Getting Ensimag students to work on Git for a few weeks

Hi,

I'd like to propose a kind of mini-google summer of code to the students of the the school where I teach, i.e. Ensimag, France ( http://ensimag.grenoble-inp.fr/ ). In short, this means having a few students working for Git for a month at no cost ;-).

Currently, the students have an end-of-year project (in equivalent of master 1) with the choice between many subjects, some of them being somehow "real-life" (i.e. actually usefull things), and other being artificial (i.e. enjoy doing it, and throw it away afterwards).

This year, I'd like to propose a subject "contribution to an existing free software", and since the one I know best currently is Git, this would take the form of "contribution to the Git project". I'd see the practical organization a bit like the google summer of code: chose a feature (the GSoC proposals on the wiki can be a good source of inspiration), and implement it with the goal of being eventually merged upstream. There would be no money involved, but the students get a grade at the end. I would anyway follow the work of the students, but a co-mentoring from a Git expert would be great.

The students work full-time for about 3 weeks (May 20th to June 16th), and are grouped by teams of 2 to 4 students. Given my bandwidth, I plan to propose only one group of 4 students this year, but we may scale up later, who knows.

We have plenty of time before this starts, but I'm just sending this email to get your feeling on it.

Any opinion? Do you like the idea?

Hi,

On Tue, 27 Oct 2009, Matthieu Moy wrote:

Do not underestimate the cost of time and nerves, both on the student's and the mentor's part.

I spent an insane amount of time on the Google Summer of Code this year, a relatively small (but not less frustrating) part of which was with Git.

Ciao, Dscho

Johannes Schindelin <Johannes.Schindelin@gmx.de> writes:

Hey, you're talking to a teacher ;-).

Yes, for sure, this takes a lot of time. But on my side that's my job, the school gives me time to mentor the students. And I'll be geographically close to the students, having regular physical meetings, ... I'd appreciate a co-mentor, but the feedback from the mailing list (like the one for non-school-related contributors) plus my mentoring can be sufficient.

On Tue, Oct 27, 2009 at 11:12:52AM +0100, Matthieu Moy wrote:

That's not much time to get familiar with a complex project like git. So you will have to do something extremely simple, which probably means that it won't be anything exciting. If it were, someone else would have done it already.

Clemens

Clemens Buchacher <drizzd@aon.at> writes:

That very much depends on which students. Some of our students are already experienced programmers when the enter the school. If I manage to attract such students, a group of 4 can do a bit more than "extremely simple".

Depending "exiting for who" ... my first contributions to free software were very trivial code, and I found it very exciting the day I got 20 trivial lines of code of mine into GNU Emacs ;-).

So, the feature may not be exciting, but I think the experience for students could be.

Clemens Buchacher <drizzd@aon.at> writes:

Let's take a look at some projects from SoC2009Ideas and Wishlist:
* "smart" HTTP transport is being actively worked on
* narrow / sparse checkout is being worked on
* directory renames development stalled, I think last was in
  http://thread.gmane.org/gmane.comp.version-control.git/99529
* git-svnserver has supposedly partial Python implementation
* restartable clone, which should be fairly easy to add to "dumb"
  protocols, and quite challenging to add to "smart" protocols;
  even without a code, having fresh ideas would be nice
* (optional) support for empty directories, needs index extension,
  there were some patches, but the area might be muddy

It all depends on what you want to achieve in this short time. Would it be to get to know OSS development workflow (submitting patches, answering reviews, etc.), or would it be solving interesting real-life problem, or perhaps solving some problem from beginning to the end (the code being accepted).

On 27/10/2009 11:12, Matthieu Moy wrote:

Have you considered projects contributing to libgit2? The students at ULCO [1] will be mentored by Andreas Ericsson.

Given that libgit2 is quite empty, there are many small tasks like starting commits, trees, tags, etc. objects.

But I guess he won't have time for mentoring more students.

[1] http://dpt-info.univ-littoral.fr/mediawiki/index.php/I2L:Projet#libgit2_-_Herv.C3.A9_Cauwelier_-_itaapy

Hervé Cauwelier <herve@itaapy.com> writes:

Right now, the project is just an idea, so I didn't consider an idea in particular. But I'd love to get people to work on libgit2 (I really think this is _the_ thing that SVN has and that Git lacks), and therefore, I'd love my students to work on this.

But as much as possible, I'll wait and decide the exact project with the students.

Thanks for the pointer.

OTOH, I don't know whether we'd get as much feedback working on libgit2 than working on Git itself.

Hi,

I'm resurecting an old thread :

  http://thread.gmane.org/gmane.comp.version-control.git/131321
  (message copied below)

and welcoming Diane, Axel and Clément (CC-ed), 3 students who have chosen this project. Right now, they're trying to get familiar with the codebase. As a first exercice, they'll try to add textconv support to git-blame. The next tasks are undecided yet (but advices are welcome).

I hope the project will be productive, both for students and the Git community!

On Tue, Oct 27, 2009 at 12:12 PM, Matthieu Moy <Matthieu.Moy@grenoble-inp.fr> wrote:

On Fri, May 21, 2010 at 07:44:42PM +0200, Matthieu Moy wrote:

Just FYI, there is a patch pending to add textconv support to git-gui.

 http://mid.gmane.org/20100415193944.GA5848@localhost

Clemens

Le 22/05/2010 13:26, Clemens Buchacher a écrit :

Thank you for this information.

As we are about to finish to add textconv support to git-blame, we'll now look to add it to git-gui blame. We wonder what would be the best choice between add a second option "Use Textconv For Blames", or transform the existing one (introduced by the patch above) into "Use Textconv For Diffs and Blames" (so the user wouldn't have the choice to use textconv only for Diffs or only for Blames; is it really usefull ?). Your advice is welcome !

Thanks, Clément