Re: X fonts selection weirdness

[David Kastrup]

Kenichi Handa <address@hidden> writes:

> I installed a fix.  Please try again.

[...]

> The above should work, but I'm now thinking about
> introducing non-default-font-script-list:

[...]

I can't help feeling that we are currently stumbling without much
orientation through a maze of ad-hoc local patches and fixes.

This does not feel like it should be the intended mode of operation:
other applications have to use fontsets, too.  So I have the suspicion
that at least for some platforms, we are approaching the problem from an
angle that has not been intended by the creators of the rendering
infrastructures we are using.

It may be that the current Emacs APIs don't allow us to match our font
stuff to what is Xft's intended operation.  In that case, it might be
better to interface at a different level.  If that means that the
behavior and customization of Xft- and non-Xft rendering will differ
more than it does currently, that would appear a reasonable price to pay
in my opinion.

In any way: our current rate of creating yet another ad-hoc variable
intended to patch around the font system in a manner that no other
application does is not useful for platform integration.  And it does
not help people understand what happens.

I have no expertise at all in that area.  I just recognize a pattern
here that feels like a bad idea.  It is a pattern occuring when
something needs to get shipped, and sound judgment is suspended while
"just make it work, we won't touch this again" is the only left
priority.  But "we won't touch this again" is not what we are aiming for
here.

If doing this right needs more time and plan and thought, then we should
invest that.  Or it is going to bite us later.

Again: I lack the technical skills to judge the subsystems here.  I
might be all wrong about this.  But it feels increasingly that we are
getting complexity that will be beyond all users, even if perhaps not
all developers.

-- 
David Kastrup, Kriemhildstr. 15, 44793 Bochum