On Fri, 17 Jun 2005 13:31:19 -0400, Christopher Ness wrote:
> On Fri, 2005-06-17 at 18:26 +0200, Peter Hercek wrote:
>> I don't think this is the problem. When I omit the last argument the current
>> directory should be used. And the merge actually starts even with the two
>> It failes after working for about 2 minutes and modifying a lot of files in
>> the working copy.
> Hi Peter,
> I kept waiting for this to show up in my list folder, but I see you
> replied off list, maybe to save me some shame of not reading the full
> help page on merge.
:-) I wish I would, I have probably misconfigured news reader on
the other machine, when I posted to the groop it returned back
complaining about illegal header. I'm curious if this will go ok.
> You are correct about W_PATH being . if not set.
> Now, as for why it is failing - I have no idea but your paths seem funny
> to me in the command. What are the changes that are being made? Have
> you tried doing a `svn diff` and getting a patch file out of SVN to see
> what is going on.
> You could then try applying it with the `patch` program. Though this is
> rather annoying. But perhaps patch would throw more interesting errors
> than what `svn merge` is giving you.
> It's not much help, just another way to look at the problem.
This is actually and interesting experiment. I generated the diff
sucessfuly. I'm not sure if it means that the problem is not on
the server. The patch can be applied with some (about 20) rejects.
Unfortunately this is not usable way in general. There were about
20,000 deletes and additions in the diff. To add and remove all
the files manually would kill me. I'll rather try to debug the
error or try to do it other way round: figure out the chagnes
we did and create patch from them ... and then just to kill our
trunk version and use vendor version with our patch ... it
should be less work this way.
To unsubscribe, e-mail: firstname.lastname@example.org
For additional commands, e-mail: email@example.com
Received on Sat Jun 18 01:00:23 2005