My 2 cents...
We were planning on using the concept of vendor branches to support
disconnected offices. I was hoping to use this tool to keep monthly
code-drops in sync. I would much prefer to retain version history, even at
monthly intervals, and it seems like I could generate a diff in either case.
I believe the only real difference is that your algorithm (as I understand
it) won't allow any deltification of the trees within the repository, so
each import will consume space proportional to the size of that import and
not proportional to the size of the changes. This is substantial to us as
our source tree and supporting files hovers at around 200 megs (don't ask).
Did I understand this right?
From: Ben Collins-Sussman [mailto:email@example.com]
Sent: Tuesday, February 24, 2004 3:35 PM
To: Perry E. Metzger
Cc: firstname.lastname@example.org; email@example.com
Subject: Re: svn_load_dirs.pl on win32
On Tue, 2004-02-24 at 14:29, Perry E. Metzger wrote:
> How would you do this? I've never successfully imported a tree in the
> same place twice -- I get weird error messages when I do it. (I
> assumed that it was just the system's way of telling me I was not
> supposed to use import twice for the same sources...)
'svn import' won't overwrite anything in the repository, and neither
will 'svn cp'. Sorry about that, it was a deliberate decision.
Here's the algorithm I was describing:
$ svn import vendor-1.0/ http://host/vendor/vendor-1.0
$ svn cp http://host/vendor/vendor-1.0 http://host/myproject
# weeks go by, I make small changes to 'myproject'.
# vendor releases new version.
$ svn import vendor-1.1/ http://host/vendor/vendor-1.1
# now we have two unrelated trees in the repository. no big deal.
$ svn checkout http://host/myproject
$ cd myproject
$ svn merge --ignore-ancestry \
http://host/vendor/vendor-1.0 http://host/vendor/vendor-1.1 .
To unsubscribe, e-mail: firstname.lastname@example.org
For additional commands, e-mail: email@example.com
Received on Tue Feb 24 21:59:26 2004