Re: Real time Java? Huh?

[jpt@diphi.demon.co.uk (JP Thornley)]

In article: <71vhli$hul$1@nnrp1.dejanews.com>  oopster@my-dejanews.com 
writes:
> 
> Is it still the case that Ada exceptions are to be avoided for 
real-time?
> Not a good thing if true...
> 

For high-integrity Ada, I've always been fascinated by the existence of 
the two opposed camps on exceptions - one group taking the view that 
having exceptions in the language makes it *more* suitable for 
high-integrity applications and the other regarding them as, not quite 
the work of the devil, but definitely not something that self-respecting 
software engineers would want to get involved with.

The Annex H Rapporteur Group spent a long time discussing the 
appropriate advice to give for exceptions in the (currently Draft) 
"Guide for the Use of the Ada Programming Language in High Integrity 
Systems" with the result that it describes three different approaches:-

1. Catch and handle all exceptions locally - so there is no exception 
propagation; program state must be well-defined after the handler has 
executed.

2. Use a single catch-all handler at the top level that makes almost no 
assumptions about program state - it probably does a restart.

3. Prove that no run-time errors can occur (so code can be compiled with 
all checks suppressed). [But you often still include the 'catch-all' 
handler to cope with hardware glitches/memory corruptions.]

Phil Thornley

-- 
------------------------------------------------------------------------
| JP Thornley    EMail jpt@diphi.demon.co.uk                           |
|                      phil.thornley@acm.org                           |
------------------------------------------------------------------------