Actually libneon probably can't be used in an app distributed through
the App Store thanks to its LGPL licensing, so eventually we either
need a replacement, like AOO, or need to bypass code that uses neon
for iOS. But for now let's keep building it...

Any LO-based app distributed through the App Store can't have
scripting or extendability anyway.

Sure, this will break the build elsewhere because of missing headers.
No big deal, I will take care of that eventually. It isn't as if there
would anybody else building for iOS anyway, as far as I know. If there
is, please make yourself heard.

Not that I recall exactly what that each one of these unit tests do
any more, and how interesting they are. But they need to be mentioned
here.

This is the file as seen on api.libreoffice.org.

Also, code in sw can be simplified under the premise that always
/org.openoffice.Office.Common/Undo/Steps > 0.

(Could easily happen when max undo steps is set to 1.)

I can't see that WriteSrc, WriteCxx WriteHxx are anything other than unused
archaic remnants of the past.

Or else it would partially break formula reference chain.

This reverts commit af70bc00.

To be replaced by a better solution.

This reverts commits 359d0fbd,
6a15a289 and
21352f88.

This change is not mature enought: needs more work

getKeysymReplacementName() returned a translated keyname, now
XKeysymToString() is called directly. This change had no negative
effect to me. I tried it with a fixed string as return value for
getKeysymReplacementName() on a german interface. INSRT is still
translated to EINFG as this is done by po files.

This cherry-picks the extra cleanups from the UNO type publishing
revert.