Learn how easy it is to sync an existing GitHub or Google Code repo to a SourceForge project! See Demo
Related tickets: #3474765, #3474534, #3480246
Please merge rev# 20941 and 20969 into 4.5.x to display correct characters for font substitution.
Thanks to Max Funk!
Perhaps also rev21212, related to ticket 3488967 but also related to font glyph substitution can be included in this ticket, or do I need to make a new ticket for that?
I was considering about merging further series of fixes for font
Here are my conclusions:
This is the minimal fix to make font substitution work on code
points instead of code units, and resolve dependency for r21259 in
merge request #3498696.
So this one should be merged separately.
This avoid font substitution for control chars which could result
in improper glyphs with some buggy fonts which says it can display
such control chars. This is a workaround for a minor problem.
So this one should be merged separately, not to complicate others.
However, since this one is nominated in this merge request, I'll
accept it but in separate revision.
* Above here, svn merge just works.
This is a quality improvement for the cases where both an auxiliary
font and the default font can display chars after first char which
triggered font substitution.
But this fix depends on r21205 which requires manual resolution of
merge conflicts. This also means that merge request for r21212 will
need a backport branch.
I may post a merge request later if I can resolve the conflict
cleanly. I think this is not required to be in 4.5.1.
From my last comment,
> So this one should be merged separately, not to complicate others.
> However, since this one is nominated in this merge request, I'll
> accept it but in separate revision.
I turn down this decision about separation. The entry in CHANGES.txt;
"Wrong chars in auxiliary fonts" told me that these can be said as a
single fix for users. It's enough to have separate revisions in trunk.
Reviewing the diff of r20941, I noticed that the quality problem which
was fixed in r21212 was a regression since r20941, which had
accidentally dropped the check to switch back to the default font.
Then we can't accept r20941 without the fix in r21212.
Now it turns out that the whole revisions, r20941,20969,21205,21212, are
linked in a single chain of dependencies.
I'll post a merge request for these with a backport branch. This request