Crash-a-lot just became critical

[Dr. Hawkins]

On Thu, Jan 8, 2015 at 2:04 PM, Richard Gaskin <ambassador at fourthworld.com>
wrote:

> Dr. Hawkins wrote:
>
>  I've been living through the frequent random crashes...
>>
>
> Computers are deterministic systems:  consistent inputs will yield
> consistent results.
>

So they would like us to believe :)

>
> If something may appear to be random, it merely suggests we haven't found
> the recipe for reproducing it yet.
>


Some pseudo-random events are as good as random for such purposes . . .

And most  people are unfamiliar with Bohr's response to Einstein's dicing
comment . . .


> All problems with computers can be resolved by identifying the differences
> between the working and non-working states.
>
> Your filing a bug report is helpful (#14350 for those wishing to help
> triage this), but the crash log may not be enough for the team to identify
> what went wrong.
>
> It might, though, and if so then we have nothing more to worry about.
>
> But if this issue is important to you (and it would seem that it is), you
> may be able to get what you want more easily by including in your bug
> report a description of what your program is doing when it crashes.
>

Just moving a mouse.  Err, fingering a trackpad.


>
> If this is really important, you may consider adding some script logging
> to your stack so you can at least know the last handler that executed
> successfully.
>
> If you haven't yet played with the messageMessages property and the
> messageHandled message to make such a log, I can provide a simple script
> for you to log all handler calls to a text file.



I'd appreciate that--but I suspect that it won't catch this.  This isn't
livecode responding to an error, but failing to do so in a way that it
doesn't trap.

It appears quite likely that the triggering event is the reference to a
non-existent control  (the wrong stack was being targeted).  In a vanilla
stack, livecode choked on the error; in a palette, execution stopped.  A
missing branch in an engine switch, perhaps?  A segfault?

I'm willing to believe that the times that it happened "after" I clicked a
button were likely already happening from moving the mouse to a location.
If that's the case, mouseEnter messages would seem to be involved in every
failure.

-- 
Dr. Richard E. Hawkins, Esq.
(702) 508-8462