On Mon, Dec 20, 1999 at 07:23:34PM +0200, Alex Dubov wrote:
> Hello everybody.
>
> I joined this mailing list not a much time ago, and I'm already having
> some disconcert with what is going on (especially, with this RTAI/RTL
> stuff). So I want to point your attention to some issues, I count as
> important:
> 1. RTlinux is a great thing, but it is not very convenient in handling
> and deployment. It will be great, if the RT functionality could be
> implemented in the mainstream linux kernel. I'm thinking about a policy
This is not likely for technical reasons.
> driven scheduler (in style of Solaris' kernel), that will support
> different classes of processes (in the beginning just RT and TS
> classes). It will be great to implement a "guaranteed response time"
> feature in the linux kernel, too (just like in the Solaris).
Try to develop a periodic task with a 100microsecond period under
Solaris and see what happens.
> 2. Another heavy drawback of the RTlinux is a kernel space processes.
> Such processes require excessive debugging and are somewhat dangerous
> for system stability. This calls for implementation of user-space event
> delivery mechanism (hopefully faster, than this in DOSEMU). Primary use
> for this feature will be it utilization in the high-level
> development/prototyping systems. It is very important to carry out a
> rapid and cheap prototyping (especially in the academy world, but for
> industry too) and computers nowadays are fast enough to revert from
> compilation to interpretation techniques for control scripts (using such
> environments as Mathworks Simulink or, may be, evolved Octave).
We are moving towards a "debug under pthreads/ run in kspace" functionality.
> 3. It will be nice to implement standard driver API. Comedy is useful
> only for data acquisition, because it's developer's API is not flexible
> enough.
We are moving that way. Specific suggestions are nice.
> These steps can carry linux to the new stage in it's development.
> Microsoft and related companies are working on WinCE V3, that will have
> (at least, theoretically) most of the listed features (included not
> listed APM that is extremely important for low power embedded apps, and
> completely uncovered by RTlinux developers; APM support requires kernel
> redesign, because many tasks require prolonged sleeps and brief
> wake-ups). In spite of financial pressure (and generous aid) provided by
I don't see why this requires redesign.
--- [rtl] ---
To unsubscribe:
echo "unsubscribe rtl" | mail [EMAIL PROTECTED] OR
echo "unsubscribe rtl <Your_email>" | mail [EMAIL PROTECTED]
----
For more information on Real-Time Linux see:
http://www.rtlinux.org/~rtlinux/
