[gobolinux-users] Call for volunteers: would you like to be a recipe committer?

[Dave Dodge]

On Fri, Jul 20, 2007 at 01:23:46AM -0300, Fábio Mierlo wrote:
> Darcs [1]!

I actually had a line at the end of my message "now somebody recommend
Darcs and we're done" but took it out just before sending :-)

> Where patches are global to the repository and are subject under
> certain conditions to reordering, as governed by the project's "theory
> of patches" [2].

Yeah, I've always found idea that there is an actual theory underlying
its operation to be pretty fascinating.  However, the current
implementation has some possible downsides:

 - Some large projects that have tried to use it have had issues with
   its speed, especially relative to things like git and hg.  Probably
   not an issue for small projects.

 - It's written in Haskell, which could be a good or bad thing
   depending on your point of view.

 - There are unresolved corner cases where operations can hang
   indefinitely, run out of memory, and so on.  For example:

     http://www.mail-archive.com/darcs-users@darcs.net/msg03440.html

> ps2: in the end, the importance is if it is client-server or distributed.
>   To me, off-course...

Right, the switch from centralized to distributed is the big step, and
on a small codebase pretty much any of the distributed systems will do
the job equally well.

                                                    -Dave Dodge