>>>>> "Stuart" == Stuart Hughes <[EMAIL PROTECTED]> writes:
Stuart> [EMAIL PROTECTED] wrote:
>>
>> I think that priority inheritance is for people who want to build
>> complex critical realtime systems that sometimes work.
Stuart> Hi Guys,
Stuart> Much of todays software today is complex, and cannot be fully
Stuart> tested. It is a fact that the software in avionics flight
Stuart> boxes today can only be exercised through a limited range of
Stuart> all its potential paths. Next time you're on a plane that is
Stuart> on final approach, be glad the designers used all the
Stuart> available mechanisms to ensure you get down in one piece,
Stuart> they may never need the defensive parts of the design, but
Stuart> things can wrong in unpredicable ways.
Stuart> I admire your confidence in your own software design skills,
Stuart> but if you ever write the software for the plane I'm about to
Stuart> take, I may decide I don't want to get on.
Hm, that's interesting, I get the opposite conclusion from the very
same reasoning...
We were critical of software design approaches that say "we can't test
the whole thing so it's possible that it will get itself in trouble,
and just in case that happens let's put in priority inheritance
because it *may* help". And you seem to be saying that this is the
approach used when building avionics, and this is a good thing.
Conversely, I prefer a design approach that relies on maximally simple
mechanisms to make the system easier to analyze and test fully, so I
do NOT have to rely on "let's throw this in because it might help"
approaches. And you're saying that this conservative design approach
is one you would *not* want to fly with?
Strange.
paul
-- [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/
