Re: [nas] nas and adobe's libflashsupport

[Jon Trulson <jon@xxxxxxxxxxx>]

On Tue, 21 Apr 2009, Paul Fox wrote:

> i've implemented the NAS client code for libflashsupport.so,
> using some minimal client code i stole from mpg123 as a template.
> it seems to work, mostly (see below).  if i install my
> libflashsupport.so in /usr/lib, and run firefox with the a couple
> of env variables which force it to choose NAS rather than
> something else (the library has a probe and search mechanism that
> will usually find native sound before networked sound), then my
> audio goes to nasd.
>
> i say "mostly", above, because i've tested locally, and between
> several pairs of machines, and there's one nasd server that
> reliably dies with "explicit kill or server shutdown", and the
> server dies:
>    Apr 21 14:53:12 jade kernel:  [262066.900419] nasd[1046]:
> 	segfault at 16 ip b7e37a73 sp bfd4caa0 error 4 in
> 	    libasound.so.2.0.0[b7dea000+c3000]
>
> all of my nasd servers are 1.9.1.  i guess i'll have to build
> a copy and debug.
>
> if anyone would like to try my libflashsupport.so, please let
> me know, and i can send you either a binary, or a patch against
> http://tcosproject.org/git/libflashsupport.git

  That's pretty interesting - especially since you took a stab at it
  :)  I'd love to see both the patch and the lib so I can try it out
  here.

  As for the crash above - this seems... interesting.  nasd uses OSS,
  however the crash listed above happened in libasound, which I
  believe is an ALSA lib...?


> paul
>
>
> i wrote:
> > just curious -- has anyone considered adding NAS support to the
> > flash plugin that adobe provides?  the intent of their plugin is
> > to allow support of alternate audio output devices and formats,
> > and i have to say, the API is pretty simple -- the client only
> > has to provide open, close, and "latency" routines, and is
> > provided with alloc, free, and "fillbuffer" routines which need
> > to be called as appropriate.  and there are implementation
> > examples for both oss and alsa.  audio samples are 16 bit stereo
> > at 44100.
> >
> > the plugin source from adobe is linked to from here, if you're
> > curious:
> >
> > http://labs.adobe.com/wiki/index.php/Flash_Player:Additional_Interface_Support_f
> > or_Linux
> >
> > i haven't tried to verify that that's the most recent version
> > available from adobe, but i did find this version which adds
> > pulse and esd support:
> >     http://tcosproject.org/cgit/cgit.cgi/libflashsupport.git/
> > in fact, i just realized that this is the source for the
> > debian/ubuntu version, so it's clearly the place to start.
> >
> > having now lived with alsa and pulse for remote audio for about a
> > year, i'm about ready to throw in the towel and switch back to
> > nas, which was far more reliable, far more efficient, and also
> > supports software mixing.  what it lacked, though, was reliable
> > support for audio from my browser, and this could fix that...
> >
> > paul
> > =---------------------
> >  paul fox, pgf@xxxxxxxxxxxxxxxxxxxx (arlington, ma, where it's 55.8 degrees)
>
> =---------------------
> paul fox, pgf@xxxxxxxxxxxxxxxxxxxx (arlington, ma, where it's 55.2 degrees)

--
Happy cheese in fear                 | Jon Trulson
against oppressor, rebel!            | mailto:jon@xxxxxxxxxxx
Brocolli, hostage.       -Unknown    | A828 C19D A087 F20B DFED
                                     | 67C9 6F32 31AB E647 B345