[Bug 605373] Review Request: qgis - A user friendly Open Source Geographic Information System

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

 



Please do not reply directly to this email. All additional
comments should be made in the comments box of this bug.


https://bugzilla.redhat.com/show_bug.cgi?id=605373

--- Comment #2 from Kevin Kofler <kevin@xxxxxxxxxxxxxxxx> 2010-06-22 19:48:20 EDT ---
So, before I start reviewing the package against the packaging criteria, I'm
comparing it with the current (orphaned) qgis.spec, mainly to check for upgrade
path issues (and found one of those among other things, see point 3 below).

1. I see you're putting the %description entries all after the %package
entries. It's more common to list both %package and %description for each
subpackage together like the old specfile did it. (But the way you're putting
them is not incorrect, just unusual.)

2. The Summary entries in %package python and %package grass are swapped.
Please fix them.

3. You dropped the theme subpackages. If you drop a subpackage from an existing
Fedora package, you have to Obsolete it for upgrade paths. Please add:
Obsoletes: qgis-theme-classic < 1.1
Obsoletes: qgis-theme-gis < 1.1
Obsoletes: qgis-theme-nkids < 1.1
to the main package.

4. One bizarre thing which was already in the old specfile is that
BuildRequires are listed per subpackage. This doesn't make much sense,
BuildRequires are per SRPM, not per subpackage. The grass BRs are even listed
twice (once in the main package and once with the subpackage). While this is
not strictly required, I'd suggest listing all the BuildRequires together.

5. You're setting GRASS_PREFIX twice, once with the snippet copied from the
original specfile and once with your GRASS_PREFIX=/usr/lib64/. The last setting
prevails, so please remove the unused snippet. And you must fix the hardcoded
/usr/lib64/ to %{_libdir} or it won't build on 32-bit.

6. VERBOSE=1 in the make command line (which you added) is not needed because
the %cmake macro already sets -DCMAKE_VERBOSE_MAKEFILE=ON, i.e. VERBOSE=1 is
the default anyway.

7. You changed the chrpath call to (i) zap the rpath from the libraries and
(ii) not zap any rpath from the executables, are you sure this is right? If in
doubt, please use chrpath on everything (both the libraries and the
executables). But then I'd like to know where the rpath is coming from in the
first place. (Have you actually checked whether there's an rpath?) Also try
just using %{_cmake_skip_rpath} in the %cmake call and see if that makes the
rpaths go away. (It's better to not generate rpaths in the first place than to
remove them.)

8. You added an additional chmod +x on libraries, are you sure this (and the
one which was already there) is needed? The current CMake should be installing
shared libraries with +x permissions already!

9. The devel subpackage doesn't need any ldconfig calls (ldconfig calls are
only needed for shared libraries, not devel symlinks to them), please remove
the 2 scriptlets you added.

10. Just a hint: instead of:
%dir %{_libdir}/%{name}
%{_libdir}/%{name}/*
you can just list:
%{_libdir}/%{name}/
in the file list, it will have the same effect (and yes, %exclude also works on
that).

-- 
Configure bugmail: https://bugzilla.redhat.com/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.
_______________________________________________
package-review mailing list
package-review@xxxxxxxxxxxxxxxxxxxxxxx
https://admin.fedoraproject.org/mailman/listinfo/package-review


[Index of Archives]     [Fedora Legacy]     [Fedora Desktop]     [Fedora SELinux]     [Yosemite News]     [KDE Users]     [Fedora Tools]