[Shootout-list] NOT ACCEPTED: Program should use exception handling.

[Isaac Gouy]

> > "NOT ACCEPTED: Program should use exception handling."
> Let me just say that I don't agree with this at all.

Discussion of what it means for a program to be accepted as completing
a benchmark is becoming more ripe. 


> While I eventually admitted defeat regarding the use of structures
and
> ingenious foo to create object systems for C

I forgot about that - let me compile a list of non-OO programs
purporting to complete the object/method benchmarks! :-)


> I have to say that disqualifying a language like Felix 
> for using its in-built error handling system is not reasonable.

The language Felix has not been disqualified! 

The Felix "exception handling" program has been shown as an
"interesting alternative program" because as the author has
acknowledged several times it doesn't implement the benchmark
as-required "in the same way".  

Noting that there are other programs that don't implement the benchmark
"in the same way" is a good argument for moving those other programs to
"alternatives" and a lousy argument for including another program that
doesn't implement the benchmark.


> The key idea here is to see what the cost is to use a language's
> error handling system.

In context that's fine - in the same way that the context for fibonacci
is that the program *must* be recursive.

We aren't just interested in the cost, we're as interested in what we
get for the cost. Can we restart? Can we step back before the exception
and restart? Are we distinguishing between different severities? 

And that makes except a lousy benchmark - but it still isn't an
argument for including programs that don't do what the program asks.


> Call/CC or Common Lisp's escape clauses should be allowed
afaict CL has 'Conditions' 
Scheme? http://srfi.schemers.org/srfi-34/srfi-34.html


		
__________________________________ 
Do you Yahoo!? 
The all-new My Yahoo! - Get yours free! 
http://my.yahoo.com