[darcs-users] Re: Subversion vs DARCS (was: Moving sf.net CVS to cryptomonkey.net Subversion)

David Roundy droundy at jdj5.mit.edu
Mon Jul 7 14:20:39 UTC 2003


On Mon, Jul 07, 2003 at 09:57:15AM -0400, Zooko wrote:
> 
> [I, Zooko, wrote the lines prepended with "> > ".]
> [David Roundy wrote the lines prepended with "> ".]
> 
> > > Section 4: why DARCS is speeecial
> ...
> > > Section 4.a: conflict management at the level of patches instead of
> > > branches
> ...
> > > Section 4.b: cherry-picking of patches
> ...
> 
> > I think this section is just a little overly optimistic.
> 
> :-)  I'm glad that you are honest enough to say so.
> 
> I'm still interested in trying it out, if only for my own edification.

Oh, and I'm definitely interested in having it tried out!  :)

> > I would love to create a graphical merge tool that would give you all sorts
> > of interesting options, but currently the state of haskell GUI libraries is
> > a bit discouraging.  I probably will start with creating an interactive
> > text merge tool, but even that is a relatively low priority.
> 
> FWIW, I usually dislike graphical tools.  It's good for the program to display 
> graphics to me -- graphics have much higher bandwidth than text does.  But it 
> is bad for the program to require input from me in the form of menu-selections- 
> and-mouse-clicks.  Menu-selections-and-mouse-clicks have *lower* bandwidth 
> than text does, and more importantly they aren't programmable.

Well, the other thing that a graphical tool can do is allow the user to
make choices in whatever order makes sense, which is why I think a
graphical tool would be nice.  It would let you see all the conflicts and
from all that information decide how to resolve each one.

> I *think* that the ideal merge tool that I want just tells me which patches 
> are the specific cause of the merge conflict, and otherwise it behaves just 
> like CVS does.

I would definitely want to have more than one merge tool, and the first to
be created would be an interactive text mode one that prompts you for each
conflict in order (of course, asking you what you want to do with it).  As
for what you'd want (basically like CVS), that is the default (except that
you aren't told about which patches conflict--that could be added to the
TODO list).

> > One thing that occurred to me that you might not have considered is that
> > darcs hasn't been ported to windows.  I took a look at mnet but didn't
> > immediately see if it ran on windows.  Darcs could certainly be ported to
> > windows (and maybe even without too much work), but I'm not going to do it,
> > as I don't have windows.  It does run on MacOS X, though.
> 
> Definitely a drawback!  But has anyone tried it on cygwin?

No, noone has tried.  I don't think it would be too hard, but it probably
wouldn't be a trivial compile either.  Darcs uses pipes to run text editors
and to do its emailing, which I imagine won't just compile under cygwin.
It also uses pthreads and libcurl to do its downloading over http.  libcurl
would be no problem, but the threads stuff probably would be.
-- 
David Roundy
http://civet.berkeley.edu/droundy/




More information about the darcs-users mailing list