[darcs-users] Future of Darcs: Thanks for the poll responses and increase in contributors

[Trent W. Buck]

"circ ular" <circularfunc at gmail.com> writes:

> 1. Why is darcs slow and is it in general or at a specific task? Is it
> the algorithms or the language?

While darcs is probably slower at non-interactive operations (like
applying a patch), I feel more productive with darcs than hg (I use both
at work) because I spend less time using it interactively.  I believe
this is because darcs' has a simpler UI, which I grok more, and which
better maps to my workflow.

(For example: to amend in hg you have to save the commit message
somewhere, manually delete the patch, then re-commit all the old hunks
plus the new hunks.  Similarly, "hg record" changes the mtime of
recorded files, so I have to waste time doing M-x revert-buffer in
Emacs, possibly losing unsaved changes.)

> 2. Is darcs competing with git or they different kind of users?

Personally, my approach is to use darcs at first, because it has the
least annoying UI.  If it ever turns out to be too slow (because the
repo has too many patches or something), *then* I'll consider switching
*that* repo to git or svn.

> 3. Why is performance slow on Windows and more exactly what needs to
> be done? Improved algorithms or perhaps C-optimization?

Probably it's simply because there are fewer developers and users on
Windows than Unix.  Thus Windows performance has been tested and
optimized less.