Re: "svn commit" performance

I am thinking about the semantic having a repository checked out
to a subdirectory of another repository's wc, since i made a
patch for #695 whcih does some checks for the parents of the
target path of checkout.

# svn co http://server1/path/project1 /tmp/project1
# svn co http://server2/path/project2 /tmp/project1/lib/project2

maybe we could add svn:externals if we find that project2
is being checked out somewhere inside project1's wc?

then have a svn_wc_condense_targets that does not condense
subdirectories that are not from the sam repo.

anyway, maybe the commit tree lock should be fixed first.

Cheers,
CLK

On Sat, Jun 21, 2003 at 12:10:05PM -0500, cmpilato@collab.net wrote:
> > 'svn commit foo/bar foo/bar/zig/zag'
> >
> > where foo/bar/zig/zag is a child of foo/bar, but comes from a
> > different working copy. I believe that in the past Subversion
> > silently dropped the second argument and only committed foo/bar.
>
> Ah. Well, I can tell you why *that* happens. Look no further than
> svn_path_condense_targets(), which is going to drop foo/bar/zig/zag as
> redundant in light of foo/bar and recursivity.

---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Received on Mon Jun 30 13:55:16 2003