hi

i've tried to push a bunch of patches from my local repo to a server
one, and darcs just hanged. then i tracked down that there is a
"special" patch that causes the hang

if i do a "strace -p" on the server's darcs process, then i get the
following loop:

--- SIGVTALRM (Virtual timer expired) @ 0 (0) ---
sigreturn()                             = ? (mask now [HUP INT ABRT PIPE
TERM])
rt_sigprocmask(SIG_SETMASK, [], NULL, 8) = 0
rt_sigprocmask(SIG_BLOCK, [HUP INT ABRT PIPE TERM], [], 8) = 0
rt_sigprocmask(SIG_SETMASK, [], NULL, 8) = 0
rt_sigprocmask(SIG_BLOCK, [HUP INT ABRT PIPE TERM], [], 8) = 0
rt_sigprocmask(SIG_SETMASK, [], NULL, 8) = 0
rt_sigprocmask(SIG_BLOCK, [HUP INT ABRT PIPE TERM], [], 8) = 0
rt_sigprocmask(SIG_SETMASK, [], NULL, 8) = 0
rt_sigprocmask(SIG_BLOCK, [HUP INT ABRT PIPE TERM], [], 8) = 0
rt_sigprocmask(SIG_SETMASK, [], NULL, 8) = 0
rt_sigprocmask(SIG_BLOCK, [HUP INT ABRT PIPE TERM], [], 8) = 0
rt_sigprocmask(SIG_SETMASK, [], NULL, 8) = 0
--- SIGVTALRM (Virtual timer expired) @ 0 (0) ---
sigreturn()                             = ? (mask now [])

and so on.

here is the repo (can be darcs get'ed):

http://ftp.frugalware.org/pub/other/people/vmiklos/

here is the patch i'm trying to apply:
http://frugalware.org/~vmiklos/patches/darcs_hang.patch

udv / greetings,
VMiklos

Thanks for the gettable repo and example.

In it helps, I can confirm this behaviour on my Mac.  I did a darcs
apply --verbose; it listed out all the extra patches, said "diffing
dir..." and never came back.

any progress? :)

sorry, i accidently removed the repo. though Dominic Steinitz in
http://article.gmane.org/gmane.comp.version-control.darcs.user/11118 also
mentions the same problem

Well, the 'diffing dir...' output is incomplete; that's not the part that takes
long.  I've been meaning to submit a patch so that it adds a new output
(something like 'resolving conflicts').  That's the slow part.

I think the only real progress will be when we switch to the new way of handling
conflicts, which might not be for a while.

We're working on it! (or rather, David and Jason are!)

The specific case here is not possible to work on, because test repo has gone
away, but the general case I believe has been addressed by "darcs 2", so I'm
marking this one "resolved in unstable".

It also sounds like a dupe of issue566, which remains open.

On Mon, Jan 07, 2008 at 05:10:53AM -0000, Mark Stosberg <bugs@darcs.net> wrote:
> The specific case here is not possible to work on, because test repo has gone
> away, but the general case I believe has been addressed by "darcs 2", so I'm
> marking this one "resolved in unstable".

as i mentioned before there is an other repo in which this was
reproducible as well, but using darcs unstable, it works pretty well. so
it seems to be really resolved in unstable, yay! :)

thanks,
- VMiklos