Re: Wanting opinions->uController OS

["Bryan J. Smith <b.j.smith@ieee.org>" <thebs413@earthlink.net>]

From: NZG <ngustavson@emacinc.com>
> It does actually, starting with the 2.6 kernel a large chunk of the uClinux 
> project was integrated into the mainstream kernel.

Yes, much, but not to the level you are comparing against.

Now I will see you in the fact that it's nice to finally have, even if
just partial, a Linux tree that does both MMU and MMUless.  But
there are lots of BSD variants that do both as well, even if they
are not as popular.

> If you can't make a point without insults, maybe you should take a
> break and think about whether it's really a valid point. 

I'm not insulting you.  Yes, I'm using strong language.  But I'm am
directly and most pertinently pointing out why you _continue_ to
miss points due to lack of historical knowledge and lack of attention
to context.  If you didn't notice, I largely _stayed_away_ from this
thread _until_ you went into a realm you shouldn't have.

To review:  
- You totally missed Theo's points on the AT&T v. UCB lawsuit
- You simplified and played a "don't question me" card

Now, if my "rebuke" was an "insult," I'm sorry you feel that way.
Yes, it was a "rebuke," I'll admit that, but it did have merit.

> Not really, Adeos basically removed the oldschool proprietary
> approach that RT/Linux relies on

RT/Linux is GPL, and it's patent from Finite State Machine (FSM)
Labs is licensed for use as GPL under an FSF-approved "Open
Patent."  No matter what RTAI does, it's still based on a similar
approach with clear lineage.

Yes, RTAI reduces a lot of efforts.  But I would still argue RT/Linux
is the way short of developing a fully blown, re-entrant kernel
like VxWorks or even "heavier" re-entrant kernels like Irix or Solaris.
Linux's design cannot scale, let alone for real-time, unlike those
other alternatives.

At the "commodity 2-4 way" hard real-time, Opteron/Solaris is
really starting to give Linux a butt whipping (as well as on the
server).  For more embedded footprint, RTAI still isn't RT/Linux
or VxWorks (among other BSD derrivatives).

But I _will_ see you in the fact that RTAI gives "near hard"
real-time for the "development buck."  Try not to throw up
absolutes and narrow-focus because you never know who might
be listening.  ;->

> and can be used without patching the Linux kernel,

That's the problem.  I'd rather maintain a strict separation between
a "real-time" kernel and a "monolithic" kernel that runs as a non-
real-time process above it.

RTAI is a nice option, but not good enough for many applications
IMHO.  In fact, it's approach is very much like the 2 implementations
available for the NT kernel -- both not surprisingly based off of a
license from FSMLabs for their code and patent.  ;-ppp

> but I agree, we are quickly moving off track, this is an thread
> for rtai.org/vxworks lists

You introduced it.  It's still Linux, so I have no issue discussing the
technical merits of embedded and/or soft-to-hard real-time.



--
Bryan J. Smith   mailto:b.j.smith@ieee.org


-
To unsubscribe, send email to majordomo@silug.org with
"unsubscribe silug-discuss" in the body.