CSL's ThreadedFrameStream (was Re: [CLAM] RT and port sizes (was Re: AudioPorts Usage))

[Stéphane Letz]

>
>
> Stephane,
> Now that you mention this disk-access-in-RT technique, do you
> (or other people listening) have any good pointer to documents
> on that kind of technique? I'm interested in it for my research.
> And not only restricted to disk but in general, which and how
> blocking operations are to be avoided. And other RT patterns.

  I don't have "paper" reference but for code reference you can look at:

- capture_client.c in jack source example-clients folder

- in ardour I guess

- our own audio library : LibAudioStream  at http:// 
libaudiostream.sourceforge.net/

Stephane

>
> Going back to clam, the point is that disc access in clam is not
> RT safe (it just blocks), so we will have to fix that, at some
> point.
> Of course, when you do callback-based clam apps, you usually
> don't need accessing the disk, so it is not a very big problem...
>
> Xavier, I've had a look at the code of CSL for the first time (BTW,
> nice naming convention). It seems interesting though I need
> to clarify general ideas on their approach. I'll try with their
> ICMC2003 paper.
>
> Thanks!
> Pau
>
>
>
>