On Mon, Nov 25, 2019 at 4:59 PM Jerry James <loganjerry@xxxxxxxxx> wrote:
>
> On Mon, Nov 25, 2019 at 5:14 AM Fabio Valentini <decathorpe@xxxxxxxxx> wrote:
> > PS: Feel free to browse the data in the linked pagure repo. I'm
> > regenerating it daily with the latest state of all fedora branches. If
> > somebody is interested in getting notified about any of their packages
> > getting broken deps (for example, in -testing, before things get
> > broken in stable), I can work out some kind of automatic (opt-in)
> > notification system :)
Hi Jerry!
> This turned up a problem in the gap-pkg-polenta package, which I wish
> I had known about sooner. So, yes, I am interested in some kind of
> notification. I miss the old days when the broken deps report was
> part of the Rawhide compose email. Is there any chance of bringing
> that back? It was very useful.
Yes, there's some interest in getting the reports working again,
though there are some limitations in DNF (most are related to rich
dependencies) that need to be worked around somehow. Once it works
reliably (and correctly), we can work in integrating it into the
infrastructure somehow (maybe as a checking step after composes).
> The report lists the mscore package, which surprised me, since I just
> built it a few days ago. It says:
>
> - mscore (src):
> o pkgconfig(Qt5WebEngine)
>
> There is no problem installing 'pkgconfig(Qt5WebEngine)' in mock on my
> machine. From the spec file:
>
> %ifarch %{qt5_qtwebengine_arches}
> BuildRequires: pkgconfig(Qt5WebEngine)
> %endif
>
> I suspect the check was run on a machine with an architecture not in
> qt5_qtwebengine_arches. In that case, this is a false positive. Do
> you have some way of telling when a BuildRequires is inside an %ifarch
> or %ifnarch block?
Ah, I see what's happening here. So mscore doesn't have ExcludeArch
set, but some feature (presumably an embedded web view?) is only
available on some architectures? I assume that the srpm is rebuilt
somewhere on an architecture that has this feature, and this srpm is
then copied into the -source repositories for all architectures, even
those where this dependency isn't available. This is not a problem
since this conditional is expanded again at build time, but the srpm
metadata contains it unconditionally (which is what matters for
repository metadata, I guess). Maybe that's even a bug :)
I do have some emulation of querying "ExcludeArch" and "ExclusiveArch"
(because I can't query that from the repos). But there's no way (that
I know of) to check for architecture-dependent BuildRequires, except
maybe parsing .spec files directly (which is something I *really*
don't want to do) or rebuilding the srpm files on the target
architecture (which probably would blow up the time required to
generate the report by some orders of magnitude).
I think this kind of "optional BuildRequires" shouldn't be something
too common in fedora, so I can probaly add this stuff to the whitelist
...
> Thanks for running this report, Fabio.
Sure. I'm glad it already proved useful in one case :)
Fabio
> --
> Jerry James
> http://www.jamezone.org/
> _______________________________________________
> devel mailing list -- devel@xxxxxxxxxxxxxxxxxxxxxxx
> To unsubscribe send an email to devel-leave@xxxxxxxxxxxxxxxxxxxxxxx
> Fedora Code of Conduct: https://docs.fedoraproject.org/en-US/project/code-of-conduct/
> List Guidelines: https://fedoraproject.org/wiki/Mailing_list_guidelines
> List Archives: https://lists.fedoraproject.org/archives/list/devel@xxxxxxxxxxxxxxxxxxxxxxx
_______________________________________________
devel mailing list -- devel@xxxxxxxxxxxxxxxxxxxxxxx
To unsubscribe send an email to devel-leave@xxxxxxxxxxxxxxxxxxxxxxx
Fedora Code of Conduct: https://docs.fedoraproject.org/en-US/project/code-of-conduct/
List Guidelines: https://fedoraproject.org/wiki/Mailing_list_guidelines
List Archives: https://lists.fedoraproject.org/archives/list/devel@xxxxxxxxxxxxxxxxxxxxxxx
