[darcs-users] performance regression in annotate darcs 2.4.4 -> darcs 2.4.98

Ganesh Sittampalam ganesh at earth.li
Wed Sep 1 15:01:34 UTC 2010


On Wed, 1 Sep 2010, Eric Kow wrote:

> On Tue, Aug 31, 2010 at 17:22:31 -0600, Zooko O'Whielacronx wrote:
>> zooko at tahoe-lafs:~/playground/darcs/bench$ darcs-benchmark run -v
>> --converge darcs-2.4.4 darcs-2.4.98.3+175
>> WARNING: Could read /home/zooko/.darcs-benchmark/tahoe-lafs.org-warm.info
>> darcs-benchmark: user error (Pattern match failure in do expression at
>> source/Benchmark.hs:178:7-21)
>
> Argh, sorry about that!  I had tried it, just not using the same
> version of Darcs as you did
>
> I'll note that this is looks like a standard pattern-match failure that
> you'd check type checking for, but when we pattern match out of an
> action, we don't get GHC reminding us that this there are missing cases
> (ie. ones where splitOn returns not exactly 2 results).
>
>  [info, context] <- splitOn "Context:\n\n" `fmap` darcsMeta cmd ["--exact-version"]

The problem (for GHC) is that pattern match failure here isn't necessarily 
a runtime error, as it depends on the monad being used.

Ganesh


More information about the darcs-users mailing list