On Tue, Apr 29, 2008 at 2:44 PM, Duncan Coutts
Yes, we typically expect 'easy' instances of the problem. Though we would like to be able to handle large easy instances too, like trying to install the latest version of every package on hackage simultaneously.
Yeah, that kind of benchmark makes me a bit nervous but is should definetely be a good stress test.
Perhaps we should have a go with profiling and see if there is anything obvious.
Absolutely.
Actually it didn't get funded as a GSoC project but Maciej Wos is looking at this problem for his 4th year project. As you say, the design space is very wide so I would not want to discourage you or anyone from looking at it.
Ah, I didn't know it was denied funding. Too bad. Good to hear Maciej is still interested in working on it.
I haven't looked into how to use this package in cabal-install yet, but I suspect it will be fairly easy. I'd appreciate any comments and help on that though since I'm not familiar with the cabal-install code base yet.
I want to make the resolver a bit more easily replaceable.
[Lot's of useful information snipped]
Thanks for the info Duncan. I'll be looking into this but don't expect super rapid progress. Where do I send patches? Cheers, Josef