Re: a joint letter on low latency and Linux

[Paul Barton-Davis]

>Perhaps an intresting point is: Linux isn't very useful as a professional
>music playform without ALSA and the kernel doesn't yet include Alsa, so

SuSe and Mandrake come with ALSA, I believe. 

>people wanting to do serious music stuff are already going to have to be
>running a non-standard kernel (or at least a kernel module), is it that
>much more to ask them to install a low-latency patch until a clean enough
>set of solutions is found?

Well, its not really the same thing. Replacing the kernel that came
with a distribution is a potentially huge undertaking. Installing a
modular device driver is quite different. Of course, the kernel
replacement could be very straightforward, but there are few
guarantees of this.

There is also the problem that I've mentioned before: if we want to
convince companies like Steinberg, Emagic and the rest to port their
stuff to Linux (*), they need to know that "stock Linux" has the kind
of performance characteristics that make this worth their while. They
already operate in a very niche market, and the diversion of resources
needed to do a port would be hard to justify when nobody but a person
prepared to patch the kernel source and reinstall a kernel could use
the result.

--p

(*) and we *do* want them to do this. for the record, i spend a huge
    percentage of my time writing apps for pro audio, and in so doing,
    it has become easy to recognize the enormous amount of work that
    these applications represent, not to mention the domain knowledge.
    Although I'd love to see a bunch of GPL-using folks writing
    replacements for Nuendo, Acid, Logic, Digital Performer, ProTools,
    GigaSampler and a host of other current commercial software
    systems, my rational self tells me that this simply is *not* going
    to happen on any reasonable time frame. The number of people with
    both the time and motivation, plus the technical expertise to
    write this code and the technical understanding of the domain is
    probably numbered below 500 worldwide, including all those people
    who already work for commercial companies. It might not even be
    that high.