Subject: Re: [linux-audio-dev] ok, jump on slashdot right now, ALSA story in progress
From: Joshua Haberman (joshua_AT_haberman.com)
Date: Sat Nov 02 2002 - 08:46:25 EET
Paul Davis <paul_AT_linuxaudiosystems.com> wrote:
> i don't want to blow my moderator points this time, so get busy. there
> is lots of confusion, mis-information and general cluelessness in the
> responses so far. get to it, provide me with some stuff to moderate
> upward ...
What a coincidence: I've had brewing in my mind in the last two days an
ALSA usability project. It would be pointless to post to /. about it,
so let me braindump here since I could actually get informed replies. :-)
Here are my primary goals:
1. Work on getting discover [0] to work with ALSA. discover is a project
developed by Progeny, designed to detect what hardware is available
and provide that information to userspace programs via the library
libdiscover. It works reasonably well, but at this point only supports
OSS. I wrote to the mailing list yesterday asking questions about
how ALSA could be supported [1].
I believe this is a very promising project: here is an example of
the output on my machine:
$ sudo discover all
Silicon Integrated Systems (SiS) SG85C503
Unknown LITEON DVD-ROM LTD163D
Silicon Integrated Systems (SiS) SiS900 10/100 Ethernet
Realtek Semiconductor Co., Ltd. RTL-8139
Silicon Integrated Systems (SiS) 5513 [IDE]
Linux IDE-SCSI emulation layer
Creative Labs SB Live! EMU10000
Silicon Integrated Systems (SiS) 7001
Silicon Integrated Systems (SiS) 7001
Silicon Integrated Systems (SiS) SG86C201
NVIDIA Corporation GeForce [NV11]
Unknown WDC WD1200AB-00CBA1
Unknown ST360021A
$ sudo discover --module all | sort -u
emu10k1
ide-scsi
rtl8139
sis900
usb-ohci
2. Modify the ALSA drivers to put something useful in the card ID. Right
now you get card0, card1, etc., which is completely useless: if you wanted
to refer to the card by number, you'd just write hw:0.
Why is the card ID important? Because it can be used to enumerate and
select cards, the way other audio APIs will let you do! Using the syntax
hw:CARD=cardid you can select a device based on human-readable strings.
Today I get:
$ cat /proc/asound/cards
0 [card0 ]: EMU10K1 - Sound Blaster Live!
Sound Blaster Live! at 0xcc00, irq 11
1 [card1 ]: ICH - SiS SI7012
SiS SI7012 at 0xd800, irq 11
My goal is:
$ cat /proc/asound/cards
0 [SBLive ]: EMU10K1 - Sound Blaster Live!
Sound Blaster Live! at 0xcc00, irq 11
1 [SiS SI7012 ]: ICH - SiS SI7012
SiS SI7012 at 0xd800, irq 11
It is possible to set these with arguments to modprobe, but that's no
reason why they shouldn't be useful by default.
3. Documentation on how to write user-friendly device selection for ALSA.
If common apps started getting this right, ALSA would provide a directly
visible value-add for end-users, since OSS can only let you choose between
/dev/dsp1, /dev/dsp2, etc (AFAIK).
Alternatively, encourage people to use PortAudio since it will provide
this transparently.
Alternatively, investigate the possibility of being able to run more
than one jackd at a time, one for each sound card. Running a single
jackd for more than one sound card is infeasable since two cards can
never be perfectly synced. The point of this being that the user would
again see the advange of being able to easily select the desired sound
card for output, with the added advantage of being able to share the
sound card.
I believe that these things alone would make ALSA *much* more usable for
simple, stereo users that perhaps have more than one sound card. The
hw:0 stuff is a drag, as is trying to find out what module goes with your
card and modprobing it.
I'm interested in input from multichannel users regarding what kind of
usability improvements ALSA could use in this arena. To be honest, I have
no idea how ALSA works with multichannel cards since I don't have any.
Comments welcome.
Josh
[0] http://hackers.progeny.com/discover/
[1] http://lists.progeny.com/archive/discover-workers/200211/msg00000.html
-- "Only Angels can sing that high, and only dogs can hear it." -- James Evans
This archive was generated by hypermail 2b28 : Sat Nov 02 2002 - 08:43:47 EET