[darcs-users] Easily managing over 100, 000 lines of code with darcs
Stephen J. Turnbull
stephen at xemacs.org
Thu Apr 1 05:00:37 UTC 2010
Max Battcher writes:
> Stephen J. Turnbull wrote:
> > Mark Stosberg writes:
> > > Our workflow is primarily built around cherry picking patches based on
> > > ticket numbers in the patch names (spontaneous branches). I wouldn't
> > > call this feature of "simpler"... I know of no VCS which an do it as
> > > well as darcs does.
> > But I *would* call it "simple." It's built around a single, memorable
> > idea, and requires no extensions to Darcs or complicated option
> > fiddling.
> Ah, but I call it "complex". The darcs patch theory, upon which such
> spontaneous branches are so easily built, is a pretty complex thing of
> mathematical wizardry.
So what? As you are at pains to point out later, it's the UI and the
user's mental model that matters. Those are simple. If the
underlying patch theory is complex, that's not relevant as long as
it's implemented correctly and not in the user's face.
What could be simpler than smiling? But that requires control of 13
separate muscles (and I bet you can't name one of them; I know I
can't). Ah, I see you're frowning ... 37 muscles (or is it 43?)!
Implementation complexity doesn't matter if the user sees only
> Why is it that git (and to a lesser extent hg) make one of the
> simpler conceptual approaches ("snapshot-based DAG") so much more
> complicated to work with in day-to-day practice?
Because it doesn't. The history DAG is an idea with very complex
ramifications (it is a representation of concurrency, which is
arguably the hardest problem in computing science today). As I
pointed out earlier, with a few exceptions (like spontaneous
branching), the git workflow corresponding to any given Darcs workflow
is rarely more complex, though Darcs often does it more smoothly.
As long as you continue to parade these canards about git, you're
going to have to accept the canards about Darcs. There is some truth
in both camps, but a lot of it is misconception.
More information about the darcs-users