Hi, On Tue, Oct 29, 2013 at 08:20:59PM +0100, Franck Villaume wrote: > Le 28/10/2013 21:18, Sylvain a écrit : > >I started a discussion with some Savannah Hackers about migrating > >Savannah to FusionForge. > > > >Context: the Savane project didn't find a new maintainer after I > >progressively withdrew from it, and I think it may be good thing to > >officially deprecate Savane in favor of something else. It's not news > >- for the past 2 years a guy had some wish to mentor a GSoC student to > >work on the stalled Python/Django rewrite (though didn't eventually), > >but each time I pointed it would make more sense to move to > >FusionForge, which was the closest active free project. And IMHO still > >is :) > > > >Since I'm going to work on FusionForge next year, I thought it would > >be a good opportunity to bite the bullet, officially deprecate Savane, > >and help Savannah "upgrade" to FusionForge along the way. > > > >Note that I'm not officially part of the Savannah Hackers for a couple > >years now, so the decision is not mine. Also they may think newer > >alternatives are more suitable (Redmine, Allura?). Last, this is > >unrelated to my work on FusionForge next year - I pointed them the > >opportunity to switch to a project I'd be working on, but this > >Savannah work would be on my personal time anyway, just like now. > > > >Anyhow, at this point of the conversation, the decision in mixed with > >various other issues, from finding more sysadmins for the Savannah > >platform, to thinking about building a separate platform from scratch. > > > > > >The goal of this mail is to mention this potential change, but more > >importantly to ask for your thoughts about such a migration from > >Savane to FusionForge: do you expect any high incompatibilities (that > >would make a migration script particularly hard)? Missing features > >(trackers triggered transitions comes to mind, but they aren't used > >much)? Architectural diverging choices (chroots maybe)? > > > >Feel free to ask me for more details on the current setup. > > > >I've been experimenting a little with FusionForge, but I'd be glad to > >hear what more experienced devs think about this :) > > Hi Sylvain, > > If features are missing, maybe it's time to implement them into > Fusionforge. > About migration, I do not think about incompatibilities but maybe > some heavy changes. Any documentation available about savane > features ? > > Can you share with us some informations about the current savannah > deployment ? (general architecture : is source code hosted on the > same machine as the php frontend ?) > > How can we help to make the decision happen ?
A good starting point for the infrastructure would be: http://savannah.gnu.org/maintenance/SavannahArchitecture/ Subsystems (vcs, downloads, front-end, etc.) are in separate virtual machines. The only link between all VMs is access to the database. The decision is currently depending on who is going to maintain Savannah in the future (dedicating a sysadmin from FSF part-time is considered, in addition to several volunteers) - and what tool they will choose, be it by upgrading the current infra, or progressively migrating to something entirely different. Or maybe the current team will get tired of discussing and just migrate to FusionForge ;) One thing that will help is assessing how difficult it will be to migrate. In particular, did any of you already wrote migration scripts for the forges fusioned in FusionForge? (I myself wrote migration scripts to the forsaken Savane 4, but relying heavily on having both source and destination databases in a single MySQL instance. Here will have to switch to PG as well. http://git.savannah.gnu.org/cgit/savane.git/tree/migrate_old_savane.sql) Cheers! Sylvain _______________________________________________ Fusionforge-general mailing list [email protected] http://lists.fusionforge.org/cgi-bin/mailman/listinfo/fusionforge-general
