Jean-Marc Lasgouttes wrote:
"Abdelrazak" == Abdelrazak Younes <[EMAIL PROTECTED]> writes:
Abdelrazak> Maybe, but I don't think this is related to this
Abdelrazak> re-ordering in that particular case. FYI, it's Lars that
Abdelrazak> changed the calls to updateMetrics, not me.
I have to admit that I did not follow the corresponding threads... So
you mean that reverting the calls as they were would fix the problem?
No, I am not saying that. I was just clarifying the matter. I think
there are two solutions:
1) make sure that the coordCache is up to date at the end of any user
action. This was probably the case before my change. And the reason why
this crash is appearing only now.
2) For any user action, make sure that BufferView::pimpl::update() is
called at least once before the coorCache is used. IMHO, this will be
elegantly solved at the frontend level who knows about the metrics.
What do you propose instead of coordcache?
Abdelrazak> I'll think about it. In any case, something not global and
Abdelrazak> that is kept up to date at any time so that we don't have
Abdelrazak> to worry about that in totally non-related part of the
Abdelrazak> code.
Be careful with that. Changing this stuff is very time consuming and
we cannot afford to stop in the middle. The rewriting of the update
mechanism was already the reason why 1.4 was delayed.
And I believe also that's one of the reason of the slowness of 1.4
compared with 1.3. I think it's better to fix that issue early rather
that later. But point taken, don't worry, I am careful with that ;-)
Abdel.
