[darcs-users] some darcs-relevant stuff from ICFP

Ganesh Sittampalam ganesh at earth.li
Tue Sep 8 19:33:26 UTC 2009


Hi,

Since I was at ICFP last week, I thought I'd let people know about the 
interesting things related to darcs I heard about there.

- There was some further discussion at HIW[1] about the darcs->git 
migration and the GHC team weren't really sure what to do - long-running 
branches were still causing pain, but on the other hand an experiment with 
git had also apparently been unpleasant. I said that I would try to do 
something about getting darcs rebase[2] implemented in order to improve 
the situation there. General frustration was also expressed with the 
uncertainty, which was causing people to not do things.

- There's a new IO system in GHC 6.12 support UTF8 properly, as Jason 
already emailed about. It also exposes the Handle abstraction, allowing 
users to write their own types of Handles (e.g. backed by in-memory files 
or mmap'd files); not sure if this'd be any use to us though.

- There's a new type inference algorithm for GADTs. Unfortunately I missed 
the talk because it was right at the end of the main conference and I was 
getting a bit conferenced out at that point, but the paper[3] looks nice. 
The upshot as I understand it is that we should need fewer type 
signatures, but it's not in 6.12 so it'll be a while yet before this helps 
us.

- Bryan O'Sullivan is writing a new benchmarking program called 
'criterion' or something similar. It's notable in that it will use 
statistical techniques to compare the results of multiple runs and figure 
out if they were interfered with by the environment, e.g. other processes 
putting load on the machine. Perhaps that'll make it more feasible to do 
some automated performance regression tests.

[1] http://haskell.org/haskellwiki/HaskellImplementorsWorkshop
[2] http://bugs.darcs.net/issue938
[3] top paper on 
http://research.microsoft.com/en-us/um/people/simonpj/papers/gadt/

Cheers,

Ganesh


More information about the darcs-users mailing list