Mirrors | Updates | Feedback | Changes | Wishlist | Team
Currently, the way that
do_paint() decides where to split the
text runs it passes to
do_text() is rather ad hoc and
messy. It splits whenever there is a change to the "attributes" of the
characters (subject to
term->attr_mask), which includes
any changes to the high-order byte of the character codes. It also splits
around characters in the U+23BA–U+23BD range because the Windows
front-end needs this.
It seems to me (BJH) that we could do with some way for the front end
to inform the run-splitting code, so that
put in splits where the front end needs them, and not put in spurious splits
where it doesn't. I don't yet have any clear idea how this should work,
Audit trail for this wish.