Subject: Re: [linux-audio-dev] Still I cannot understand why...
From: D. Stimits (stimits_AT_idcomm.com)
Date: Mon Dec 17 2001 - 11:33:09 EET
Ivica Bukvic wrote:
>
> > "if we get everyone to switch to alsa, then everyone can share audio
> > interfaces".
> >
> > "if we teach everyone how to use LD_PRELOAD and they accept it, then
> > everyone can share audio interfaces without even switching to ALSA."
>
> Well this sounds like a win-win situation and we should advertise it as
> such, and I mean advertise it BIG TIME. I think that the idea of not
> having to worry about polling sound signals to the dsp resource sounds
> so appealing, since that makes one less thing for a programmer to worry
> about :-).
LD_PRELOAD is like performing open heart surgery with a heart bypass
machine. If you do this with the wrong thing, it is fatal. Anything that
can be run with LD_PRELOAD can be put into the libraries or other code
by the normal mechanism, LD_PRELOAD is good for testing and debugging,
it isn't the sort of thing to mess with. If you must preload a function,
why not just get the library that has the original and get that changed?
Then everyone has it without the hassle and danger. Even if you set the
variable from a shell environment, it can have security problems.
D. Stimits, stimits_AT_idcomm.com
>
> > "if we get everyone to switch to alsa, and they use the poll interface
> > via ALSA to control their workflow, and/or only read/write avail_min
> > chunks of data, then they'll be ready for their app to work in a
> > synchronous execution system, of which JACK is an example."
> >
> > Just to keeps things complex...
>
> When the benefits are so great, the complexities are easy to overcome :)
>
> Ico
This archive was generated by hypermail 2b28 : Mon Dec 17 2001 - 11:27:37 EET