Subversion problems

Michael Matz matz at kde.org
Thu Feb 17 21:59:41 GMT 2005


Hi,

On Thu, 17 Feb 2005, Stephan Kulow wrote:

> I think I know how we can combine both. I think I could hack cvs2svn to
> only create an incremental diff against the last imported revision. This
> would mean, we could import all of cvs into svn but mark it read only,
> then mark some of cvs read only and after the next incremental update,
> the svn paths for that cvs modules are open and at the same time we
> could e.g. switch anoncvs to anonsvn (taking longer update cycles for
> that service though).

IIRC we talked about a way, which doesn't need fiddling with 
partial read-only and still reduces commit-downtime to only a few hours:
* create initial svn 
  from a copy of current CVS (takes long, although we have now a fat 
  machine)
* leave CVS open that time, but disallow any creation of tags 
  and branches (!).
* leave svn closed, remember highest rev R.

Then (assuming cvs2svn is appropriately hacked):
* close cvs
* use cvs2svn on it to fill svn ignoring all revision up to R (takes let's 
  say two hours)
* open svn
* done.


Ciao,
Michael.




More information about the kde-core-devel mailing list