https://bugzilla.rpmfusion.org/show_bug.cgi?id=4993
FeRD (Frank Dana) <[email protected]> changed:
What |Removed |Added
----------------------------------------------------------------------------
CC| |[email protected]
--- Comment #13 from FeRD (Frank Dana) <[email protected]> ---
Crap, I didn't realize my email reply wouldn't make it into the bug. That's my
fault, sorry. I had written a response to #11...
========
(In reply to Akarshan Biswas from comment #11)
> (In reply to Nicolas Chauvet from comment #10)
>
> > Are you sure the libva dependency needs to be explicitly added ?
>
> yes, video acceleration api depends on libva. We are using the system libva
> for the build.
That's not the question, the issue is whether the dependency needs to be
explicitly stated in the spec file itself.
Fedora (and thus RPMfusion) packaging policy is to let rpmbuild handle library
dependencies automatically wherever possible. If chromium is being linked with
any /usr/lib64/libva*.so*, that'll be discovered automatically when the
binaries are packaged up, and those specific libraries will be added as runtime
dependencies. That's generally preferable to specifying them automatically.
Typical pattern is to set e.g. "BuildRequires: libva-devel", to ensure the
appropriate headers and unversioned .so files are present at compile time, and
then let rpmbuild figure out the runtime requires for itself. In standard
compile-and-link scenarios involving shared libraries, it does a far better job
than we can in the spec file.
See:
https://fedoraproject.org/wiki/Packaging:Guidelines?rd=Packaging/Guidelines#Explicit_Requires
========
...Unfortunately, looking at the new .spec file, you did exactly the opposite
of that. rpmbuild will add all of those Requires itself (at the individual
shared library, rather than package, level), they should not be explicitly
listed in the .spec.
--
You are receiving this mail because:
You are on the CC list for the bug.
You are the assignee for the bug._______________________________________________
rpmfusion-developers mailing list -- [email protected]
To unsubscribe send an email to [email protected]
