Ramkumar Ramachandra wrote on Sun, Sep 26, 2010 at 01:48:18 +0530: > Hi Daniel, > > Daniel Shahaf writes: > > Daniel Shahaf wrote on Sat, Sep 25, 2010 at 16:49:08 +0200: > > > 0:/tmp/svn% $svnrdump dump file://$PWD/r1/trunk/A/B 2>&1 | grep > > > Node-path > > > Node-path: trunk > > > Node-path: trunk/A > > > Node-path: trunk/A/B > > > Node-path: trunk/A/B/E > > > Node-path: trunk/A/B/E/alpha > > > Node-path: trunk/A/B/E/beta > > > Node-path: trunk/A/B/F > > > Node-path: trunk/A/B/lambda > > > > > > Nice :-) > > > > > > > Is it? If I try to dump a subdir, then I /shouldn't/ get a "Create > > /trunk" entry in the dumpfile... instead, it should be the user's > > business to create that directory in the target repository out of band, > > before loading the dump into it. > > svnrdump behaves exactly like svnsync in this manner.
svnsync allows you to sync a subdir of a repository (i.e., svnsync $REPOS/trunk/A/B $MIRROR ), but does it also create /trunk/A/B in the mirror? But for now I still think that svnrdump invocation I quoted above shouldn't have outputted a 'create /trunk' entry in the dumpfile :-). @all, what do you think? > See the tests > only_trunk_dump, only_trunk_A_with_changes_dump, and > descend_into_replace_dump that I imported from svnsync. They all > exercise this subdirectory feature. Maybe the tests are not > exhaustive? > > -- Ram
