cocoon-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Hunsberger, Peter" <Peter.Hunsber...@stjude.org>
Subject RE: Gump
Date Wed, 03 Mar 2004 16:12:10 GMT
Stefano Mazzocchi <stefano@apache.org> writes:

<snip silliness/>

> 
> > Seriously, there have got to be some mathematicians around that you 
> > can pose the question to?
> 
> The problem is not graph theory, it's algorithmical complexity: if it 
> takes two months for gump to build cocoon, the whole thing wouldn't 
> converge.
> 
 
I think it's one and the same: find a graph traversal algorithm that
works (in some reasonable O) for this case and you've solved the
problem? 

Then again, maybe a genetic fitness type algorithm might work: assign
each release a "fitness" based on how many other projects can compile
with it and always try the version with the highest fitness first.  Any
release with a fitness < some X dies off (or at least has a very low
probability of being picked to "breed" with some other project). No
graph traversal, just one big gene pool (so to speak).

> I'll find the answer, don't worry.

No worries on this side, just some interest in observing the problem
solving process....



Mime
View raw message