Andre Poenitz wrote:
On Thu, May 31, 2007 at 02:25:57PM +0200, Abdelrazak Younes wrote:
Jean-Marc Lasgouttes wrote:
"Abdelrazak" == Abdelrazak Younes <[EMAIL PROTECTED]> writes:
Abdelrazak> Jean-Marc Lasgouttes wrote: Possibly yes. I have a patch
Abdelrazak> pending for that issue that is not OKed AFAIS. Could you
Abdelrazak> try it?
I tried this one but cannot compile:
Abdelrazak> Sorry, try this one.
This seems to work
OK, as this is a fix to a crash and that it also addresses Andre' main
complaint about the memory consumption due to the boost::signal in the
Inset class I'll commit.
I just compile LyX with your latest fix and my first idea to make it
crash with multiple views was 'successful'.
To reproduce:
- create a doc with a dozen chars and an inset at the end.
Math is fine, ERT will do as well.
- place the cursor inside the inset.
- File -> New window
- in the new window (in which the first line is not fully visible
initially btw) delete a few of the non-math char
- switch back to window 1
<BOOM>
I cannot reproduce. In any case if this doesn't work, this is not at all
the signal solution which is at fault but the fixIfBroken() method which
does not covers this case. As simple as that. As I said you don't know
what you are talking about.
I maintain that LyX is not ready to have multiple views. The feature
should be disabled for 1.5.0.
Year you said that already many times. I think everybody understood.
Which is of course independant of whether this fix is correct or not. If
you have the time you could try to implement the alternative solution
proposed by Alfredo (checking inset validity from top to tip in
fixIfBroken). But I don't really understand why you think the signal
solution is bad.
Because it cannot not work.
Still no proof.
Alfredo's solution (also known as
"registered cursors") will need quite some effort, but it is at least
theoretically feasible.
Blablabla.
Abdel.
