DJ wrote:
> I would use Waba more if it didnt run dog slow on my OS 2.0 120. No
> offense to the Waba team, you've done a remarkable job, but you should
> see how slow the suck-o-meter is on this thing.
I have a 120 here at my office which I test Waba on. Yep, it's dog slow
on 'em. Not too fast on an eMate either. Hey, but it works! Will we be
able to speed things up? It remains to be seen.
There are two reasons for the slowness. First, the Newton has very poor
access to low-level graphics. To draw a line, Waba must ask
*NewtonScript*, another interpreter (!), to create a line object, then set
up the graphics context, then draw the line. NewtonScript of course in
turn asks the low-level C++ QuickDraw engine to do it. All this
handshaking is very very expensive. The result here is that Waba for the
MP2000 is slower than Waba for the Palm Pilot, at least when it comes to
drawing. How do we fix this? We can't do much about that until Paul or
someone else comes through with low-level access to QuickDraw. The other
option is to write Waba apps that use real NewtonScript widgets, buttons,
etc., rather than scribbling Palm-style ones on the screen. Waba can do
that already, but most Waba apps are written for the Palm and WinCE, and
not with the Newton in mind.
Second, it's interpreted code. On the Palm side of things they're working
on a new project, called WabaJump, which compiles Waba bytecode directly
to Motorola DragonBall machine code for the Palm Pilot. Runs much faster!
Do I have the wherwithall to write a bytecode->ARM compiler? Heck no.
But Palm has just announced that they're jumping to StrongARM, with the
very distant possibility that WabaJump might just write the compiler for
me. :-) We'll have to see. In the mean time, if you want to speed
things up, join the development team and see what you can do to clean up
the Waba code.
Sean
-- This is the Newtontalk mailinglist - http://www.newtontalk.net To unsubscribe or manage: visit the above link or mailto:newtontalk-request_at_newtontalk.net?Subject=unsubscribe
This archive was generated by hypermail 2.1.2 : Sun Sep 09 2001 - 19:50:28 EDT