[flext] Flext and jMax
Maurizio De Cecco
jmax at dececco.name
Tue Jun 8 17:56:16 CEST 2010
>Hi Maurizio,
>thanks for making me aware jMax is still existent!
You are welcome :->
>I'm not sure if there are any remaining traces in the flext source code, still documenting my efforts to support jMax when it was being developed at IRCAM.
>As for the port to current jMax you should be aware that i'm the only developer and i have no free capacities to work on flext other than fixing urgent problems.
That's ok. I have some time (no users means no urgent problems :) and i can write, test and contribute the needed code, if you are ok with that.
But i'll need you to answer my questions :->.
>But as flext supports only a subset of both the Pd and Max APIs which to my knowledge is practically the same as in jMax, i don't see any major problems with that.
It should be more true today that at the time, because the version in development support directly a part
of the PD APIs and the PD style for writing object methods (jMax have a different object model,
and fixed signature methods).
>There is one thing that might be problematic: I remember jMax to be Multi-CPU aware, which also Max5 is to some point. This is very hard to include into flext - in fact the Max5 port is using a thread lock to avoid problems.
No, not really; jMax separate the UI, written in Java, from the computation; the two part run as two process, possibly on two CPUs, but the engine is single threaded; i plan to add multi-thread support later, but
i think to guarantee a single threaded execution for each object (at least for the control part); anyway, we will see later.
>Are you already (very) familiar with the way how flext deals with the different platforms supported.
No, not at all, this is the problem :->
>A good starting point might be flprefix.h. Huh, i just realized there's still a definition of FLEXT_SYS_JMAX....
I'll take a look, and write back.
Thanks for answering,
Maurizio
More information about the flext
mailing list