When I try to apply the attached patch to the http://allmydata.org/ 
source/tahoe/trunk repository, using darcs-2-pre, I get:

-------
HACK wonwin-mcbrootles-computer:~/playground/allmydata/tahoe/trunk$  
darcs apply /Users/wonwinmcbrootles/Desktop/Allmydata\ GNUmakefile\  
signal-error.txt
darcs: bug in get_extra commuting patch:
Tue Mar 18 17:15:36 MDT 2008  robk-tahoe@allmydata.com
   * confwiz: reworked confwiz look and feel

   this changes the confwiz to have a look and feel much more consistent
   with that of the innosetup installer it is launched within the  
context
   of.  this applies, naturally, primarily to windows.

Attachments

• Allmydata GNUmakefile signal-error.txt

I'm not sure what's going on here, but I do notice that darcs 1.0.9rc1 can't
parse this patch bundle, so I presume it's generated by a darcs-2 version? Any
chance we could see the repository that generated this patch bundle?

David

On Mar 21, 2008, at 10:22 AM, David Roundy wrote:

> I'm not sure what's going on here, but I do notice that darcs  
> 1.0.9rc1 can't
> parse this patch bundle, so I presume it's generated by a darcs-2  
> version? Any
> chance we could see the repository that generated this patch bundle?

Hm, it was generated by a new contributor to allmydata.org, using  
this binary:

http://eric.kow.free.fr/download/darcs-1.0.9-i386-linux.gz

Which we found on darcs.net:

http://wiki.darcs.net/DarcsWiki/CategoryBinaries#head- 
c038d6893073983986c777aec342eeead0d3d4da

However, there could be some kind of problems with his system.  Also,  
he hit Down Arrow on his keyboard when typing in his e-mail address.   
Are the funny Down Arrow control characters in that field the reason  
that it doesn't parse?

If I ever talk to him again, I'll ask him if he can give us the 30 MB  
compressed tarball of that repository from his machine, if he still  
has it...

Regards,

Zooko

I don't think the down arrow could cause darcs itself trouble, but it is quite
possible that it made this patch tricky to send via an email program.  It could
even expose a bug in our mime encoding (which as I understand things is not
particularly correct).

David

I just ran into the same error message and was about to report it. I'm using
darcs  1.0.9rc1. For me it happens with a GHC libraries/base repo. I can upload
the offending repo / patch if it helps.
-Tim

Please do, especially if it is small or if you can make a small repo which
exhibits the same behavior.

I just ran into the same bug too, using darcs-2.
Got a patch by mail, saved the whole mail (using Thunderbird) and tried to
apply, failure.

darcs: bug in get_extra commuting patch:
Sun Nov 25 18:44:57 CET 2007  der_eq@freenet.de
  * Added patches to make regex-posix work with 6.8

It worked when I just saved the attached file and applied that.
The repo is http://www.haskell.org/~gentoo/gentoo-haskell/

I'm attaching both files here.

Attachments

• kolmodin-non-working.eml

Lennart, when I try applying this patch, it says that it has already been
applied.  Do you think you could reconstruct a version of the repository where
this failure happens and put it online?  Thanks!

When I try to apply the non-working patch to the repo it doesn't work, even if
the patch is already applied (still using darcs-2).

If you still would like a repo without the patches, it's easy to construct one
yourself by unrecording the patches in the bundle.

Cheers.

Ah, sorry! I had gotten confused and tried to apply kolmodin-working.dpatch
(which, sure enough... worked).  Yes, when I try to apply
kolmodin-non-working.eml, I indeed get the bug in get_extra commuting patch error.

I'd be interested to see if a non-bytestring version of darcs can apply this.

You might be interested to hear this follow-up: I've installed the program
mimedecode on Ubuntu.  When I ran it on the non-working patch, it produced a
file which was very similar to the working patch, except that the piece of
context that talks about this file:

Sun Nov 25 18:44:57 CET 2007  der_eq@freenet.de
  * Added patches to make regex-posix work with 6.8

had an extra newline after the patch comment.  Removing that newline by hand
allowed the patch to apply cleanly.

I guess we need to looking to the bit of darcs code that MIME encodes these
patch bundles?

Attachments

• non-working-decoded.dpatch

Tim: I have a hunch that your bug is just issue279, which has been fixed by
darcs2, and that the one zooko and kolmodin are experiencing is different (and
hopefully more minor than it looks).  So it would be good to see if you still
get the same error message with darcs 2.

I don't understand what goes on here. Is this really two issues: The first being 
the one reported by zooko and kolmodin and the second being the one reported by 
tim? And is the issue reported by tim just another case of issue279 (which also 
relates to issue175 and issue256)? And has that latter issue been solved, thereby 
solving the problem experienced by tim? Answers to these questions and any other 
comments to this issue would be most welcome.

Thanks and best regards
Thorkil

So I think this bug involves patch bundles that have somehow gone through some
kind of corruption in transit (Zooko and Lennart being affected).  

I've done a little bit of work trying to understand get_extra and now have a
slightly better handle on the problem (see
http://wiki.darcs.net/Troubleshooting#bug%20in%20get_extra%20commuting%20patch )

Now we still need to spend some more time characterising the nature of this (see
the wiki for other examples).  I believe it's just another case of mistaken
identity in patches but could be wrong.

What would be nice is if we could find a way to reproduce the problem, not just
the error message, but recreate the conditions that allow such bad patch bundles
to be generated.

No longer reproducible and probably the standard problem with darcs-
1/darcs-2 formats (i.e. failed permutivity).