Re: [Libreoffice-commits] core.git: Fix CustomTarget_postprocess/check_dynamic_objects for --enable-qt5

Am 07.02.2018 um 14:21 schrieb Stephan Bergmann:
> On 07.02.2018 11:46, Rene Engelhard wrote:
> > On Wed, Feb 07, 2018 at 09:57:46AM +0100, Stephan Bergmann wrote:
> > > On 07.02.2018 09:33, Stephan Bergmann wrote:
> > > > commit bf23fe3a4a3eebe3d08fa26341a486b83d80c1c7
> > > > Author: Stephan Bergmann <sbergman@redhat.com>
> > > > Date:   Wed Feb 7 09:11:35 2018 +0100
> > > >
> > > >       Fix CustomTarget_postprocess/check_dynamic_objects for
> > > > --enable-qt5
> > > >       The dependency of Library_vcl on qt5 libraries (via
> > > > vcl/qt5/Qt5Font.cxx) is
> > > >       there ever since b66a7cbd8491fe436126e11975c360f47ae346ed "QT5
> > > > first stab on
> > > >       implementing CommonSalLayout", but no idea whether that is by
> > > > design or is a
> > > >       hack that should eventually be cleaned up.
> > >
> > > Any thoughts on the above?  It might e.g. by undesirable for distros
> > > that do
> > > --enable-qt5 but split libvclplug_qt5lo.so into a Qt5-dependent
> > > package of
> > > its own (and don't want the base vcl package to depend on Qt5).
> >
> > Yes, I already "complained" about this, jmux had a look/wanted to have
> > a look. This is a blocker for me enabling qt5. I am so not making -core
> > depending on qt5.
>
> So whoever will eventually fix that, please remember to revert the
> relevant part of this bf23fe3a4a3eebe3d08fa26341a486b83d80c1c7, too.

There are two commits to fix this:

https://gerrit.libreoffice.org/#/c/47279/
https://gerrit.libreoffice.org/#/c/47408/

I simply didn't have enough time to test these. They compile and "work
for me". But this eventually breaks embedded font handling and some code
I changed didn't made much sense. There were some reinterpret_cast stuff
to fit those embedded font handling into the normal font handling in
some obscure way, I don't yet grok. Then there is the hb_font handling
on Windows, which - I guess for some resource handling - is places in
the PhysicalFontFace instead of the LogicalFontInstance.

But currently I don't have time to test this with all the stuff (AKA a
Windows and Mac build, embedded font handling).

Probably it would be easier to just move the Qt5 stuff; but then I
wanted to clean this up and move the platform specific code into their
respective VCL modules. I also started writing documentation, about the
way the font handling works / should work.

Fact is I'm currently overwhelmed by non-LO work. If someone else has
some time do the testing I would be happy. I can push this, but don't
have time to fix regressions.

And I would like to mentor some Qt5 GSoC, which I have to replace the
KDE stuff on the ideas page.

And I have rebased patches, if that helps.

Jan-Marek