[Live-demo] Google Summer of Code ideas for the LiveDVD

[Hamish]

Cameron wrote:
> I've set up a page for Google Summer of Code ideas here:
> http://wiki.osgeo.org/wiki/Live_GIS_Disc_GSoC_2010
> 
> What other ideas should we add?

Keep in mind that documentation writing, website overhauls, and other non-
coding chore-work, while important, are specifically outside the scope of
GSoC. One of the requirements to get paid is a code deliverable which
the student need to upload to Google at the end of the summer, which
contains their & only their work. Which in a practical sense & for better
or worse mostly means it is useful for succinct lone-wolf style projects.

On the other hand, projects which bring together multiple within-mentoring
organization projects (& dev teams) as well as multiple mentoring-orgs is
to be encouraged. (I suspect gaining buy-in and SCM check-in trust from a
wide pool of projects will require a special & widely known and hands-
on mentor, otherwise it makes me nervous about giving the student a
project which is destined to stall.)

So working to get test-suite* & buildbot scripts implemented in all OSGeo
projects could be a suitable project, but a project to create a common
dataset probably would not be funded.

[*] (but isn't that supposed to be already-solved as an incubation
milestone?)


> Hamish, I wonder whether you would want to mentor a student
> who migrates liveDVD packages to .deb files?

The SoC admin team (which includes me) have decided this year to split
our time amongst all the student projects instead of trying to that as well.
And so I won't be a primary mentor on any project, and would only consider
backup-mentoring if it dealt with code which I was the sole maintainer of,
& so was absolutely necessary.

Putting stuff in a .deb package is rather simple. Getting stuff packaged
to pass Debian QA review is a whole other story and much more important.
(the great advantage of .deb over .rpm has very little to do with the
technical merits of the two package formats, & has everything to do with
the standards & peer-review process, IM(biased)O)

To do a good job the mentor for such a project should really be a full
Debian Developer (DD) already, which I am not. You could ask for
volunteers on the DebainGIS mailing list if you like, but in most cases
I suspect it will mainly come down to debianizing Java packages, which
really means working to get those Java programs to work with a version of
Java which does not require a click-through agreement. Either by patching
the program or by helping to improve the open JRE itself.

So while on the surface this looks like a .deb project, I really think
under the surface it's an inter-project Java one. Once the java problems
are solved the .debs can flow relatively easily..


regards,
Hamish