[Oberon] Re: Robot & critical real-time systems with XO/2... not easy...

[Fernando Passold]

Hello Mr Freed, Mr Danforth and Mr Drake,
and the others who give me
some inspiration,

> NO has been used very successfully in real-time control/data-acquistion 
> apps.  
> Speak to Mike McGaw.  He talks every now and then on comp.lang.oberon,
> so you should be able to extract his email address from those archives. 
>  I'm
> sorry, but I don't have his address at home, which is where I'm working 
> today.

    Ok... I finally discover the problem and the solution.
The mesage of run-time erros is related to low resources of XO/2 to
allocate another task to its task's handler... Unfortunately the
mesage generated is not clear... But before make a lot of
combinations, adjusting the time parameters of mine real-time
tasks and before to start the main tasks, mine software performs
a test to verify if the task's handler of XO/2 was really
allocated a memory to execute the pretended task... Of course, It
take me a lot of time to discover this. Quite 1 week, but I reach
the conclusion that I must enlarge the mainly deadline of mine
real-time tasks from 1 [ms] to 2 [ms]. And that was enought to
make my system works fine. With, of course, a quite exasting process
of compile-run-and-testing the time parameters of mine critical
real-time task (especially the parameter "duration").
Well... I'll hope I could finish my job in more 5 days that
rest to me.... It's so hard to deal with critical real-time
systems... It's 01:57 am and I am so tired that I go home now!!!
My wife is supposed to be borring with me...

thanks a lot about your interest,

Fernando Passold