To: perot@pallas.amp.uni-hannover.de
Subject: Re: The $c9 Function                                                            


On Fri, 3 Jan 1997 10:18:55  Peter Rottengatter <perot wrote: 
>
>I exchanged a few mails with Andreas, and he basically confirmed all : It 
>is in all machines, it's function is indeed the same as appl_yield, and 
>it indeed does not need any GEM arrays. Andreas has a binding for MagiC 
>that is
>
>_appl_yield:
>        move.w   #$c9, d0
>        trap     #2
>        rts

Excellent !
This is what I was hoping for.

>So he even chose almost the same name as appl_yield. What he did not say 
>anything about is whether there are any official documents. He is a 
>little strange on this subject, you need to very carefully ask him about 
>documents that are not by himself. I'll continue to do so.

Good, please do !
Since Andreas has not felt bound to maintain compatibility to undocumented
TOS features in other areas, I take this as a strong indication that he has
documentation on this function.


>He was very enthusiatic about the performance of this call. He said this 
>function is one of the best means to demonstrate the superior performance of 
>MagiC.

Hmmm, does he mean that programs which regularly call this cooperate better
with each other, than programs that rely on preemption can...?


>Summarizing I'd say this is the function to go for.

Yes, I think so too.  But I'd be much relieved to hear that there is also some
official documentation for it, because otherwise we cannot rely on its being
implemented in the other new AES currently under development.

-------------------------------------------------------------------------
Regards:  Ronald Andersson                     mailto:dlanor@oden.se
                                               http://www.oden.se/~dlanor
-------------------------------------------------------------------------
