https://bugzilla.redhat.com/show_bug.cgi?id=1885430
--- Comment #19 from giovanni.cabiddu(a)intel.com ---
I tried a build reverting that myself, and something else looks off.
Previously the build was creating the library files with this suffix:
.so.%{soversion}.%{version}
Now using the Makefile target it creates them with this suffix:
.so.%{version}
Is that a bug in the new Makefile target? This is the reason the guidelines state that
that globs should not conceal the soname.
That is expected. The upstream library is
now installing
.so.%{soversion}
.so.%{version}
Before, instead, the upstream library was not using the correct version in the
real name. So we had
.so.%{soversion}
.so.0.0.0
In the spec we were renaming .so.0.0.0 into .so.%{soversion}.%{version} and
then creating symlinks to .so.%{soversion}
In the new version of the spec I removed the glob in the %files section and
re-introduced a %global soversion with the difference that now there is a
soversion for each library we install since both might evolve independently.
Is this ok?
Spec URL:
https://raw.githubusercontent.com/intel/qatlib/v20_10/rpm/qatlib.spec
SRPM URL:
https://raw.githubusercontent.com/intel/qatlib/v20_10/rpm/qatlib-20.10.0-...
Regarding the git repo, I put a request for qatlib:
https://pagure.io/releng/fedora-scm-requests/issue/30688
--
You are receiving this mail because:
You are on the CC list for the bug.
You are always notified about changes to this product and component