Deferred updates/slow redraw on OS X

> I've complained occasionally about the state my images can get into  
> where there are frequent excruciating waits while each SystemWindow  
> redraws itself many times slower than normal - these are often  
> triggered by debuggers popping up but I also have images where just  
> opening a new browser window can sometimes set it off.  (See http://
> macos.tuwien.ac.at:9009/482103917.asHtml for one discussion about the  
> problem).  I can now characterize it a little better, I think: what's  
> happening is that the deferred updates flag is getting set to false,  
> and yet the WorldState still has a canvas directly on the Display  
> form.  If I disable deferred updating, and nil out the canvas on  
> WorldState, then I'm pretty sure the problem goes away (and I don't  
> notice any particular change in UI responsiveness, though I realize  
> there must be some extra copying going on).
>
> Does this ring any bells for anyone?
>
> Avi
>
>

>  > Does this ring any bells for anyone?
>
> There is a possibility that this is caused by a debugger popping up via
> a non-ui process. When this happens (which just so happened quite a
> number of times to me) you see all sorts of weirdo effects - starting
> from the debug button being in the wrong place, over redrawing stuff
> extremely slowly, up to bogus recursive errors. The cause? Interference
> of the debug process with the main Morphic UI process. The code in
> Debugger>>openOn:context:label:contents:fullView: is questionable - it
> forks off the debugger in a separate process but that process is not
> synchronized with Morphic. What that code probably *should* say is
> "WorldState addDeferredUIMessage:[...]" or somesuch. You might want to
> try this and see if it addresses your problem. I wouldn't be surprised
> if it did.
>
> Avi Bryant wrote:
> > I've complained occasionally about the state my images can get into  
> > where there are frequent excruciating waits while each SystemWindow  
> > redraws itself many times slower than normal - these are often  
> > triggered by debuggers popping up but I also have images where just  
> > opening a new browser window can sometimes set it off.  (See http://
> > macos.tuwien.ac.at:9009/482103917.asHtml for one discussion about the  
> > problem).  I can now characterize it a little better, I think: what's  
> > happening is that the deferred updates flag is getting set to false,  
> > and yet the WorldState still has a canvas directly on the Display  
> > form.  If I disable deferred updating, and nil out the canvas on  
> > WorldState, then I'm pretty sure the problem goes away (and I don't  
> > notice any particular change in UI responsiveness, though I realize  
> > there must be some extra copying going on).
> >
> > Does this ring any bells for anyone?
> >

> > Does this ring any bells for anyone?
>
> There is a possibility that this is caused by a debugger popping up
> via a non-ui process. When this happens (which just so happened quite
> a number of times to me) you see all sorts of weirdo effects -
> starting from the debug button being in the wrong place, over
> redrawing stuff extremely slowly, up to bogus recursive errors. The
> cause? Interference of the debug process with the main Morphic UI
> process. The code in Debugger>>openOn:context:label:contents:fullView:
> is questionable - it forks off the debugger in a separate process but
> that process is not synchronized with Morphic. What that code probably
> *should* say is "WorldState addDeferredUIMessage:[...]" or somesuch.
> You might want to try this and see if it addresses your problem. I
> wouldn't be surprised if it did.
>
> Cheers,
>   - Andreas

>
> Avi Bryant wrote:
>> I've complained occasionally about the state my images can get into  
>> where there are frequent excruciating waits while each SystemWindow  
>> redraws itself many times slower than normal - these are often  
>> triggered by debuggers popping up but I also have images where just  
>> opening a new browser window can sometimes set it off.  (See http://
>> macos.tuwien.ac.at:9009/482103917.asHtml for one discussion about
>> the  problem).  I can now characterize it a little better, I think:
>> what's  happening is that the deferred updates flag is getting set to
>> false,  and yet the WorldState still has a canvas directly on the
>> Display  form.  If I disable deferred updating, and nil out the
>> canvas on  WorldState, then I'm pretty sure the problem goes away
>> (and I don't  notice any particular change in UI responsiveness,
>> though I realize  there must be some extra copying going on).
>>
>> Does this ring any bells for anyone?
>>
>> Avi
>>
>>
>
>
>
>

> >
> > There is a possibility that this is caused by a debugger popping up
> > via a non-ui process. When this happens (which just so happened quite
> > a number of times to me) you see all sorts of weirdo effects -
> > starting from the debug button being in the wrong place, over
> > redrawing stuff extremely slowly, up to bogus recursive errors. The
> > cause? Interference of the debug process with the main Morphic UI
> > process. The code in Debugger>>openOn:context:label:contents:fullView:
> > is questionable - it forks off the debugger in a separate process but
> > that process is not synchronized with Morphic. What that code probably
> > *should* say is "WorldState addDeferredUIMessage:[...]" or somesuch.
> > You might want to try this and see if it addresses your problem. I
> > wouldn't be surprised if it did.
> >
> > Cheers,
> >   - Andreas
> This is fixed in 3.9alpha. A similar fix should probably go into 3.8.
> Karl