Recent Experience with Darcs

[Curt Sampson <cjs%netbsd.org@localhost>]

On 2008-07-30 06:03 +0000 (Wed), David Holland wrote:

>  > - merge tracking
> ...
> darcs has some of the bits you need but I'm not sure it has all
> of them.

FWIW, I've been both using Darcs recently and following some of the
stuff on their mailing list.

While Darcs has its good points, I don't think it's an option worth
seriously considering, unless perhaps we have several people in the
group who are decent Haskell programmers and are willing to work on the
deficiencies. Here are a few random points I picked up.

* The "patch theory" Darcs uses is pretty hand-wavy and, in the opinion
of more than a few smart people, no better than the informal models of
many other systems.

* The UI has definite issues, particularly in terms of merging. It's
easy to accidently lose a pair of conflicting changes, and this does not
seem to be a big issue for the Darcs developers. I've also found it to
generate spurious conflicts that SVN doesn't in the same situation.

* Performance, even in Darcs 2, is still a problem.

* Building it is going to be a serious issue for platforms such as the
VAX. ghc is losing the C back-end as of version 6.10, so we'd have to
write native code generators for any systems with less-popular CPUs.

I speak here as a big Haskell fan, big enough that I've gone to some
trouble to make programming in Haskell my day job.

I should also point out that ghc, one of the biggest Haskell projects
out there, and one with no lack of top-notch Haskell programmers, has
decided to move from Darcs to git.

cjs
-- 
Curt Sampson       <cjs%starling-software.com@localhost>        +81 90 7737 
2974   
Mobile sites and software consulting: http://www.starling-software.com