deep build dependencies and I'm having a lot of trouble
getting everything built.

I have requested some build root overrides since yesterday and per the
recommended command they don't seem to be complete yet.

I've noticed general sluggishness of the Koji web interface and a large
number of Koschei builds in process.

New OCE and smesh coming to rawhide

Fedora is currently two major and one minor release behind on OCE due to
the fact that smesh had not been updated to some library/API changes.
Secondarily smesh would not build on GCC 6 but that has now been fixed.

I plan to build OCE 0.17.1 and newly released smesh 6.6 in the next day or
so.

On a tangent, I was able to figure out required rebuilds with yum/repoquery
but I can't seem to find the magic incantation for "dnf repoquery".

Is there a documented "best practice/known good" command in the wiki and I
just can't find it?

Thanks,
Richard

FreeCAD Fedora 24 packages missing all library dependencies.

I recently got a rash of bug report that FreeCAD on f24 was segfaulting.
After looking at a few things I noticed that the f24 builds seem to be
missing all library based dependencies.

Has this happened to anyone else?

Thanks,
Richard

Blender: New error with min/max function not in scope

I was performing a rebuild of Blender after fixing a FTBFS error with
openCOLLADA and ran into an interesting error:

[ 78%] Building CXX object
source/blender/imbuf/intern/dds/CMakeFiles/bf_imbuf_dds.dir/DirectDrawSurface.cpp.o
cd
/builddir/build/BUILD/blender-2.76/cmake-make/source/blender/imbuf/intern/dds
&& /usr/lib64/ccache/c++ -DHAVE_STDBOOL_H -DNDEBUG -DWITH_DDS
-D_FILE_OFFSET_BITS=64 -D_LARGEFILE64_SOURCE -D_LARGEFILE_SOURCE
-D__LITTLE_ENDIAN__ -D__MMX__ -D__SSE2__ -D__SSE__
-I/builddir/build/BUILD/blender-2.76/source/blender/imbuf/intern/dds
-I/builddir/build/BUILD/blender-2.76/sour

GCC 6 Binaries segfaulting

I have one package where several of the binaries used for unit testing are
segfaulting.

I'm guessing that's not likely a direct GCC 6 issue but some googling leads
me to believe that it could be an ABI breakage with a dependency that has
not been rebuilt.

Is that the most likely cause?

Thanks,
Richard

openCOLLADA: Help with GCC6 narrowing conversion

Can someone point me in the right direction? My package openCOLLADA is
FTBFS in rawhide with the following (repeating) error:

/builddir/build/BUILD/OpenCOLLADA-3335ac164e68b2512a40914b14c74db260e6ff7d/COLLADABaseUtils/src/COLLADABUURI.cpp:57:2:
error: narrowing conversion of '-1' from 'int' to 'char' inside { }
[-Wnarrowing]
};

I tried using the gcc6 porting guide but I'm not qutie enough of a
programmer to understand how to apply it in this case.

Rawhide builds complete in mock but fail on koji

I'm getting the following error when I build on koji but all mock builds
complete.

Help with Rawhide build error with GCC 6.0

With the release of GCC 6.0 in Rawhide I'm having a build
warning/error[1,2] with OpenImageIO I'm not sure what to do with (other
than adding a flag to ignore it).

Upstream is looking into it but currently thinks that the pugixml API is
requiring a method that GCC 6.0 doesn't like:

[ 3%] Building CXX object
src/libutil/CMakeFiles/OpenImageIO_Util.dir/filter.cpp.o
cd /builddir/build/BUILD/oiio-Release-1.6.9/build/linux/src/libutil &&
/usr/bin/c++ -DNDEBUG -DOpenImageIO_Util_EXPORTS -DUSE_EXTERNAL_PUGIXML=1
-DUSE_FIELD3D=1 -DUSE_FREETYPE -DUSE_OCIO=1 -DUSE_OPENCV
-DUSE_OPENEXR_VERSION2=1 -

Upstream release monitoring issue

I was notified by a user that a newer version of trustedqsl was available
so I looked into why I wasn't notified.

For some reason it still thinks 1.13 is the current version when it is 2.2:

<a href="https://release-monitoring.org/project/4998/" title="https://release-monitoring.org/project/4998/">https://release-monitoring.org/project/4998/</a>

I've tried tweaking things but can't seem to find the magic formula to get
it to work.

Any suggestions?

Thanks,
Richard

How to test appdata for a package?

I' m working on a package for a review and have added what I believe is a
good appdata file, at least it passes appdata validation during the build,
but how in the heck are you supposed to be able to test what the result
will look like in gnome-software?

I thought I successfully used gnome-software --prefer-local before but no
matter what I do it doesn't find the package.

I've even gone as far as to try appstream-util install /path/to/appdata.xml
but I'm not sure if that does what I want as the documentation is pretty
light for the options.

Thanks,
Richard

yum: Critical path update in testing for 4 months?

Usually I don't dig too deep but I saw a "critical path" update for yum on
F22 (which should be using dnf anyway).

So why is a critical path update with +10 karma still in testing?

(I know how, there's no auto push to stable based on karma for this update)

Thanks,
Richard

Dealing with rolling release versioning

I have a project that stopped providing versioned releases and went to a
rolling release model using the date.

In this case these are not "pre" or "post" releases or snapshot releases.

Is there any reason not to use the date as the version? It's in YYYYMMDD
format so there shouldn't be a upgrade path issue but this isn't explicitly
covered in the packaging guidelines that I can find.

Thanks,
Richard

Automate nag emails for pending ACL/pkgdb requests?

This is not about any particular instance, but browsing around pkgdb for
various reasons I've run across ACL/pkgdb request that haven't gotten
approved (or rejected). I know we all get busy but it's not right to ignore
(intentionally or not) these requests from other packagers.

Some might be in favor of auto approval after a defined period of
inactivity but what if someone goes on an extended vacation? How long is
too long?

I think the least disruptive approach would be to start sending nag emails
after a certain period (1 week?).

Thoughts?

Thanks,
Richard

Guidelines unclear for unretire of a branch

I would like to unretire the epel7 branch of LibRaw, per the guidelines[1]
this woudl require a re-review, but that would only make sense if the whole
package was retired, not an epel branch.

Why would we re-review something that still have active branches
(specifically rawhide)?

Thanks,
Richard

[1] <a href="https://fedoraproject.org/wiki/PackageDB_admin_requests#Other_requests" title="https://fedoraproject.org/wiki/PackageDB_admin_requests#Other_requests">https://fedoraproject.org/wiki/PackageDB_admin_requests#Other_requests</a>

flmsg.i686: E: missing-call-to-setgroups-before-setuid /usr/bin/flmsg

I have a review request[1] where rpmlint complained about:

flmsg.i686: E: missing-call-to-setgroups-before-setuid /usr/bin/flmsg

Which is caused by bundled mongoose server. There is an issue logged
upstream where the developer is reluctant to add setgroups:

<a href="https://github.com/cesanta/mongoose/issues/301" title="https://github.com/cesanta/mongoose/issues/301">https://github.com/cesanta/mongoose/issues/301</a>

Is that a valid reason or does the problem still need to be addressed?

Thanks,
Richard

[1] <a href="https://bugzilla.redhat.com/show_bug.cgi?id=1218758" title="https://bugzilla.redhat.com/show_bug.cgi?id=1218758">https://bugzilla.redhat.com/show_bug.cgi?id=1218758</a>

"Fixing" mcelog on AMD cpu's

All of my computers at home are AMD based and a while ago I noticed that
the mcelog service always fails. It has been reported on BZ [1]

The easy solution was to just disable it but I wasn't sure how important it
was at the time. Turns out it's just flat out incompatible with modern AMD
processors[2,3].

So what do we do?