Folks,
I've been fiddling around with building my Subversion repository, and I was
blowing it away and rebuilding it so often I made a script to do it
automatically. But now I've run into a weird situation. Here's what I'm
trying to do in my script:
1) Blow away the directory containing the repository and run svnadmin create.
2) svn import from a skeleton directory layout (i.e., dirs only).
3) svn import from existing source into one of the dirs created in #2.
4a) svn2cvs --dumponly some CVS sources.
4b) Hack the dumpfile a tad.
4c) svnadmin load the dumpfile with --parent-dir into a different dir from #2.
Whenever I do this, svnadmin load blows up and tells me to run a recovery
(i.e., DB_RUNRECOVERY error). Obviously I tried removing #4b to see if that
helped; it didn't. But here's the odd thing: if I remove #3, it works. If
I remove #4 it works. If I swap #3 and #4, it works. It only fails doing
both steps in that particular order (with or without my hack).
I've noticed that cvs2svn really really wants to create a repository from
scratch; is there some reason it doesn't like loading into a repository that
already contains some files? (Note: #3 and #4 land in two entirely
different paths in the repository.) Is there perhaps some way to tweak the
dumpfile to fix this (I tried tweaking the "Revision-number" lines with no
success)? I could live with reversing the two steps, I suppose, but it
isn't entirely accurate from a chronological POV, and besides this is (as I
mentioned before) mostly a learning experience for me.
TIA for any help.
-- Buddy
---------------------------------------------------------------------
To unsubscribe, e-mail: users-unsubscribe@subversion.tigris.org
For additional commands, e-mail: users-help@subversion.tigris.org
Received on Tue Jan 13 00:20:33 2004