james-tigris@jrv.org writes:
> I have a bunch of old PVCS (the old DOS version) archives I've like to
> place in an SVN repository. I'd like to retain the data and the
> authorship and timestamp information and place it in an existing
> repository.
>
> I've got a program that is producing an output essentially similar to
> the output of 'svnadmin dump'. Is this the right direction to go -
> can 'svnadmin load' safely merge a dump stream into a repository?
It seems like a good way to go, yes. 'svnadmin load' has no other
purpose than to merge a dump stream into a repository :-).
> If so I have a few questions:
>
> 1. What do I do for a UUID, or should I omit that line altogether?
Just omit it.
> 2. What do I do for the Revision-number: fields?
Well, you need to make up revision numbers. Just start with 1 and
increment. If you're merging into an existing repository, the numbers
will adjusted as they go in (and the 'svnadmin load' output will make
clear that this is going on).
As for timestamps, I'm not sure what will happen with those. You
might want to try an experiment first (and let us know the results).
> If this is an undesirable approach for whatever reason can this be
> accomplished via repeated commits on a client to get the data into the
> repository, followed by propsets to change the timestamps and
> authorships?
You can always change revision properties (such as timestamps and
authors) following a load, just as you could following a series of
regular client-originated commits. I.e., these two things are
unrelated: the method by which the data got into the repository, and
what you do with that data afterwards.
-Karl
---------------------------------------------------------------------
To unsubscribe, e-mail: users-unsubscribe@subversion.tigris.org
For additional commands, e-mail: users-help@subversion.tigris.org
Received on Thu Aug 21 17:05:49 2003