Comment # 3 on bug 4368 from Kevin Kofler

> APPROVED

Thanks! That was quick.
Can you please set the fedora-review+ flag?

> 1.  MUST use verifiable sources, can you account for this difference?  (maybe used a different snapshot?)

Looks like Qt upstream respun all its tarballs on:
http://download.qt.io/snapshots/qt/5.7/5.7.1/latest_src/submodules/
on December 5. I will update to the latest tarball on import. But the Qt 5.7.1
packages in Fedora (including QtWebEngine) need updating too.

> builds/installs: ok
>
> runtime not tested (qt-5.7.1 available only for rawhide a the moment)

I'll have a try in a VM (downloading some Rawhide ISOs now, but it takes some
time), but I don't expect nasty surprises, this uses the same trick as
freetype-freeworld.

> 2.  SHOULD confirm qtbase dependency
> %{?_qt5_version:Requires: qt5-qtbase%{?_isa} >= %{_qt5_version}}
>
> for the fedora package we currently use a tighter,
> %{?_qt5:Requires: %{_qt5}%{?_isa} = %{_qt5_version}}

This is actually one of the many lines copies as is from the current Fedora
Rawhide package:
http://pkgs.fedoraproject.org/cgit/rpms/qt5-qtwebengine.git/tree/qt5-qtwebengine.spec#n277

I should probably fix the >= to > in both packages.

> 3.  SHOULD consider using -debuginfo shrinking hacks on all archs (including x86_64)

Well, we had this discussion already. :-) This is really a tradeoff between
quality and size of debugging information. On 32-bit architectures, we have no
other option than using -g1 because the linker will run out of address space
with -g. On x86_64, we have no such issue.