Re: a joint letter on low latency and Linux

[Larry McVoy]

> > I and others have pointed out that if you need hard real time then
> > what you do is use RT Linux and you can get exactly what you need.
> > I've heard two arguments against this:
> > 
> > a) ``Other operating systems offer "soft realtime" and we don't want to
> >    port our code from that to the RT Linux model.''  Translation:
> ....
> > 
> > b) ``If we take the RT Linux path then Linus will never add the low latency
> >    features we want.''  I love this argument.  It implies knowledge that there
> >    is a better way, that if you use the better way then there is no need to do
> >    what Linus doesn't really want to do.  Exactly.
> > 
> > In fairness to the low latency application people, I think that the
> > RT Linux folks need to provide skeleton "apps" that show how to solve
> > problems in the RT Linux space.  Whether that happens or not, the RT Linux
> 
> Larry McVoy,
> 
> Then you fundamentally misunderstand the nature of audio and various
> media programming. These folks, including me need to have a reliable
> jitter free process model from the deepest points in the kernel to
> userspace... Including the ability to handle lots of floating point
> operations through out that path.

Perhaps you would care to explain why it is that you can't do exactly
what you want with RT Linux?  If you did that, then you might find either
(a) that you are mistaken about the limitations of RT Linux, or (b) the
RT Linux folks will provide what you need.

All I am hearing is that it is too hard to do it inside of RT Linux and
that just sounds like whining (to me at least).  What I am not hearing
is that it can't be done inside RT Linux, in fact, quite the opposite.
There have been many statements of "We don't need that much real time,
we just need XYZ".  I'm sick to death of that sort of statement - I've
lived through that in SunOS and IRIX and I can promise you that if Linus
gives in on the first demand, you'll promptly be back with the next.
Give it enough time and Linux == IRIX.  Is that what you want?  No,
you'll scream, I just want _my_ one little feature that helps my one
little application and it won't hurt the kernel at all.  

What I want to know is how are you any different than the same set of
people who screwed up IRIX/SunOS/AIX/HPUX/name-your-favorite-Unix-here?
All of those people asked for exactly the same things, made exactly the
same set of promises that it wouldn't hurt, and they all turned out to
be wrong.

We have a nice little OS here.  Instead of asking for the same stupid 
ideas that can clearly be seen to be the wrong approach, how about 
thinking out of the box?