Leo - thanks for clues. We should do as much of this ourselves; I just
hoped that infra might have this automated for SourceForge from before.
I'll test out the process then say when the import capture is going to
happen so everyone knows the changeover point.
I've already got rsync backups of CVS and SVN. It's the cvs2svn that's
new to me.
If anyone has any particularly active places, then do say so and we can
find a way to either minimise the period of disruption or do that area
separately. I'm going to be particular affected on TxTDB.
Andy
On 13/05/11 10:46, Leo Simons wrote:
So, I asked around on IRC. It seems the precise procedure for doing
these imports isn't very documented, but here's what I gather.
Infrastructure do not import CVS. The simple suggestion is to tar up
the relevant parts of the CVS repository and add a tarball of that
into SVN. If you want more than that, grab an rsync of sourceforge CVS
[1], use cvs2svn to make a svn repo, then create a svnadmin dump file
from that [2].
To import SVN, provide infrastructure with a svnadmin dump file. Grab
your SVN repo from sourceforge, and run svnadmin dump on the bits that
you want imported. Gzip up the dump file and scp it to your local SSH
user account on people.apache.org [3]. Update the jira ticket with the
location of the dump file (along with where it should be imported).
Infra cannot process the jira ticket (INFRA-3628) without the dump
files. So a bit of work is required for someone that's an admin of the
sourceforge project!
cheers...
Leo
[1] http://sourceforge.net/apps/trac/sourceforge/wiki/CVS#Backups
[2]
http://sourceforge.net/apps/trac/sourceforge/wiki/SVN%20adminrepo#ImportingfromotherreposincludingotherSCMs
[3] I'm simply assuming this is a reasonable place. Anywhere
web-accessible would be fine I guess.
On Mon, May 9, 2011 at 8:20 PM, Andy Seaborne
<[email protected]> wrote:
https://issues.apache.org/jira/browse/INFRA-3628
Andy
