On Tue, Aug 12, 2008 at 2:03 PM, Claus Reinke <[EMAIL PROTECTED]>wrote:
[snip]

>
>
> Much as I'd have liked to advocate darcs 2 over git for ghc, I
> couldn't really have justified that until the little bugs are known to
> be fixed. Of course, I doesn't help that darcs is so terribly slow
> on ghc (and David indicated that this might be an inherent limitation
> of darcs on large projects).


Claus,

I believe Eric is going to address your concerns in detail, but I wanted let
you know correctness is very important to us.  Let me just give a quick run
down of the testing that darcs uses currently:

1) We have over 100 scripts that do functional testing (last time I checked
I think it was nearing 150).  Pretty much any command that darcs supports
gets tested here at least in the simplest or ideal cases.  These tests are
ad-hoc and some things get tested more thoroughly than others.  We even have
a test suite now for `networking' functionality.  We like it when users
submit reproducible examples because we add them to our test suite.

2) We use buildbots to help ensure that darcs stays working on some of the
more popular platforms.  The buildbots run all the above tests.

3) We have a collection of unit tests for the core of the patch theory.
Eventually we'll expand our unit tests out to every module, but this is a
start and also where we need it first.

4) We use GADTs + phantom types + existential witness types to enforce some
of the properties of patch theory.  For example, once you have a sequence of
patches any manipulation to their order must respect the properties of patch
theory, or else it won't compile.  This makes it so that our patch
manipulations are safe.  This beefed up type safety is an on going effort.

We are currently working towards a 5th type of testing, a performance
regression test suite.

Please let us know if you have any suggestions on how we can test more
effectively.

Thanks,
Jason
_______________________________________________
darcs-users mailing list
[email protected]
http://lists.osuosl.org/mailman/listinfo/darcs-users

Reply via email to