Google Summer of Code 2007

Google Summer of Code 2007

[Shawn O. Pearce]

ShadeHawk and robinr brought up Google's Summer of Code on #git the
other day.  I had also been thinking about seeing if we cannot get
Git involved with SoC, so here goes... ;-)

The application deadline for organizations is March 12th.
The earliest that we can submit an application is March 5th, so we
still have time to kick ideas around and see if the community is
interested in participating in SoC.

Google's FAQ has a lot of details, but the important part which
lists what should be included in an application can be found here:

  http://code.google.com/support/bin/answer.py?answer=60303&topic=10727

I don't know how the great SoC filter works for organizations,
but last year's list (found at http://code.google.com/soc/) has a
number of projects listed on it that are actively using Git for their
version control.  It would be nice if the SoC program was able to
benefit multiple projects in one shot, by helping to improve Git. :)

I'd be happy to help put the application together and to probably
also serve as a mentor, but I'm only willing to put in the effort
if others are genuinely interested and will also help out.

Thoughts?

-- 
Shawn.

Re: Google Summer of Code 2007

[Johannes Schindelin]

Hi,

On Sun, 25 Feb 2007, Shawn O. Pearce wrote:

> Thoughts?

The projects I would suggest are

1. Make more complete tests.
2. Use sparse to fix errors revealed by static code analysis.
3. Make more things builtin.

Regarding 2: this would be more of a sparse project IMHO. It could mean 
that sparse actually learns about fixing trivial bugs. (You need to 
inspect the result, but verifying is faster than writing.)

Ciao,
Dscho

Re: Google Summer of Code 2007

[Andy Parkins]

On Sunday 2007 February 25 07:59, Shawn O. Pearce wrote:

> Thoughts?

What about the oft-asked for windows port?  From what I've seen, none of the 
Unix types have any incentive to do this, so a financial incentive might be 
just the thing.

Perhaps the windows port is already far enough along that this wouldn't be a 
big job.  In which case the project could be windows shell integration (or 
whatever other candy-based device a windows user would need to easily use 
git).

Project title: "Make it so that one .exe click-and-install would make the full 
power of git available in an idiot^H^H^H^H^H typical-GUI-using-developer 
friendly form"

Given that the best thing for git in the long term is more users, and that no 
windows support is the commonly given reason why other SCMs are chosen 
instead, this would be a good use of the resources.

I'm off to shower, I feel dirty even talking about Windows :-)


Andy
-- 
Dr Andy Parkins, M Eng (hons), MIET
andyparkins@gmail.com

Re: Google Summer of Code 2007

[Matthieu Moy]

Andy Parkins <andyparkins@gmail.com> writes:

> What about the oft-asked for windows port?  From what I've seen, none of the 
> Unix types have any incentive to do this, so a financial incentive might be 
> just the thing.

+ 1. It seems the windows port somewhat has the status "we want it but
I'm not going to do it" for most git contributors.

I'm personnally using unix 99% of the time, but I also have a piece of
windows installed on my laptop. For example, I'm working on a small
project using windows-only software, and being able to use git
comfortably from this setup would be cool.

Also, I'm relucant to migrate to git completely even for unix-only
projects, since I know that a hypothetical future windows-port will be
made hard by the non-availability of the windows port.

> Given that the best thing for git in the long term is more users, 

That's what I feel. The problem with distributed SCMs at the moment is
that there are too many of them, not sharing enough users. For
example, I had to choose a replacement for CVS with some colleagues
recently. I argued about using a distributed one, but the argument
against, and the decision was "SVN is what they're more likely to use
later, let's go for SVN no matter how bad it is".

> and that no windows support is the commonly given reason why other
> SCMs are chosen instead, this would be a good use of the resources.

Sure. One of the best examples being Mozilla. I hardly see them using
a piece of software that's not working properly on windows, since the
majority of users (and probably of the developers, I don't know) are
running windows.

-- 
Matthieu

Re: Google Summer of Code 2007

[Johannes Schindelin]

Hi,

On Wed, 28 Feb 2007, Matthieu Moy wrote:

> It seems the windows port somewhat has the status "we want it but I'm 
> not going to do it" for most git contributors.

That is not really fair, is it? _Most_ git contributors have helped 
Windows users in one way or another (why don't we have a symlink()ed HEAD 
anymore, hmm?).

However, I do have the impression that the percentage of people in 
complain-only mode is way lower in the non-Windows camp.

Ciao,
Dscho

RE: Google Summer of Code 2007

[Raimund Bauer]

On Sun, 25 Feb 2007, Shawn O. Pearce wrote:

> Thoughts?

What about sub-project support?
Seems also to be on the list of things that would be nice to have, but not
needed enough by the people doing the heavy work here.

-- 
best regards

  Ray

Re: Google Summer of Code 2007

[Brian Gernhardt]

On Feb 28, 2007, at 4:35 AM, Raimund Bauer wrote:

> On Sun, 25 Feb 2007, Shawn O. Pearce wrote:
>
>> Thoughts?
>
> What about sub-project support?
> Seems also to be on the list of things that would be nice to have,  
> but not
> needed enough by the people doing the heavy work here.

+1.  Something I'd definitely work on myself...  But hacking git  
isn't what I'm getting paid for.  :-(  So let's see if someone else  
will do it.

~~ Brian

Re: Google Summer of Code 2007

[Shawn O. Pearce]

Brian Gernhardt <benji@silverinsanity.com> wrote:
> On Feb 28, 2007, at 4:35 AM, Raimund Bauer wrote:
> >What about sub-project support?
> >Seems also to be on the list of things that would be nice to have,  
> >but not
> >needed enough by the people doing the heavy work here.
> 
> +1.  Something I'd definitely work on myself...  But hacking git  
> isn't what I'm getting paid for.  :-(  So let's see if someone else  
> will do it.

Most (if not all) Git hackers are unpaid.

Err, I think we are all paid by someone, but that pay is to produce
things totally unrelated to Git.  Because if we weren't paid,
we probably couldn't afford fancy luxuries like food.  :)

Me, I hack on Git because I find it fun (it is a nice change of pace
from the dreary hacking I do for food) and I like using the software
to help me through my dreary hacking work.  But I'm certainly not
getting any money for my Git hacking.


Re: SoC, we should setup a page in the Git Wiki listing these great
ideas out - we are already up to a few, and subprojects has gotten
at least a couple of votes now as something the community would
like to see.

-- 
Shawn.

Re: Google Summer of Code 2007

[Brian Gernhardt]

On Feb 28, 2007, at 10:47 AM, Shawn O. Pearce wrote:

> Most (if not all) Git hackers are unpaid.
>
> Err, I think we are all paid by someone, but that pay is to produce
> things totally unrelated to Git.  Because if we weren't paid,
> we probably couldn't afford fancy luxuries like food.  :)

I mostly meant that as a "I don't have the time to work on a large  
project like that because I have to do other things for work".  And I  
actually have non-computer related things to do when I'm not  
working.  Shock and horror.  ;-)

> Me, I hack on Git because I find it fun (it is a nice change of pace
> from the dreary hacking I do for food) and I like using the software
> to help me through my dreary hacking work.  But I'm certainly not
> getting any money for my Git hacking.

I hack on Git because I use it for work (yay being a contractor and  
getting to pick my own tools!) and periodically need to fix a problem  
here and there.  If I had more time, I'd probably do more.  But as is  
small things I submit patches, large things I go "Please fix this".  :-)

~~ Brian

Re: Google Summer of Code 2007

[Jakub Narebski]

Shawn O. Pearce wrote:

> ShadeHawk and robinr brought up Google's Summer of Code on #git the
> other day.  I had also been thinking about seeing if we cannot get
> Git involved with SoC, so here goes... ;-)
> 
> The application deadline for organizations is March 12th.
> The earliest that we can submit an application is March 5th, so we
> still have time to kick ideas around and see if the community is
> interested in participating in SoC.

Well, certainly there is a number of areas to work on. This includes:
 * much talked on (and even with two different independent
   implementations) _subproject (submodule) support_, which would
   certainly help using git for large modular projects like KDE,
   Mozilla or distributions.
 * lightweight checkout aka. .gitlink idea, to have file which
   would point to object directory, refs directory, index file
   and current branch file. It could help submodule support.
 * partial/sparse checkouts, where you can checkout for example
   only Documentation directory, work on it, but commit full tree.
   Sometimes it better suits than using submodules
 * gitweb caching and other gitweb improvements: bringing together
   all gitweb implementations. Perhaps gitweb maintainer could
   come of it. Or at least gitweb admin for kernel.org
 * builtinification and libification
 * lazy clone aka remote alternates, if it can be done at all...
 
> Google's FAQ has a lot of details, but the important part which
> lists what should be included in an application can be found here:
> 
>   http://code.google.com/support/bin/answer.py?answer=60303&topic=10727
> 
> I don't know how the great SoC filter works for organizations,
> but last year's list (found at http://code.google.com/soc/) has a
> number of projects listed on it that are actively using Git for their
> version control.  It would be nice if the SoC program was able to
> benefit multiple projects in one shot, by helping to improve Git. :)

The other SCM which participated in SOC2006, Subversion and Monotone,
both are backed by organizations. Mercurial didn't participate, but has
a page with ideas for SOC2006.

If Eclipse is to participate, perhaps one of Eclipse projects could be
git plugin for Eclipse, or Java implementation of Git (perhaps with some
Java improvements :-), or perhaps some generic distributed SCM plugin
framework.

-- 
Jakub Narebski
Warsaw, Poland
ShadeHawk on #git

Re: Google Summer of Code 2007

[Matthieu Moy]

Jakub Narebski <jnareb@gmail.com> writes:

> The other SCM which participated in SOC2006, Subversion and Monotone,
> both are backed by organizations. Mercurial didn't participate, but has
> a page with ideas for SOC2006.

There was a SoC project too for bzr, to create a GUI (called olive).
Also backud by organization (Canonical).

-- 
Matthieu

Re: Google Summer of Code 2007

[Shawn O. Pearce]

Jakub Narebski <jnareb@gmail.com> wrote:
> Shawn O. Pearce wrote:
> > ShadeHawk and robinr brought up Google's Summer of Code on #git the
> > other day.  I had also been thinking about seeing if we cannot get
> > Git involved with SoC, so here goes... ;-)
> > 
> > The application deadline for organizations is March 12th.
> > The earliest that we can submit an application is March 5th, so we
> > still have time to kick ideas around and see if the community is
> > interested in participating in SoC.

I've put together the list of ideas posted to the mailing list on
the Git Wiki:

  http://git.or.cz/gitwiki/SoC2007Ideas

Links to each message on gmane are also included.  The ideas need
to be extended out some, as many of them might not be recognized
or understood by someone from outside of the Git community.

-- 
Shawn.

Re: Google Summer of Code 2007

[Martin Langhoff]

On 3/2/07, Shawn O. Pearce <spearce@spearce.org> wrote:
> I've put together the list of ideas posted to the mailing list on
> the Git Wiki:
>
>   http://git.or.cz/gitwiki/SoC2007Ideas

There, I've added a couple I think I can help mentoring ;-)

cheers,


martin