On 13-Mar-01 at 17:46, Alexander Mai ([EMAIL PROTECTED]) wrote:
> On Tue, Mar 13, 2001 at 06:12:06AM -0500, Rick Scott wrote:
> > Is this a test?? A key is being pressed while a menu is posted. This key
> > is a mnemonic for somthing, and we are therefore trying to Activate it.
> > Problem is that this sub-class of Primitive, whatever that is, does not
> > have a class ArmAndActivate proc. At first glance, we should not be making
> > the call unless (*act) != NULL.
> >
>
> The interesting part is why does this happen _sometimes_ and
> not very often. I couldn't find nearly a pattern of reproducing this.
>
> I even don't remember when/how I managed to locate this piece
> of code where I added this fprintf(). I'll have a look at the core
> dump now.
If I remember correctly, nedit does some "interesting" things with
menus/mnemonoics/accelerators to get the behaviour that they want. It still
seems like a bad thing to assume that a Primitive sub-class has an
ArmAndActivate proc. We should be checking the (*arm).....
>
> BTW, do we have any remaining memory leaks in our simple tests?
> test/Xm/failed.known doesn't seem to be up-to-date anyway
> (would you check in a new copy of this?)
>
> --
> Alexander Mai
> [EMAIL PROTECTED]
>
>
