[darcs-users] bug report: darcs appears stuck in busy loop
droundy at abridgegame.org
Sun Mar 28 12:30:46 UTC 2004
On Thu, Mar 25, 2004 at 08:07:48PM -0800, Adam Megacz wrote:
> David Roundy <droundy at abridgegame.org> writes:
> >> Using 0.9.17 the pull runs for over an hour at 97% cpu...
> > Just to let you know, I've been working on this for a few days now.
> Hey thanks. BTW, Brian, this is why I haven't merged your changes yet ;)
> > I was hoping to figure it all out before responding, but that looks
> > unlikely. I've been tracking down the root cause of the problem, and have
> > eliminated a number of possibilities, but expect it'll be at least a few
> > more days before I can get it figured out. :(
> No problem. Thanks for letting me know! I don't know if it's exactly
> the same problem, but I encountered this behavior once before.
Alas, I seem to be defeated. It looks like the slowness involved can't be
avoided without a major change in how darcs deals with conflicts. In
particular, every patch involved in a conflict "knows" about all the
patches that were involved in the conflict before it, which is problematic
when there are too many patches involved in a conflict.
In your problematic pull, there was a single (intermediate) patch that
required 300,000 lines to describe, because it conflicted with essentially
every previous patch. :( It is possible that there is a bug somewhere
involved, but if so I haven't been able to find it. I do believe the pull
will finish eventually (I've been looking at intermediate steps), but have
no idea how long it would take. In the meantime, I recommend making these
changes by hand (and then unrecording them in the repo in which they were
originally made). Mostly, it seems to simply be a job of hitting M-Q in
emacs about 100 times, but it's hard to see what else was changed.
More information about the darcs-users