On Wed, Apr 06, 2005 at 04:22:18PM +0200, Jean-Marc Lasgouttes wrote: > >>>>> "Martin" == Martin Vermeer <[EMAIL PROTECTED]> writes: > > Martin> Is this a new feature? Or a regression fix? > > Don't now :) > > Martin> As you seem to know these things, I'm sure you also know a > Martin> really code-efficient way to do this :-) > > In math_nestinset.C::doDispatch, I see plenty of code setting cmd to > LFUN_FINISHED_RIGHT in such cases. I supposed the feature is supposed > to exist already but does not work anymore. This is probably the > reason why pageup/down is trapped in math insets (see my other post). > > I am not able to follow the logic, though.
You're in good company.
I see this code (I changed the indentation to agree with the logic):
---
} else
if (cur.posRight() || idxRight(cur)
|| cur.popRight() || cur.selection())
;
else
cmd = FuncRequest(LFUN_FINISHED_RIGHT);
break;
---
In words: move right within a cell; if I hit the end, skip to the next
cell; if I run out of cells, pop out of this inset; if any of these
succeed (or if selection is on), do nothing further.
When is LFUN_FINISHED_RIGHT ever reached? At end of document?
A second question (am I being just dense or what?): even if the
LFUN_FINISHED_RIGHT statement is reached, what does it actually do? I see
that cmd is redefined, with LFUN_FINISHED_RIGHT as its action, but it's
too late for this dispatch loop. Where is the newly built command being
dispatched?
This could explain a lot, e.g., why LFUN_END and LFUN_HOME don't manage
to leave the inset even if the cursor is at the end or start. Nothing a
simple cur.popRight or popLeft couldn't fix.
- Martin
pgpdSN36mzwZH.pgp
Description: PGP signature
