13 Jun
2005
13 Jun
'05
8:36 p.m.
right. so we've already established that the actual
nature of the
problem on linux is different, and much closer to the windows situation
- a huge variation in hardware types and configurations. jwz solved his
"problem" by throwing the problem away, and moving into a domain where
the problem he had wasn't solved, it was eliminated.
yes, we have established the actual nature of the problem. on linux.
the problem is: not enough apps, and what few apps they are:
'not-standard' and 'not-synced' with the distro's world view of
things.
the lesson of jwz is this, and its a hard one: if the user even has
to -think- about setting up their audio system, even just a tiny
little bit, _BEFORE_ using an app (but not necessarily during or
after), then linux will be 'behind the big guys'.
when i multi-track on OSX, i don't even think about the 10 channels
of digital i/o i'm getting from my firepod, nor do i care about the
megahertz, nor do i have to set a damn iota of it up. i just use the
app (cubase/logic/intuem) the way it wants me to use it and i feel
good about it. the OS handles the bullshit.
accept this config/usability challenge, or don't.
(y'know, maybe make the linux ISO boot CD the solution to ALSA, not
the problem...)
the competing APIs is definitely a problem. the OSS
guys continue to
refuse to accept ALSA,
WHAT OSS guys, and why aren't they paying attention to
linux-audio-dev if they're such great OSS guys? i mean, c'mon. its
weak to draw such lines in the sand, yo. show me a great OSS app,
and then show me 10 ALSA apps, before you make such conundrums out to
be the excuses they aren't!!!
and continue to promote the benefits of their API
and libraries. The layers that have been built on top of them
(PortAudio, JACK, the arts audio api, gnome-sound, etc) continue to
compete with each other in various ways.
yeah, and this sucks ass. to multitrack, depending on my app, i
could go one of 3 different ways, and only and all because some
developer decided that writing a few honky lines of code was worth
more than an e-mail or two to someone else who had already done 'the
glory work'. bah!
what is happening on linux is similar to the windows
world: multiple
audio APIs each of which serve a different purpose (windows own MM api,
ASIO, GSIF).
if you want to see a community using honky-ass technology, look no
further than the VST/ASIO crowd. man, its wonky, but whatever:
they're using it.
or didn't you notice?
OS X has a head start here because they forced
everyone,
even the email client writers, to use a callback model for audio I/O via
CoreAudio. if we could do that on linux, the biggest headaches could be
solved quickly. but as you note, we can't.
actually, what i like about the MidiShare effort is that it seems to
have been written by people who don't really give a shit what the
model is, they bend it to their own model until its sexy.
in other words, the french.
in all
fairness, if there were a hardware vendor willing to follow
the 'known working' path to ALSA glory, we wouldn't be having this
discussion .. or, at least, if we knew of such a vendor (i'm sure
they're out there, those quiet linux VAR's who pack it all up and
send it off, operational-like).
jwz would have refused to buy the equipment from that vendor. he would
have insisted that it should work on the gear he already has, since ALSA
says that its supported. "and now they tell me i have to buy a machine
from this Linux Audio Systems people just to play music at the same time
as making a skype call. wtf?"
linux guys giving the 'jwz should just by an sb-live' argument are
weaker than piss. i mentioned jwz in the first place because you
should know: linux audio needs to be a pop star. until it is, it
won't be working well enough.
there are millions more people doing audio on windows
than on osx, and
yet windows requires *at least* as much work to get setup for pro-audio
as linux. so what conclusion do we draw from that?
sorry, but no. i set up my firepod, a wonderful audio i/o interface,
and from that point on it just worked. setting it up consisted of a
bit of humble-pie windows driver-.exe downloading, exactly *one*
'confirm that you want to do this' step, and from that point on it
was magic.
perhaps what the ALSA crowd need is a crash-course on making
AUTOPACKAGE shell scripts which leave the stupidity of entering the
root password up to the user, and from that point on just provide
what everyone wants: a working AUDIO subsystem.
--
;
Jay Vaughan