subversion-users mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Nico Kadel-Garcia <nka...@gmail.com>
Subject Re: Removing data from subversion
Date Wed, 28 Nov 2018 12:14:44 GMT
On Wed, Nov 28, 2018 at 6:42 AM Ahtesham Kazi <ahtesham.kazi@gmail.com> wrote:
>
> Greetings,
>
> I have read that subversion does not allow deleting obsolete not required data.
> Is there a way we can fork from some point onwards ?

Kind-of sort-of. There's a lot of discussion in the Red Book and
mailing list. To prune content, you'll be committed to switching to a
new Buversion repository. Pruning in place is *dangerous*, it's one of
the oldest requested features, and as best I can tell it's never going
to happen.

The fastest and simplest is to snapshot a working copy of the limited
content that you want and make a new repository, with no history. That
is least likely to include bulky content from commits that touched
multiple branches. Second fastest, if your Subversion content is not
complex or heavily dependent on svn: settings, is to use git-svn to
import the desired branch and only the desired branch, apply "git gc
--aggressive" to the git repo, and use "git-svn" to export it back up
to an entirely new Subversion repository. This works surprisingly
well, though it can get quite confused about "svn:keywords" and
similar attributes.

The trickiest is to do a dump, exclude, and reload into a new
repository. That's better able to preserve the history, unsullied,
with really good provenance of subtle tuning like "svn:keywords". But
I've personally given up on the approach. Subversion is aimed at
conserving history, and the history of embedding bulky content
accidentally or embedding keywords or embarrassing software errors
accidentally and very reasonably wanting them *out* of the repos is
why I've used the other approaches.

Mime
View raw message