Subject: Re: [linux-audio-dev] A Python loop sequencer, was re: using csound as a sampling engine
From: Benno Senoner (sbenno_AT_gardena.net)
Date: ke syys 08 1999 - 19:00:30 EDT
On Wed, 08 Sep 1999, David Slomin wrote:
> On Wed, 8 Sep 1999, Benno Senoner wrote:
>
> > I think sooner or later we should all join our efforts and contribute
> > to the Audiality project (with first versions running in userspace on
> > ordinary Linux kernel , then on RT-Linux with sub 1msec lanecies).
>
> Don't forget that there's never such a thing as "one size fits all",
> only "one size fits most" at best.
Agreed.
>
> I don't see myself as ever running Audiality (much less writing code that
> relies on it) because I do not like the idea of having to run RT-Linux.
> I'm a hobbiest musician, not a professional, so I'm not willing to
> dedicate a machine to the job; I only run general purpose OS's. This
> doesn't mean I think there's no place for Audiality, but I don't think
> it's realistic to expect _all_ the Linux audio developers to target that
> single niche.
Yes, this is the main reason I stressed the kernel folks so much to
improve the userspace realtime behaviour, and in fact
Ingo did it: now we can implement much lower-lantency soft-synths
as Windoze on the same hardware, RELIABLY, with no glitches during
system load.
RT Linux is way to difficult to install for Joe average user.
Since shared-mem/IPC is pretty fast, we could implement a version of Audiality
by letting run the audiality "daemon" as normal SCHED_FIFO process,
on a a regular linux kernel (with the hope that the low-latency patches will go
into the mainstream kernel).
RT-Linux is maybe only needed by the *VERY* audio professional,
but Joe average will go with userspace engines.
If RT Linux will go into the mainstream kernel someday, then we could provide
both engines (RTLinux engine and userspace engine), to provide easy porting
to other POSIX architectures.
Since the userspace engines are almost POSIX code + some OSS/ALSA code,
it would be not too difficult to port the engine to other POSIX OSes like
Solaris, FreeBSD etc.
(But as far I know FreeBSD still lacks of the POSIX Realtime extensions
( sched_setscheduler() are fake calls with no effect),
therefore you can't run your 4ms-latency soft-synth on this OS)
>
> Besides, I don't think I've ever even written a program "for Linux". I
> write for Posix or for the JVM and test/use it on Linux. Here's a new
> poll for the day: How many of you developers on the LAD list only target
> Linux?
I think it will be like with other projects: there is a release for the primary
OS and other groups will provide the ports.
But actually Linux is the only free OS capapable of this hardcore
realtime-audio stuff.
regards,
Benno.
This archive was generated by hypermail 2b28 : pe maalis 10 2000 - 07:27:11 EST