bug#72268: 31.0.50; Color emoji rendering

[Eli Zaretskii]

> From: Manuel Giraud <manuel@ledu-giraud.fr>
> Cc: rpluim@gmail.com,  luangruo@yahoo.com,  72268@debbugs.gnu.org
> Date: Thu, 25 Jul 2024 18:15:46 +0200
> 
> > More importantly, I don't think I understand the user-level meaning of
> > this.  You say "compressed color Emoji", but what does that mean in
> > practice?  Does it mean no font which supports color Emoji display
> > will ever work on OpenBSD unless the FreeType library is rebuilt with
> > PNG support?
> 
> I'm not sure myself.  "compressed color Emoji" is a copy-paste I made
> from sthen@ own words.  Maybe it is refering to the PNG compression.

Since it says "compressed color Emoji", it probably refers to the
font.  If we cannot establish whether there are any color Emoji fonts
that don't need "compressed" PNG, we should probably not mention the
"compressed" part at all, but just say that one needs FreeType with
PNG support to use any color Emoji fonts.

> >If so (but then why say "compressed" above?), I think we should say so
> >and include a bit more details about rebuilding FreeType on OpenBSD,
> >like that Reddit discussion mentions.  OTOH, if some Emoji fonts can be
> >used on OpenBSD, we should say that as well, and perhaps mention those
> >fonts, if they are free.
> >
> > IOW, PROBLEMS is supposed to describe the problems and their solutions
> > or workarounds in user-understandable terms, because that file is for
> > Emacs users, not for Emacs developers.
> 
> I am wary about adding the described workaround because it modifies the
> system's FreeType library for all applications.  And it does so by
> relying on a ports (think a 3rd party package) of libpng.  I don't know
> what would happen should someone remove the libpng package after having
> applied this workaround.  Also, as explained in this recipe, it will be
> gone after each system upgrade.

We can include the caveats in the text, so that users could make
informed decisions for themselves.

> My idea was more to record this limitation into "etc/PROBLEMS" for
> future reference.

That's not what PROBLEMS is for.  If a problem has a solution or a
workaround, we should describe them.  Maybe it's just me, but I see
rebuilding FreeType as a no-brainer, so if I were in your shoes, I
would not have hesitated to rebuild it.  I'm quite sure there are
others who would think the same.