Hi,
it seems that somebody is interested to discuss the RTL architecture,
and a few even in favor of the HAL/RTAI idea.
I got answers, both personally and from the mailing list, and I want to
clarify some points.
First I do not think that 2.4.xx should include a native hard real time
support but should definetly go to a modularization based on the
HAL/RTAI concept. Unfortunatly I am unable to contact "LINUS_GOD who art
in heaven" to have him evaluate the idea.
Second the concept above can be relaxed only if Linux is rewritten
following QNX and not UNIX architecture. But that's the microkernel
concept, with all the services as tasks with different privileges, that
Linus refused in the debate with Tannenbaum. However modules can bring
to a balanced compromise.
Third I do not want to be the focus of anything in RTL community. In my
message I stated that the NMT "promogeniture" (read it: the right of
Victor & C. to be the refrees of a bazaar develpment of RTL) should be a
firm point. What I complain about is their reluctancy to accept the
bazaar, as any change and discussion about the basics are rare and RTL
remain primarely an NMT effort based on NMT ideas.
Fourth before talking I've implemented a working version of HAL/RTAI.
>From that I learned that I still have to learn a lot, but that it is
definitly the best way to go, and I will go that way anyhow. The point
is that the discussions I would like within in this group should be.
also but not only, of the type: is it necessary to patch that and this?
why an UP machine has less jitter that a standard dual SMP? What happen
if we overload the apic bus in real time applications? And many others
of the same kind.
Ciao, Paolo.
--- [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/
