Created attachment 1947096 [details] conflict report (/var/tmp/tmt/run-*/rpmdeplint/execute/data/check-conflicts-x86_64/output.txt) Description of problem: Latest tbb build is stuck in gating: https://bodhi.fedoraproject.org/updates/FEDORA-2023-4acfa3bbcc While the reason is the tier0 test, there are other failures worth considering. One of them is fedora-ci.koji-build.rpmdeplint.functional The logs have already been deleted but the failure can be reproduced manually: git clone https://github.com/fedora-ci/rpmdeplint-pipeline.git cd rpmdeplint-pipeline tmt run --all --debug -e RELEASE_ID=f39 -e TASK_ID=97820120 (The task ID comes from the corresponding build in Koji.) I'm attaching the report of the failure but in essence tbb-doc.x86_64 and tbb-doc.i686 conflict on several files, they can't be installed in parallel but their conflict isn't declared. Another way to reproduce the problem is downloading and installing the build's RPMs: koji download-build --arch={x86_64,i686,noarch} tbb-2020.3-16.fc39 dnf -y install ./*.rpm Version-Release number of selected component (if applicable): tbb-2020.3-16.fc39
Those files should really be identical, and so wouldn't conflict. But there are spurious ordering differences between some of the HTML files, e.g. (with line breaks inserted to show the differences more clearly): --- i686/usr/share/doc/tbb-doc/html/a01521.html 2023-03-01 12:09:31.227326646 +0000 +++ x86_64/usr/share/doc/tbb-doc/html/a01521.html 2023-03-01 12:09:31.230326648 +0000 @@ -596,10 +596,9 @@ <div class="ttc" id="aa01513_html_a51df5166cb93eb4aabf15c0604cb6257"><div class="ttname"><a href="a01513.html#a51df5166cb93eb4aabf15c0604cb6257">tbb::blocked_range::grainsize</a></div><div class="ttdeci">size_type grainsize() const</div><div class="ttdoc">The grain size for this range.</div><div class="ttdef"><b>Definition:</b> <a href="a00017_source.html#l00082">blocked_range.h:82</a></div></div> <div class="ttc" id="aa01513_html_aa5b02cb98cc0ad6c0cd5d29bb1271b7c"><div class="ttname"><a href="a01513.html#aa5b02cb98cc0ad6c0cd5d29bb1271b7c">tbb::blocked_range::do_split</a></div><div class="ttdeci">static Value do_split(blocked_range &r, split)</div><div class="ttdoc">Auxiliary function used by the splitting constructor.</div><div class="ttdef"><b>Definition:</b> <a href="a00017_source.html#l00131">blocked_range.h:131</a></div></div> </div><!-- fragment --> -<p class="reference">References -<a class="el" href="a01513.html#aa5b02cb98cc0ad6c0cd5d29bb1271b7c">tbb::blocked_range< PageValue >::do_split()</a>, -<a class="el" href="a01513.html#aa5b02cb98cc0ad6c0cd5d29bb1271b7c">tbb::blocked_range< RowValue >::do_split()</a>, +<p class="reference">References <a class="el" href="a01513.html#aa5b02cb98cc0ad6c0cd5d29bb1271b7c">tbb::blocked_range< PageValue >::do_split()</a>, <a class="el" href="a01513.html#aa5b02cb98cc0ad6c0cd5d29bb1271b7c">tbb::blocked_range< Value >::do_split()</a>, +<a class="el" href="a01513.html#aa5b02cb98cc0ad6c0cd5d29bb1271b7c">tbb::blocked_range< RowValue >::do_split()</a>, <a class="el" href="a01513.html#a51df5166cb93eb4aabf15c0604cb6257">tbb::blocked_range< Value >::grainsize()</a>, <a class="el" href="a01513.html#a3e06d6e390716f14508a0d77732cccfa">tbb::blocked_range< Value >::my_begin</a>, <a class="el" href="a01521.html#a11e731b20c82fec4346d3be2931d06e9">tbb::blocked_range3d< PageValue, RowValue, ColValue >::my_cols</a>, It seems like Doxygen doesn't produce consistent output here, which is annoying.
The simplest solution is probably just to add the Conflicts: tags, rather than fighting Doxygen.
N.B. tbb-doc isn't a multilib-enabled package, so users will never see tbb-doc.i686 in the x86_64 repo anyway. So they can only encounter this problem by downloading the i686 package and trying to install it manually. Ideally the tbb-doc would be BuildArch: noarch, but that fails: BuildError: The following noarch package built differently on different architectures: tbb-doc-2020.3-18.fc39.noarch.rpm
This should be fixed in https://bodhi.fedoraproject.org/updates/FEDORA-2023-72982bd6c6 but the tests seem broken
Looks good to me, thanks. fedora-ci.koji-build.installability.functional is indeed broken (incompatible with DNF5, to be specific) but I hear a fix is on the way. I checked the functionality manually and waived the test. fedora-ci.koji-build.rpmdeplint.functional failed b/c of slow infra, if I may guess. Again, both the test and the infra it runs on are beyond my direct reach but I will look it later. Anyway, I think it's not a blocking test and the build is free to go...
Great, thanks for checking it