[CM] rts

Rick Taube taube@uiuc.edu
Fri, 31 Mar 2006 14:08:04 -0600

well i dont mean give up TOTALLY, just (possibly)  try a different  
solution. trying to do it all in 100% lisp is really hard, its like  
working with one hand tied behind your back. as it stands rts seems  
to work really well only in openmcl.

maybe moving the rts scheduler to C and using callbacks to eval  
musical processes that get registered from lisp is a better way to  
go. im trying run some tests right now as a mater of fact. if it  
worked there would be some huge advantages:

-rts could run in every cffi/common lisp and any scheme with callbacks
-have complete control over process pioroty etc
-give conless scheduling in lisp and no boxing/unboxing doubles

since we already have a good model for rts scheduling it probably  
isnt hard to move it to C.  of course my c abilities are really poor  
-- it took me 2 hours to figure out how to do a Makefile that creates  
a .dylib :/ but as of 1 hour ago i was able to sprout my first  
ptheads in cmucl sbcl and openmcl on darwin and get rts time from c.  

> Hi Rick,
> Am 31. März 2006, 07:10 Uhr (-0600) schrieb Rick Taube:
>> do you have rts loaded as well? that might be a problem -- im about
>> ready to give up with rts in sbcl.
> Please don't do that! I really need rts and sbcl working. It does seem
> to work here (with cm 2.7.0) and Ralf Mattes is going to help me
> implementing a pdsend class for integrating cm with pd which would
> solve about anything I could wish for in the moment...
> --
> Orm