Very early in Mstation's days we interviewed Jaroslav Kysela about
the ALSA project. This project set out to upgrade the existing free
kernel sound drivers in significant ways to match up with modern
soundcards and modern ways of working with sound. Since we last
spoke a lot of work has been done. The drivers are now reasonably
easy to install and the number of cards supported has expanded
dramatically. Back then there were no high-end cards supported and
now there are a few.
ALSA is quite widely used now I think. Do you have any numbers
for users at all?
It is difficult to determine real numbers of users. We have more than
ten thousand downloads per months from the main site. I am very glad that
a major part of the Linux community knows about ALSA at this time.
When we last talked the main thing coming up was integrating
ALSA with the 2.3 kernel. There have been a few twists and turns
in the meantime but how is that job going now?
Right, because we were not sure about the API and the situation for the kernel
integration was not so good at 2.3 time, we postponed our goal.
Unfortunately, although all sound related developers agreed on ALSA
integration into 2.5, the process is not as fast as we wish. Linus is busy
with the new block I/O layer, so we have to wait awhile again.
Has that process been a consultative one or do the ALSA
project coders mainly have to react to changes?
We accept reactions from kernel developers to improve and integrate our
code into the main Linux tree. It is bi-directional talk. Unfortunately,
due to lack of time for responding kernel developers, the communication is
spare right now. The fact they are not able to verify our APIs at the
moment worries me a bit.
The ALSA soundcard matrix is looking very impressive these days with
quite a few reasonably high-end cards on it. Did you have to persuade
most of them that Linux drivers were a good thing or did they come
looking for you?
I must admit, that manufacturers for all supported high-end cards are
very helpful in the sense of hardware availability and documentation. On
the other hand, some other manufacturers still deny the open source world
from being happy with commercial platforms. I think that there is only one way
to change this situation - users should buy only "open source" hardware
for their Linux machines and then when the user base is large enough,
I am sure that hardware vendors will change their thinking.
What points would you make in trying to persuade card maker/marketers
to release their specs so that Linux drivers can be made?
We have very nice document available at
http://www.alsa-project.org/call.html which explains reasons for hardware
vendors to support open source software.
Is there anything major coming up for ALSA in terms of features and
Right now we do not expect more rapid changes in our driver / library
framework. We are happy, and we offer quite stable drivers for audio and
music hardware. Also, the presence of projects like JACK or LADSPA - which
simplifies the development of applications working with digital audio and
allows easier interoperability among different pieces of code, creates a
good starting point for many good sound applications.
Our goal is to finish the documentation for application developers and
improve ALSA library to let some parts be more tuneable. Also, the support
for wave-table hardware is not finished. And, in our minds are projects
like the sharing of exclusive devices (mainly for digital-audio) and so on.