These Astronomy_%26_Astrophysics tests now succeed:
export/examples/Articles/Astronomy_%26_Astrophysics_dvi3_systemF
export/examples/Articles/Astronomy_%26_Astrophysics_pdf4_systemF
export/examples/Articles/Astronomy_%26_Astrophysics_pdf5_systemF
The separation between cpp files and .h files was only necessary for
the monolithic build feature.
Moreover, try to move the stuff in lyx_SOURCES to liblyxcore_a_SOURCES.
Two issues though:
- Box::contains creates a link error, so put all the code in Box.h
- Compare::Compare is an issue too, and I am not sure how to fix it.
For now, it is thus kept in lyx_SOURCES.
Moreover, version.{cpp,h} are moved to lyx_SOURCES to avoid rebuilding
liblyxcore.a at every commit.
I do not know until which version gcc has complained when trying to
mute a warning that it does not support. Recent versions do not do
that.
For now let's just do gcc 7, the limit should be higher.
* remove wrong (sub)namespaces. This are inserted by newer versions of
designer, although they are wrong
* add vertical spacer
* make horizontal spacer expand
If the url contains 'ctan', try to determine the correct url depending
of the components to deside if using 'https://www.ctan.org' or rather
https://mirrors.ctan.org.
Without 'ctan' try to check https:// instead of ftp:// or http://
The order is not user-friendly, but correct. Before this patch, the elements in the UI didn't correspond to the implemented behaviour. The problem is the order: the first element is always MathML, the second HTML, etc. I think we should rather use some `userData` in QComboBox items to decouple the order from the semantics.
It will allow to merge a lot of code for MathML and HTML: same parsing, slightly different outputs. The refactoring also shows clearly what parts are missing in HTML (while being easily fixable). The change is a prerequisite for implementing font nesting (although it is not done yet).
The new class is not yet used within InsetMathFont; it will come in the next commit.
Open question: should we use InsetMath::mode_type somewhere? I believe so, to determine whether the current default is italics or not in MathML.
