On 25-Aug-05, at 5:27 AM, Jesse Welton wrote:

> On Wed, Aug 24, 2005 at 07:05:43AM -0700, John M McIntosh wrote:
>
>> The move to quartz is not only for throttling issues, but also to
>> address code changes required for the macintel version.
>>
>
> I understand that.  I asked what triggers throttling problems because
> at this point I'd rather stick with an older VM, as long as I'm
> unlikely to experience problems from throttling.  I haven't noticed
> any, so I wonder under what conditions they occur.  If I'm not going
> to trigger them, I have little reason to update other than to test the
> new VM.

Once you go over 50 frames per second. This is evident if you run say  
a 3.8.5b1 VM or lower and try the
hilbert code. The other trigger is when the morphic caching logic  
encounters an issue such that it flushs
the screen on every draw event which means a flush for each pixel  
draw. This bug was I believe recently
fixed and relied on an interaction with the debugger to trigger.

Karsten Wolf had done some work with flush/cache logic in the VM a  
year or so back, I'm reviewing that
to see if we can say flush always, and revert to caching if the flush  
rate goes over 50 fps.

The crash Tony got is of more concern since I don't see the obvious   
thing I am doing wrong to trigger it.

--
======================================================================== 
===
John M. McIntosh <johnmci at smalltalkconsulting.com> 1-800-477-2659
Corporate Smalltalk Consulting Ltd.  http://www.smalltalkconsulting.com
======================================================================== 
===