Re: OSSv4 on OpenBSD

2009-05-26 Thread Stuart Henderson
On 2009-05-25, Daniel Gracia Garallar danie...@electronicagracia.com wrote:

 Sure and additional framework should make easy porting other projects to 
 OpenBSD, but as far as audio programming is related, native audio 
 support is nicely implemented and rock solid.

most of the applications which already have some way to use multiple
audio backends are reasonably straightforward to port to sndio - there's
quite a lot of example code in the ports tree.

 Just missing some samplerate convert not relying in aucat! So I can use 
 it on several devices at once, but that's a patch -filtering is the hard 
 trick- I'll work into :)

keeping samplerate conversion out of the kernel was a deliberate decision;
you can run multiple copies of aucat with different sockets/devices.



Re: OSSv4 on OpenBSD

2009-05-25 Thread Alexandre Ratchov
On Sun, May 24, 2009 at 07:48:27PM -0400, Predrag Punosevac wrote:
 A friend of mine who is an avid NetBSD user kept complaining about how
 bad is audio on NetBSD. After getting sick of hearing complains, 
 I asked on OSS mailing lists about OSSv4 support for NetBSD and OpenBSD.
 I actually got a very interesting answer 
 
 http://www.4front-tech.com/forum/viewtopic.php?t=3133
 
 I recall OSS being discussed on this mailing list after OSS went 
 open source and changed the license. Can Jake or any other developers 
 in charge of audio on OpenBSD explain the issues involved in porting 
 OSSv4 to OpenBSD? 
 

Jacob explained why the job is technically hard for OSS
devs. Also since ports don't use the OSSv4 API, it would be
of limited usefulness.

 I personally have fantastic experience with our audio but I would 
 think that OpenBSD could benefit at least from extra audio drivers.
 Am I very wrong? Sorry for the noise.
 

in theory it could be possible to borrow certain OSS drivers and
to tweak them to become part of our kernel. But:

 - if the sound card is not documented (OSS uses NDAs for
   certain drivers), we would take code that we don't
   understand and that we're unable to maintain, leading to low
   quality code. By doing this we indirectly support
   development model using NDAs, which is wrong. Not motivating
   for me.

 - for documented sound cards, we often already have
   drivers, and given the difference between our kernel
   internals and OSS internals, it's easier to just write
   our own driver.

IMO the most important is to have quality support for _all_
integrated audio devices (azalia, ac97, cmpci) so audio just
works and support few well-identified professional cards
(envy, usb) for advanced applications. It's more about
quality than supporting a lot of devices.

-- Alexandre



Re: OSSv4 on OpenBSD

2009-05-25 Thread Daniel Gracia Garallar
Actually, when audio is a concern, I'm quite happy with the audio(4) 
framework of sio_open(3) and friends.


I've just finished a remote PMR control app where real-time audio is 
needed, and all the bells and whistles are up to the task: multiple 
devices support -I'm working with four Behringer USB audio cards-, 
full-duplex, mixer control et al.


Sure and additional framework should make easy porting other projects to 
OpenBSD, but as far as audio programming is related, native audio 
support is nicely implemented and rock solid.


Just missing some samplerate convert not relying in aucat! So I can use 
it on several devices at once, but that's a patch -filtering is the hard 
trick- I'll work into :)


Regards!

Dani

Jacob Meuser escribis:

On Sun, May 24, 2009 at 07:48:27PM -0400, Predrag Punosevac wrote:

A friend of mine who is an avid NetBSD user kept complaining about how
bad is audio on NetBSD. After getting sick of hearing complains, 
I asked on OSS mailing lists about OSSv4 support for NetBSD and OpenBSD.
I actually got a very interesting answer 


http://www.4front-tech.com/forum/viewtopic.php?t=3133

I recall OSS being discussed on this mailing list after OSS went 
open source and changed the license. Can Jake or any other developers 
in charge of audio on OpenBSD explain the issues involved in porting 
OSSv4 to OpenBSD? 

I personally have fantastic experience with our audio but I would 
think that OpenBSD could benefit at least from extra audio drivers.

Am I very wrong? Sorry for the noise.


audio(4) and all the current audio drivers would need to be
modularized to not conflict with OSSv4.  OpenBSD doesn't use
modules by default, so users who would want to use OSSv4 would
be running an unsupported system.

I have tried taking small bits from 4Front drivers (for cmpci(4)
and azalia(4)), but it has not been very helpful, for various
reasons.  I've learned more by looking at FreeBSD and ALSA drivers.

some of the 4Front drivers were developed under NDAs, so the only
documentation available to us is the driver source.

having 2 vastly different audio APIs is not helpful, at all.
arguably, OSSv4 would be a third (or fourth even) audio API that
we would be supporting, as OSSv4 is different than OSSv3, which
we already support with ossaudio(3).

even though OpenBSD and NetBSD share the same basic audio code,
there are numerous differences, starting with aucat(1) and
sio_open(3) and going all the way down to the low level drivers.
it appears this diversion is going to continue.  I've tried
sending patches for simple bugs azalia(4) to NetBSD devs that
never got acted on, and they have a GSoC project to add support
for stream mixing in the kernel.




Re: OSSv4 on OpenBSD

2009-05-24 Thread Jacob Meuser
On Sun, May 24, 2009 at 07:48:27PM -0400, Predrag Punosevac wrote:
 A friend of mine who is an avid NetBSD user kept complaining about how
 bad is audio on NetBSD. After getting sick of hearing complains, 
 I asked on OSS mailing lists about OSSv4 support for NetBSD and OpenBSD.
 I actually got a very interesting answer 
 
 http://www.4front-tech.com/forum/viewtopic.php?t=3133
 
 I recall OSS being discussed on this mailing list after OSS went 
 open source and changed the license. Can Jake or any other developers 
 in charge of audio on OpenBSD explain the issues involved in porting 
 OSSv4 to OpenBSD? 
 
 I personally have fantastic experience with our audio but I would 
 think that OpenBSD could benefit at least from extra audio drivers.
 Am I very wrong? Sorry for the noise.

audio(4) and all the current audio drivers would need to be
modularized to not conflict with OSSv4.  OpenBSD doesn't use
modules by default, so users who would want to use OSSv4 would
be running an unsupported system.

I have tried taking small bits from 4Front drivers (for cmpci(4)
and azalia(4)), but it has not been very helpful, for various
reasons.  I've learned more by looking at FreeBSD and ALSA drivers.

some of the 4Front drivers were developed under NDAs, so the only
documentation available to us is the driver source.

having 2 vastly different audio APIs is not helpful, at all.
arguably, OSSv4 would be a third (or fourth even) audio API that
we would be supporting, as OSSv4 is different than OSSv3, which
we already support with ossaudio(3).

even though OpenBSD and NetBSD share the same basic audio code,
there are numerous differences, starting with aucat(1) and
sio_open(3) and going all the way down to the low level drivers.
it appears this diversion is going to continue.  I've tried
sending patches for simple bugs azalia(4) to NetBSD devs that
never got acted on, and they have a GSoC project to add support
for stream mixing in the kernel.

-- 
jake...@sdf.lonestar.org
SDF Public Access UNIX System - http://sdf.lonestar.org