SoC git projects announced

Hello,

I'm sure those involved already know, but for the rest of us - it looks like we have three projects in the SoC (assuming I'm reading it right):

http://code.google.com/soc/git/about.html

Good news eh?

Andy

Andy Parkins <andyparkins@gmail.com> writes:

Congratulations to people involved.

After recently noticing a few horrors in the existing code that were slipped in the codebase a long time ago, I have been thinking about a bit tighter patch application criteria for everybody (not just student projects). Both mentors and students, please work hard ;-)

And always, have fun!

Andy Parkins <andyparkins@gmail.com> wrote:

Google granted us 3 student projects this summer. We had 18 applications this year, and at least 8 of them were really quite good. Our summer of code mentoring 'team' (Sam Vilain, Johannes Schindelin, Petr Baudis, Martin Langhoff, Martin Waitz, Brian Gernhardt, and I) were really hoping that Google would grant us more projects, as we really wanted to accept them.

Google accepted over 900 students this year, from a pool of over 6000 applications. Competition for slots was definately fierce. Organizations received about 15% of their total applications as slots; Those orgs that were more popular (and received more applications) were given more slots than those that were less popular. I guess Git isn't very popular. :-)

I'd like to thank our mentors for their efforts during the application review process, and all 18 of our students for applying. It really was a very difficult decision for us to pick just 3.

So with all that said, I'd like to welcome our students, and encourage them to get in touch with their mentors. Though I think Dscho might be in the process of moving between countries right now... so he might not be too responsive to email for a few more days.

[Cc: Shawn O. Pearce <spearce@spearce.org>, 
     Andy Parkins <andyparkins@gmail.com>, git@vger.kernel.org]

Shawn O. Pearce wrote:

Could you tell us which projects were applied to (and if there were more than one application per project, how many applicants there were), and which ones of that were good?

I do hope that at least some of applicant which wasn't accepted by Google would do some work on git, perhaps not to the extent of the whole [paid] summer...

Jakub Narebski <jnareb@gmail.com> wrote:

Sure. The other projects were (and in no particular order except the way my brain functions on Fridays):

- Shell script to C conversions: 6 other projects submitted.  Those 6
  are in addition to the one that we did accept by Carlos Rica
  Espionsa, so we actually received 7 applications for this project.

  Dscho and I wanted to accept one more of these projects, but we
  just ran out of slots.  That other student has already said he
  is still interested and may try to contibute a little bit this
  summer anyway.

- Git backed Subversion server: 2 projects submitted.  One of the
  students has started to become a regular on #git, and has said
  he may still work on this project, assuming he can find the time.

  That student's project was also one that we wanted to accept,
  but again, we just ran out of slots.  Martin Langhoff had
  volunteered to mentor him, along with Sam Vilain.  So we may
  still see something from this.

- EGit enhancements: 1 project submitted.  I wanted to accept this
  project too, but throw new google.NotEnoughSlotsException().

- .gitlink lightweight checkouts: 2 projects submitted.  One of these
  projects had a very well written proposal.  It got ranked pretty
  high, and was one we were considering accepting.  But uh, ENOSLOT.

- Blame based merge strategy: 1 project submitted.  Personally I
  like the idea of this project, so I had I bias to wanting to see
  it get accepted.  But we had a merge conflict on slots... :-)

- More complete test suite: 1 project submitted.

- Revision control plugin for GNOME GVfs/Nautilus: 1 project
  submitted.  This particular project wasn't in our ideas list.
  We liked it, we just did not feel we had a mentor who could do the
  project justice.  A little too far away from our area of expertise.

- Managed C# implementation of Git for Mono: 1 project.  This student
  actually submitted this project also to Mono, and Mono actually did
  pick it up (with Miguel as the mentor).  Since Mono got more spots
  than we did, and accepted him, I'm happy we didn't accept him.  :-)

  I've actually gotten in touch with this student, and with Miguel,
  and offered up what co-mentoring assistance we can provide,
  since the student is reimplementing Git in C#...

As you can see, we did have a recurring theme, ENOSLOT. Google only offered us 3. We had 8 on our short list. The 3 that we did select were partly just luck on the student's part, as to some extent we really did just flip a coin to select between the available project ideas. It was a very difficult decision for us.

In the end it really came down to the following reasons why we selected the 3 particular projects that we did:

- We only had one Gittorrent proposal, and it was very good.
  Srijak Rijal was working off-list with Sam during a good part of
  the application review process to discuss it.  Gittorrent is a
  nice idea that has had a difficult time getting off the ground.
  Perhaps having Srijak work on it full-time this summer will
  do just that.

- Dscho really wants to get more of the core utility programs
  rewritten in C, as it makes the system more portable, and easier to
  work on.  (His opinion, which I share.  Junio is free to disagree,
  and sometimes does.)  We had a really good application for that
  from Carlos Rica Espinosa, and again, Carlos worked through the
  application review period on it with Pasky, Dscho and myself.

- I want a library API for Git, from C, as sometimes dammit its
  easier to call the C routines than it is to deal with pipelines.
  The only application we received for this was from Luiz Capitulino,
  it was excellent, and he discussed the topic to some length here
  on the mailing list, and off.

In all three cases the students showed (even if only off-list) qualities that some of our better on-list contributors show: willingness to work, initiated things on their own, are smart, capable of analyzing the problem and producing a solution, they followed through on things, etc... It was difficult for us to not appreciate these qualities in them.

Of course they were not the only students that showed those qualities. Many of the projects we didn't accept were equally good.

Trust me, this selection process has been very difficult on me. I haven't felt well all week, simply because we couldn't take everyone we wanted to. I have felt (and am stilling feeling) pain for the students we could not accept.

Shawn O. Pearce wrote:

It's bit a pity that EGit projects was not accpeted as Eclipse project (perhaps it was not submitted also to Eclipse; but IIRC Eclipse had better Git integration as a project).

I have put preliminary version of http://git.or.cz/gitwiki/SoC2007Projects If I remember correctly one of the requirements was publishing a git repository, perhaps using repo.or.cz...

Shawn O. Pearce wrote:

Could you please add those ideas (projects) to SoC2007Ideas wiki page?
  http://git.or.cz/gitwiki/SoC2007Ideas
Thanks in advance.

Hi,

On Thu, 12 Apr 2007, Shawn O. Pearce wrote:

Right. I am at sea at the moment, but expect to be feet-dry tomorrow. I hope to get settled by Sunday, and a regular Git list reader by Monday.

Ciao, Dscho