> Don't variation selectors appear *after* the character they apply to?
Markus Kuhn:
MK> Yes.
> If so, they'll be a pain to implement in a terminal emulator.
> (Remember that unlike PuTTY, XTerm doesn't use a backing store, and
> modifying already existing characters will cause flicker.)
MK> May be, it's time xterm buffers strings as well a bit before
MK> outputing them.
XTerm's got a backing store, and the code for delaying refresh when
under load is already there (that's what disabling ``smooth
scrolling'' does). The only thing that remains to do is to work out a
suitable heuristic about when exactly to delay a refresh (Emacs 21 is
too aggressive about that IMHO)... and implement millisecond-reso-
lution select on the 2^27 platforms that we support.
Sorry, I'm too busy playing with Cedilla.
MK> In any case, it's not worse then combining characters and other
MK> ligature substitution.
Except when doing Vietnamese (which Cedilla fully supports in the next
version) or Polytonic Greek (which is almost there -- I only need to
implement rotations in order to get a pretty glyph for Dasia),
combining characters don't happen all that often. On the other hand,
when using e.g. luit, there will be circumstances under which width
selection will happen at every bloody character.
MK> In the end, with Unicode, you'll always need to be prepared to
MK> look ahead in the character stream and undo a glyph that you have
MK> written at the end of a read() string.
Which doesn't mean you actually want to undo your work at every
character. IMHO, Unicode f*cked up by putting combining characters
after the base ones (and one of the reasons Cedilla works so well is
that it inverts this convention on input), why do you insist on
furthering this?
Peter Anvin (offtopic):
PA> Not to mention that xterm is notorious for stray pixels for
PA> various reasons.
As far as I am aware, this should have been fixed a good few years ago
in XFree86 XTerm (but not necessarily in the SI version). If you
still see this problem (using current XFree86 XTerm), please send a
reproducible bug report.
Oh, and that should not happen with Cedilla.
Juliusz
P.S. Did I mention Cedilla?
--
Linux-UTF8: i18n of Linux on all levels
Archive: http://mail.nl.linux.org/linux-utf8/
