|
|
1
2
3
|
|
Editor: mjflick
Time: 2012/08/18 04:25:22 GMT+0
|
|
Note:
|
changed:
- CvS - GNUArch/UsingGnuArch - Git - [Hg] - SvN/SvNTasks
-
[CVS] - GNUArch/UsingGnuArch - Git - [Hg] - SvN/SvNTasks
If you'd like to help, please contact savannah-hackers-public@gnu.org or join #savannah at FreeNode?.
Tasks needed for base support - done:
- DONE Support SVN in Savane
- DONE Clean-up the architecture to allow new SCMs?
- DONE Type of write access: svn+ssh (provides more isolation than a site-wide WebDAV? access where all the repositories belong to www-data)
- DONE Think where to place SVN data: along with the CVS repositories seems good - we reuse the IP and the ViewCVS?/ViewVC? install. It's the same kind of access, with no user login allowed (so as to preserve history integrity).
- DONE Document migrating CVS to SVN: SvN
- DONE: alter the Savane repository creation - do not try to enforce a repository layout
Tasks needed for better maintenance:
- TODO Automatically import SVN dumps on user request (as SF does - http://sourceforge.net/docs/E09#import) - beware of data loss if the project admin makes a mistake!
- TODO? Performance tweaks for serving 2500 repos at once: use ramfs for temporary files (goal: no hard-disk access during read-only access [I don't think SVN has read locks, though, unlike CVS, and thanks to the transaction / atomic commit design]?).
- TODO On-commit e-mail notifications (aka hooks management - http://sourceforge.net/docs/E09#scripts)
- TODO Automatically migrate a CVS repository to SVN repository on user request; cvs2svn comes to mind, but see other tools at SvN
- TODO Your suggestions :)
Other SCM at Savannah
CVS - GNUArch/UsingGnuArch - Git - Hg - SvN/SvNTasks