[GSoC] Git projects for Google Summer of Code 2008 final evaluation

We are now according to Google Summer of Code 2008 timeline
  http://code.google.com/opensource/gsoc/2008/faqs.html#0.1_timeline
after final evaluation deadline (September 2: 12:00 PDT / 19:00 UTC).

Therefore I and perhaps other on git mailing list would like to know what is the status of all git GSoC 2008 projects. We could then write summary about each project at http://git.or.cz/gitwiki/SoC2008Projects page.

So students and mentors, please write what do you think current status of project is: is it done, is it ready, is it perhaps already merged in, or can be merged in at any time. Please write also what wasn't done, or what should be done different way (hindsight and all that).

Students, could you please tell us if you plan to work on git further, and in what range (how much time can you get to work on git). What do you think about git development community? What have you learned from participating in GSoC?

Mentors, could you tell us your side about how it was working as a mentor for Google Summer of Code? Perhaps some tricks of trade? Have you herd perhaps about non-Git projects involving Git, like Git# project(s) from Mono, DVCS support for KDevelop from KDE, Git plugin for Anjuta IDE from GNOME; how did they fare?

Thank you all very much for your work on Git projects on this Google Summer of Code.

Below there is list of various Git's Google Summer of Code 2008 
projects, in the sequence the are on Git Wiki
  http://git.or.cz/gitwiki/SoC2008Projects

1. GitTorrent
 
Student: Joshua Roys
Mentor: Sam Vilain

2. git-statistics

Student: Sverre Rabbelier
Mentor: David Symonds

3. Gitweb caching

Student: Lea Wiemann
Mentor: John 'warthog' Hawley

4. Eclipse plugin push support
 
Student: Marek Zawirski
Mentor: Shawn O. Pearce

5. git-merge builtin

Student: Miklos Vajna
Mentor: Johannes Schindelin

6. git-sequencer

Student: Stephan Beyer
Mentor: Christian Couder, Daniel Barkalow

On Thu, 4 Sep 2008, Jakub Narebski wrote:

The code's written, and we've been reviewing it, but the end of the summer kind of crept up on us (me, anyway) along with the 1.6.0 timing, and we didn't get it in mergeable shape in time for the evaluation. Post-evaluation, Stephan's been continuing to clean up the code as we review it. While Christian and I will probably go on finding more things to improve about it, it's probably also ready for a wider audience now.

	-Daniel
*This .sig left intentionally blank*

Jakub Narebski wrote:

Hello all,

VCS::Git::Torrent can only transfer data under limited circumstances at the moment, and doesn't do peering. Part of that is actually due to work done after the deadline, unfortunately for me :-) Anyway, I am definitely interested in continuing to work on git (and GitTorrent) as it is the VCS I use personally. It is, however, the start of the semester here and I am currently focusing on that and job searching as this is my last semester.

I had a really good experience working with the git community over the past few months. It was really cool to see very talented people working together. It was a little intimidating too. Another thing I learned about was the 'O' in OSS, or being 'Open' about development. This is a completely different paradigm for me, as all of my previous experience is personal development or 'corporate' development (things done at work), neither of which required telling anybody about anything until completion. Regular status updates that were requested during the second half made that much more clear and, in fact, facilitated my understanding in this area.

So, thank you all again, and I hope to continue to work with you in the future.

Joshua Roys

Hi,

my reply belongs to:

Jakub Narebski wrote: [...]

It is done, but not ready.

This means, we, my mentors and I, are currently still reviewing the code. But I am pinched for time until the end of September due to an important exam. My mentors also have not much time and so everything is a little slowed down. When they ACK, it will be sent to the list.

During the GSoC application phase I have done a breakdown and I declared two items as "bonus" if I am really quick with the rest. But I wasn't that quick, so none of the bonus items have even begun.

Those were:
 6. [bonus] rewrite the users in C
 7. [bonus] discuss (with git development team and/or mentor)
    sequencer features needed for gitk and implement them

Now I think (7) is even more needed/interesting, e.g. to be able to do interactive rebasing with gitk.

And (6) shouldn't be that hard, but it will be even more interesting when sequencer is libified. I have some ideas for that, but I think, this is not so important in near time. Perhaps, there are other parts that need some libification first. :)

Yes!

But due to my current lack of time I have ignored some topics on the list that I am interested in. :-(

This is fluctating.

Strange people. Lunatics. Geeks.

:-)

In the beginning: portable shell is a mess. ;) Well, I guess I have learned a lot more...

Regards,
 Stephan

Heya,

Replying to:

On Thu, Sep 4, 2008 at 23:15, Jakub Narebski <jnareb@gmail.com> wrote:

I consider it a successfull project, although it might not be what some hoped it would be, I am happy with it. I plan to submit it for inclusion in /contrib in the near future. I hope to work on improving move detection in git log when time allows.

I very much plan to continue development, now that my day-to-day life seems to have stabilized (after being on vacaction to the USA for a month). I enjoyed working with the git community, especially the 'real time' help on #git has been very usefull!

On Thu, Sep 04, 2008 at 11:15:56PM +0200, Jakub Narebski <jnareb@gmail.com> wrote:

builtin-merge is merged. I did not have it in my proposal, but support for custom merge strategies is also merged. Lately I was working on a better merge-recursive API, that is mostly in next now.

Time did not permit, but I was about to rewrite merge-octopus, merge-one-file and merge-resolve in C as well.

I hope I'll have time to do it later in my free time, but I can't promise anything, being a fulltime student as well. ;-)

Jakub Narebski <jnareb@gmail.com> wrote:

Marek's work is fully merged into egit.git. Most of it merged before the pencil's down date, but the final push (and bonus fetch) UI parts merged just after it.

In my opinion this turned out to be a very successful project, and Marek has become a really good contributor. I'm happy to have found him, and even happier I stole him away from the Eclipse Foundation (where he also had a GSoC application filed).

I suspect he won't be contributing 6k lines of code per month now that school semesters start again, but I think he's interested enough (and had enough fun doing it) to stick around. Actually its really good he can't contribute 6k/month, I don't have time to code-review 6k/month. :)

Regarding...

Jakub Narebski wrote:

Project is done, it's merged to master of egit, as Shawn pointed. I also consider it successful, for sake of my experience too!

I was hoping to do some optional task after push GUI (2nd part of GSoC), but instead I was designing/polishing the new GUI and existing clone wizard. Created code was also base for later added (after GSoC) fetch GUI.

The downside is that there was not enough time for even starting stuff like binary delta (re)implementation. This became huge task, as there are serious libxdiff related license-issues. So it probably needs to be implemented from scratch basing on existing research and pack file format knowledge, but not the code.

I enjoyed coding here, so I except to keep on working on egit. As you may see (looking on my activity on the list) I'm pretty busy last time - doing overdue exams, getting credits or having vacations (that's especially hard labor...). So I won't work on egit next weeks for sure. Later, I'm back to university again, so my activity may vary/be limited during some periods. But I except to take some missing feature and work on that in free time.

Folks that I don't know how spend that much time on Git producing portions of nice code...

Actually, I learned here what community in open source projects means. Even if Git is not considered user-friendly, its community definitely is.

Among other things, I learned here using Git ;), doing development in more open source way (e.g. where patches review/comments is natural process), and somewhat doing development of product that is intended to be highly reusable and effective.

(...)

Thanks too!

Hi,

On Thu, 4 Sep 2008, Jakub Narebski wrote:

My 2 cents: a full success. Details are public, so no need to repeat all that here, right? ;-)

Was also nice to read it on kernel trap (last sentence in the announcement of Git 1.6.0).

Ciao, Dscho

On Thu, 4 Sep 2008, Jakub Narebski wrote:

By the way, I have found via LWN that Perl has written nice summary of theirs Google Summer of Code projects[1]. Among other they very much praise that students did blogging about their progress:

  "use Perl | Summer of Code recap"
  http://use.perl.org/articles/08/08/29/1224242.shtml

  Most of the students did a great job of blogging their progress, which
  I think is an important part of Summer of Code for the rest of the
  community.

What do you think about it (for example about doing it in a future if our informal Git Development Community would participate in next Google Summer of Code programs)?

[1] I plan to add quick summary of final results to GSoC2008 wiki page, http://git.or.cz/gitwiki/SoC2008Projects, of course if GSoC manager(s), mentors, or students don't do it first...

Hi,

On Mon, 15 Sep 2008, Jakub Narebski wrote:

I think that it is better to require frequent interaction on the mailing list, or at least on IRC.

Blogging is nice for those who want to follow the progress of a project, but do not want to get involved. In this case, it is even better than having the people discuss the issues of the project openly, as reading a blog does not require constant monitoring like the list does.

However, blogging costs time.

I think that in the case of Git, those who are interested are those that will work on (as opposed to "with") those projects, too. So I think the time is better spent on the mailing list.

Especially given the fact that the interested parties can guide the course of the project early, i.e. before a blog entry would be written (just think about the Git.pm thing; both Lea's and Joshua's project could have benefitted from enhancing the existing Git.pm according to their needs).

My cent, Dscho

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

...

I agree with Dscho completely. We are an email based community. Students should be involved by email, not by blogs. If we were the Ruby community (which near as I can tell is blog based) then blogging might make more sense. We're not. We're an offshoot of the Linux community, which is very much email-centric.

Exactly. Email also costs time, but typically we don't see traffic on the list that is just broadcasting something pointless. In general people only start a topic if there is going to be worthwhile discussion attached to it, and often that results in a useful patch being applied at some point in the future. IOW that time for email is just necessary in the course of development, so it isn't wasted.

If the community of semi-interested people really wants blogs to read, RSS scrape a news feed like Gmane. Ohloh already has an RSS feed defined around the emails sent to git ML with the subject "[ANNOUNCE]".

I however prefer email, and dislike blogging, so I'll continue to use the git ML in the future, and so will any student I mentor in any potential future GSoC program. FWIW I blogged here at new day-job on company internal blog for about a half day before I gave up and said "damn, that really is stupid".