[Shootout-list] Re: ring of processes

[Bengt Kleberg]

Aaron Denney wrote:

...deleted

> 
> Well, I wouldn't call it cheating, so much as "the natural way to solve
> this problem doesn't test what you want to test".

this is for the time beeing my main problem. i seem to be unable to 
force the ''correct'' solution by just juggeling processes, 
communication patterns and message types. not while still keeping with 
the other basics i want (simple, easy to understand, linear with just 
one input parameter, etc)


...deleted
> True, but I don't think it matters.  Fix some number of processes, then
> let the number of messages be a varying input parameter.

i had that, but a large number of processes makes it neccessary to have 
huge amounts of messages to hide the process startup costs.
one the other hand a small number of processes makes it much more likely 
that a mesh-style channel-between-all-processes ipc can be used.


...deleted


to keep the test simple (easy to understand and implement) i am leaning 
towards more rules on how one must solve the problem, and less smart 
formulations of the problem itself.

due to this change of heart it could be that the current ringmsg test 
should be fixed according to your smart problem description. i would be 
happy to try and write such a program in erlang.

than the ''lots of rules'' test i am now thinking about could be called 
sys5msg instead.

or, since i was here first (not a very good arguemnt, i know :-) i am 
allowed to update the ringmsg test with additional rules and all other 
entries than erlang (haskell, so far) would have to recode.

anybody out there that has strong feelings about which way to go?


btw: i have not managed to do the ringmsg in mzscheme yet. i have the 
sys5msg parts done, but need to write the program that uses them.


bengt