threads / discuss / 26982

GSoC 2011 : one more interested student.

Subject: GSoC 2011 : one more interested student.

## tl;dr

6 messages between Apr 4, 2011 and Apr 7, 2011.

replies: 5people: 5as markdown or json

Cyril Roelandt· Apr 4, 2011, 00:38 UTC · lore

Hi.

I am currently studying Computer Science and Engineering. I have been using Open Source software for quite a while now, but I have not contributed a lot of code to Open Source projects - except for a few patches to Gnome.

Being a git user, I would be interested in working on this project. I believe that taking part in the GSoC would be a good opportunity for me to discover the guts of a version control system.

* Complete some libgit2 features
Implementing a few key features seems like a great way to understand the 
way a VCS works. I would be interesting in working on the diff command 
and on the network related commands (git-push, git-fetch).

* Build a minimal Git client based on libgit2
It seems to be a quite popular task :) Hacking on this small git client 
would be a nice experience and would help me understanding how libgit2 
works.

* Build in more external commands
I am wondering why these commands should be rewritten. You know what 
they say : "if it is not broken, do not fix it" :-p Do these commands 
currently have performance issues ?

I will be working in a research lab until the end of June, but I will have more free time in July/August, that I could use to hack on git, which means I may have less time than most US students, who can enjoy a slightly longer summer break. Do you think it would be possible to work one of these projects anyway ?

Best regards, Cyril.

Nguyen Thai Ngoc Duy· Apr 4, 2011, 12:05 UTC · re: Cyril Roelandt · lore

Re: GSoC 2011 : one more interested student.

On Mon, Apr 4, 2011 at 7:38 AM, Cyril Roelandt <tipecaml@gmail.com> wrote:

> * Build in more external commands
> I am wondering why these commands should be rewritten. You know what they
> say : "if it is not broken, do not fix it" :-p Do these commands currently
> have performance issues ?

On Windows, yes. Shell scripts are just not designed to work on Windows. Bash and Perl also pull msys and perl packages in Windows installation, which makes it big and less Windows friendly. Building in commands also give them a higher chance of being libgit2-ifized later on.

-- 
Duy

Ramkumar Ramachandra· Apr 4, 2011, 15:57 UTC · re: Cyril Roelandt · lore

Re: GSoC 2011 : one more interested student.

Hi Cyril,

Cyril Roelandt writes:

Show 8 quoted lines

> I am currently studying Computer Science and Engineering. I have
> been using Open Source software for quite a while now, but I have
> not contributed a lot of code to Open Source projects - except for a
> few patches to Gnome.
> 
> Being a git user, I would be interested in working on this project.
> I believe that taking part in the GSoC would be a good opportunity
> for me to discover the guts of a version control system.

Thanks for writing.  We'd love to see more student proposals.

Show 9 quoted lines

> * Complete some libgit2 features
> Implementing a few key features seems like a great way to understand
> the way a VCS works. I would be interesting in working on the diff
> command and on the network related commands (git-push, git-fetch).
> 
> * Build a minimal Git client based on libgit2
> It seems to be a quite popular task :) Hacking on this small git
> client would be a nice experience and would help me understanding
> how libgit2 works.

Yes, libgit2-related work in general seems to be very popular among GSoC students this year :) Not to worry- you can make your application stronger by sketching out a solid proposal and showing us some code.

> * Build in more external commands
> I am wondering why these commands should be rewritten. You know what
> they say : "if it is not broken, do not fix it" :-p Do these
> commands currently have performance issues ?

As Nguyen already pointed out, the rewrite is mostly motivated by portability reasons.

Show 5 quoted lines

> I will be working in a research lab until the end of June, but I
> will have more free time in July/August, that I could use to hack on
> git, which means I may have less time than most US students, who can
> enjoy a slightly longer summer break. Do you think it would be
> possible to work one of these projects anyway ?

>From my personal experience, most of the GSoC ideas listed on the Git

wiki page are quite demanding. They're almost as engaging as a full-time job. I'd be especially worried about the mid-term evaluations. However, if you feel that you can take on a project, please do so by all means -- include a convincing timeline in your proposal.

Good luck with your application.

-- Ram

Junio C Hamano· Apr 4, 2011, 18:11 UTC · re: Ramkumar Ramachandra · lore

Re: GSoC 2011 : one more interested student.

Ramkumar Ramachandra <artagnon@gmail.com> writes:

> From my personal experience, most of the GSoC ideas listed on the Git
> wiki page are quite demanding.  They're almost as engaging as a
> full-time job.

GSoC is _supposed_ to be a full-time job during schools' summer break and you are expected to devote 40-hours a week (I checked with people who run the program last week).

Junio C Hamano· Apr 4, 2011, 22:31 UTC · re: Cyril Roelandt · lore

Re: GSoC 2011 : one more interested student.

Cyril Roelandt <tipecaml@gmail.com> writes:

Show 9 quoted lines

> * Complete some libgit2 features
> Implementing a few key features seems like a great way to understand
> the way a VCS works. I would be interesting in working on the diff
> command and on the network related commands (git-push, git-fetch).
>
> * Build a minimal Git client based on libgit2
> It seems to be a quite popular task :) Hacking on this small git
> client would be a nice experience and would help me understanding how
> libgit2 works.

It seems that libgit2 project got GSoC slots out of the git project even though they didn't apply themselves ;-) But if you are interested in the git project proper, there also are ideas on the wiki page.

> * Build in more external commands
> I am wondering why these commands should be rewritten. You know what
> they say : "if it is not broken, do not fix it" :-p Do these commands
> currently have performance issues ?

Performance won't be the primary issue; we would want to lose POSIX shell and Perl dependency for Windows folks and that is the primary gain from that part of the project.

Jeff King· Apr 7, 2011, 00:53 UTC · re: Junio C Hamano · lore

Re: GSoC 2011 : one more interested student.

On Mon, Apr 04, 2011 at 03:31:01PM -0700, Junio C Hamano wrote:

Show 8 quoted lines

> > * Build a minimal Git client based on libgit2
> > It seems to be a quite popular task :) Hacking on this small git
> > client would be a nice experience and would help me understanding how
> > libgit2 works.
> 
> It seems that libgit2 project got GSoC slots out of the git project even
> though they didn't apply themselves ;-)  But if you are interested in the
> git project proper, there also are ideas on the wiki page.

Just to be clear, nothing has any slots yet. We will get a bunch of proposals (and yes, there are a _ton_ of libgit2 proposals), and then the mentors as a group will rate them and decide which ones we want. And then Google will tell us how many slots we get, and if we have more good proposals than slots, we decide what to cut.

So libgit2 may get zero slots, but given the interest, I hope that it will get something.

We could potentially spin libgit2 off as its own organization next year. I've been surprised by how much student interest there is in it this year.

-Peff