[Savannah-hackers] what would you like to see

[Rudy Gevaert]

I am savannah-hacker, but only take care of moderation/support/bugs.
But:

I will try to do my masters thesis on Savannah.  If I may from my
university and if I can find a promotor (this will be the hardest
part).  This would officially start next year in September.

The best way for me to be able to do my thesis on Savannah is when I
can propose some scientificly subjects about it to my promotor.

I can't say:  I'll do the moderation for a year, fix bugs and respond
to support requests.  They won't make it :)

So if anybody has some ideas, just let me know.

Nic Ferrier already gave me the following idea:
I think it would be a pretty interesting task to rebuild it as a
distributed application:

- sessions stored in the database, this means sessions should not
  contain complex objects (no instances of classes) because
  serialization of them to database types is such a pain; there is a
  lot of analysis of just that decision that could provide a good
  thesis.

- the splitting of the CVS structure to allow massively parallel
  hosting would also be interesting. Simple hashing could do the
  job. One could also think about redundancy (keeping a single CVS
  repository on multiple machines).

- the addition of new applications via web interfaces ("web
  services") is another interesting area: for example all the apps
  that will happen when I get the RSS feeds going.

I think all of these together would make an interesting thesis with
the title:

  "Distributing load: moving the GNU/CERN savannah system from a
  single server to 20; technical, administrative and social issues."


Mathieu Roy contributed the following:

Making savannah a distributed tool as you told me some days ago may be
good starting point.
Maybe you should, if you interested in, implement the mail interface
to the trackers (that we are currently heavily working on) with gpg
authentication. It may be an interesting work and no so trivial,
because it requires to understand a technology of the future (gpg).

As there are many people on this list, I'm looking forward to some
ideas.

-- 
Rudy Gevaert                address@hidden              
Web page                    http://www.webworm.org
GNU/Linux for schools       http://www.nongnu.org/glms
Savannah hacker             http://savannah.gnu.org