Philip Martin <philip@codematters.co.uk> writes:
> Branko ÄŒibej <brane@xbc.nu> writes:
>
> > >Negative, should be the same. And both need to be fixed, it seems!
> > >And I wonder that one tweak to repos_diff.c would do the trick...
> > >
> > No. It would fix the merge and cp bug, but would introduce a bug in
> > add (on a directory).
> >
> >
> > Hmmmmm. Lemme see if I can make that bit conditional.
>
> I don't understand (Again! Why is feedback so hard?) How does
> repos_diff.c affect adding directories? Are you referring to a merge
> that adds a directory, with the working copy already having a
> versioned directory of that name? In that case merge does nothing for
> the directory unless it is scheduled for deletion. What should the
> feedback be in that case? Why are directories a problem and not
> files?
I'm not sure, but I think that "svn copy repos_url wc_path" is a
two-step process:
1. basically, `svn co repos_url wc_path'
2. then, we link up the wc_path to its parent dir with a kind of 'add'.
That's why the output is segmented into what looks like two commands:
A newdir1/foo
A newdir1/bar
Checked out revision 45 <---- here's the division
A newdir1
Disclaimer: This is a guess based on ancient memories of Karl's
implementation of `svn copy repos_url wc_path' -- I have not confirmed
this as fact.
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Received on Fri Aug 9 02:18:18 2002