[darcs-users] Re: Where Free Software revision control is going

[zooko at zooko.com]

Last I heard [1], bzr was planning not to do patch commutation.  Therefore, it 
will not do most of darcs's unique features.

On the other hand, the new merge algorithm being cooked up by Bram Cohen
(Codeville), Ross Cohen (Codeville), and Nathaniel Smith (Monotone) might be
able to do most of darcs's unique features while avoiding the tangly merge
conflict problems.  We'll see.

On the gripping hand, speculation about the future doesn't butter parsnips.
Darcs works very well for some kinds of tasks today, and Codeville and Monotone
may also work well -- I haven't personally tried them yet.  In my somewhat
well-informed opinion, any of these projects could turn out to be excellent
tools in six or twelve months: darcs (with efficient-merge-conflict-magic),
Codeville or Monotone (with their new merge algorithm), Bazaar-NG a.k.a. bzr
(even if it doesn't have a good merge algorithm, but it turns out that a tool
without a good merge algorithm is still nicely useful, or else if bzr adopts a
good merge algorithm).  Same with many contenders such as git, mercurial,
vesta, etc. etc.

I've really enjoyed observing these developments, as it is a rare case of Free
Software innovating instead of copying.  A necessary part of innovation is that
you have to have experiments which try something new and fail.  The pressure to
"pick winners" is one of the things that retards innovation.

Regards,

Zooko

[1] http://www.bazaar-ng.org/doc/darcs.html