"Scott Hughes" <shughes@dalsemi.com> wrote in message
news:d5qndo$n46$1@sea.gmane.org...
> When I migrated my repo over from vss, I did a lot of svndumpfilter
> editing to shuffle all my projects around. Somewhere in there, I forgot a
> "drop empty revs" flag. Now when I edit one of my dumpfiles, I see about
> a hojillion or so of these "empty revs" in the dump. Since I often do a
> full "dump | tar | gzip" for backup, the space isn't that big of a concern
> since I'm sure those messages compress well. However, the time to do this
> backup is actually a bit longer than I think it would be without them.
Well, I ended up writing an app in C# to do the job. It's pretty simple and
easy to modify. So, although it currently only removes padding, it's very
simple to modify it to remove/change any revision given your own
requirements. It's not commented at all, but I think it's pretty short and
readable (much more so, IMO, than svndumpfilter). Though, I can't guarantee
that it handles all records from a dumpfile properly. My specific dump file
was 400 megs and had a good mix of ascii and binary data. I had munged it
up with over 90k records of empty revision padding (do to some multiplying
bunnies as I filtered multiple projects from the same file in a
dump-filter-load loop)... Without the empty revisions, there were only 600
revisions total. It only took a few seconds to process the entire 400 meg
file.
I'll share a copy of the source with anyone who wants it... As long as you
use it at your own risk.
Scott Hughes
shughes@dalsemi.com
---------------------------------------------------------------------
To unsubscribe, e-mail: users-unsubscribe@subversion.tigris.org
For additional commands, e-mail: users-help@subversion.tigris.org
Received on Mon May 16 23:27:48 2005