Problem with mq extension - qpop then qpush fails

Paul Moore p.f.moore at gmail.com
Tue Apr 24 10:57:10 CDT 2007


On 24/04/07, Patrick Mézard <pmezard at gmail.com> wrote:
> Paul Moore a écrit :
> > On 24/04/07, Giorgos Keramidas <keramida at ceid.upatras.gr> wrote:
> >>>    No username found, using 'UK03306 at UK004134.uk.int.atosorigin.com' instead
> > And indeed if I set the [ui] entry, I still get the problem. There are
> > no rejects that I can see.
> >
> > Actually, I wonder - could this be related to CR/LF issues? I'm
> > working on a Windows box, and (for whatever reason) I have CRLF line
> > endings.
>
> It is very likely related to CRLF issues. Please look at
> <http://www.selenic.com/mercurial/bts/issue483> for details.

Bleh. Thanks for the pointer, it does seem likely this is the issue.
This sort of nasty bug is a real pain - it's horrible to get right,
but pretty vital to providing a painless Windows experience.

(I wouldn't mind, but apart from Notepad, everything on Windows can
cope quite happily with Unix LF line endings these days. You just
can't make all that history vanish, no matter how much you may want to
:-()

> I work with mq without encoding filters, using GNU-Win32 patch.exe and
> with a patched version of mercurial/patch.py : the patch call includes a
> "--binary" option (see an old post from Andrei Vermel about this one).
>
> Andrei also have a patch for transplant I think.
>
> Please ask if you need more configuration details.

I may try to set things up, but it's probably not worth the effort.
The code I'm working with is ultimately from a Subversion repository
and I'm only using hg to track offline work. I was planning on using
mq so that I could keep a queue of patches to commit back to svn one
at a time.

It's not important enough to spend lots of time getting it to work.

Thanks,
Paul.



More information about the Mercurial mailing list