[darcs-users] *practical* differences between darcs' patch model and git/mercurial's?

[Florian Weimer]

* Stephane Bortzmeyer:

> On Sun, Oct 21, 2007 at 12:46:34AM -0700,
>  Adam Megacz <megacz at cs.berkeley.edu> wrote 
>  a message of 33 lines which said:
>
>> with git or mercurial, even a "passive" developer who pulls these
>> two patches will need to add an additional content-free patch to
>> merge them.
>
> I regard this as a huge practical problem and this is the main reason
> why I hate Mercurial.

The flipside is that the darcs approach means that you cannot easily
reproduce intermediate tree states, which makes me feel somewhat
uncomfortable.

There's also the issue that darcs' cherry-picking is not universal.  If
it works, it's great, if it doesn't (you cannot commute the
patch-to-be-picked past something you don't want), you may need to do
major history surgery or create a patch from scratch, as in other
systems (but if I'm not mistaken, such duplication tends to cause darcs
to run into the exponential trap).