Dov Feldstern wrote:
Martin Vermeer wrote:
On Sun, Sep 23, 2007 at 02:43:02AM +0200, Dov Feldstern wrote:
Attached find a patch which fixes this issue.
The actual fix is what's being done in the LFUN_FINISHED_* lfuns.
The fixes in cursorLeft and cursorRight aren't relevant to the
problem we're dealing with, but I'm pretty sure they are necessary
for the cases they are meant to deal with (in fact, I think that
currently the function being called is a void, although we're using
it as a return?).
I don't really understand all the implications of this change --- we
should understand those before applying. But it does fix the current
issue.
The flow seems a little strange to me:
*) LFUN_CHAR_FORWARD calls cursorRight
*) cursorRight is in charge of almost all the cursor movement,
except for the case of exiting an inset. In this case (end of last
paragraph), cursorRight does nothing.
*) if cursorRight did nothing (well, if it didn't set needsUpdate
and a few other things) then LFUN_FINISHED_RIGHT is called.
*) LFUN_FINISHED_RIGHT is where we move forward in this case, by
just advancing the cursor position (and with this fix, set the font
according to the new position).
I don't understand why everything can't be done in cursorRight, and
what the LFUN_FINISHED_* are needed for?
Dov
No, these are a bit mysterious to me too.
Do I understand correctly that this fixes only the issue you have with
ERT in a RtL setting? Have you tried in english text, to exit an ERT
at end of paragraph and continue typimg?
No, this fixes non-RTL issues as well. Plain English, \emph, etc. ---
they all seem to work correctly now: upon exiting the inset, the font is
set (in LFUN_FINISHED_*) to the font of the inset itself (not the text
inside the inset, but of the inset itself!), which in turn was already
set to the font that was active when the inset was created --- so that
effectively, upon exiting an inset, the font is set back to what it was
upon entering (which IMO is the correct behavior).
If there are no objections, I'll commit this tomorrow evening, then
we'll found out if this also has any ill-effects... I'd still be happy
if someone could clarify more about the flow, though...
Committed to trunk as r20477.
(This is not relevant to branch --- at least not in its current form ---
as it is due to the move of current_font from text to cursor, which was
not ported to branch.)
Pay attention to any funny things going with font settings now...
Dov
