On Fri, May 06, 2005 at 08:19:06PM +0200, Andre Poenitz wrote:
> On Fri, May 06, 2005 at 04:01:26PM +0100, Angus Leeming wrote:
> > As soon as you have acknowledged the recursive nature of the current
> > update(), then something like this was bound to be needed. Well done for
> > getting to grips with it all.
>
> I understand (now) that update() might be called recursively (again).
> However, this does not serve any purpose but rather should be considered
> a bug.
>
> Maybe a simple solution would be something like
>
>
> ... update()
> {
> static bool in_update = false;
> static bool update_was_canceled = false;
>
> if (in_update) {
> update_was_canceled = true;
> return;
> }
>
> in_update = true;
> do {
> update_was_canceled = false;
>
> ... do current update stuff
>
> } while (!update_was_canceled);
> }
>
> The idea is that a nested update does nothing but inform the outermost
> update that it has to retry.I didn't try precisely that, but I did (in update) put in a test for isUpdating() and returning if true. Didn't help with the cursor up/down crash though. Might be good to have anyway. It was just inside the if (buffer_) statement. > Could anybody who is getting these 'fast typing crashes' reliably please > try that? > > Andre'
pgp5nJcXam28o.pgp
Description: PGP signature
