When I was playing with remote display, implementing dirty regions
correctly made a *big* difference.  Without them, running Morphic
"remotely" on a 300 MHz K6-2 was slow.  "Slow" as in, you press the
button, and it takes a few seconds for the world menu to pop up.  :) 
After implementing it correctly, it "only" took like .75 seconds.  :)


(By the way, I assume we are talking about the mechanism where only the
changed portions of the screen need to be redrawn, instead of redrawing
the whole screen....).

Once you have it all working, try opening a few BouncingAtomsMorph's and
experience the depression all over again.

-Lex




"Jay Carlson" <nop at nop.com> wrote:
> Jay Carlson [mailto:nop at nop.com] writes:
> 
> > I'm not using Morphic.  Until I can get this stupid ~2.2M transfer limit
> > nailed, I don't have any images I can show off---and you can't point and
> > laugh until then :-)
> 
> Well, I finally managed to shrink a Morphic-containing image below 2.5M and
> got it transferred.
> 
> So now I'm pointing at my Helio and laughing....
> 
> First of all, it really is slow.  No, really.  Multiple seconds between
> tapping the screen and the world menu coming up.  Of course, I've released
> the pen by then, and it all goes away and I have to repeat.  I guess I
> should implement dirty regions in ioShowDisplay() now, but even then I see
> over a hundred calls to ioShowDisplay() as the menu comes up :-(  Is update
> coalescing/delaying mandatory for VMs planning to run Morphic?
> 
> Second, unlike MVC, Morphic doesn't try to fit its menus to screen size.
> And browsers are totally busted unless they get enough room to open.  Go
> ahead, resize a morphic world to like 160x160 and then open a browser, try
> it....
> 
> Third, you really do need more than the red button to live in Morphic and I
> don't have enough  modifiers yet.
> 
> > Anyway, consider this another request for the new shrunken Morphic-like UI
> > replacement mentioned in the OOPSLA reports.  I could really use it.
> 
> :-)
> 
> Jay