On Thursday 18 March 2004 12:30 pm, Ned Konz wrote:
> On Thursday 18 March 2004 10:30 am, Andreas Raab wrote:
> > If you are looking into this please ignore the comments about stuff now 
being
> > obsolete - the only solution for this is to revive the VM-level interrupt
> > key handling.
>
> Or, for that matter, back out the CS <g>.

I have another idea. We could have the userInterruptWatcher poll (use a 
timeout on waiting for the interrupt semaphore), and if it's been "too long" 
since the last time that events were checked for, then do a peekKeyboard or 
equivalent (get more events and flush all the non-keyboard ones).

Then respond to the interrupt key (if any) the usual way.

I'll try this.

-- 
Ned Konz
http://bike-nomad.com/squeak/