Re: Persian joining and Pango.

["Stephen J. Turnbull" <stephen xemacs org>]

Aidan Kehoe writes:

 > The thing is, we already have working RTL on TTYs! Input and display are
 > grosso modo there already on mlterm and PuTTY. Getting things working on a
 > GUI should be *easier*, especially with Pango, Uniscribe and the OS X
 > analogue available.

I think aiming at "just getting things working" is a mistake in the
long run.  It's not obvious to me that we can avoid implementing the
UBA at some level of XEmacs.  That may be a Lisp library (possibly
accelerated with C routines).  Thing is, although it may make sense to
delegate all physical redisplay to 3rd-party C libraries, many
applications will want to be able to predict the results of redisplay.
OTOH, it may be better to implement the UBA ourselves so that such
applications can be more tightly integrated with the UBA
implementation.

Cf. the bickering between eliz and larsi over Gnus summary display on
emacs-devel.

 >  > WIll the redisplay.c changes break tty display when compiled with gtk?
 > 
 > No. It’s just about possible that Stephen Turnbull, and only Stephen
 > Turnbull, has configured his TTY XEmacs to use different TTY fonts for
 > different character sets, but XEmacs is not configured to do that by
 > default, and I’m sure no-one else has.

Heh.  *Anything* can break redisplay, and not just for those who enjoy
pushing the envelope.