[GSoC update] git-remote-svn: Week 1

[GSoC update] git-remote-svn: Week 1

[Ramkumar Ramachandra]

Hi,

I'm sorry that this comes a couple of days late owing to the fact that
I flew down the Copenhagen just a couple of days ago, and am coping
with the jetlag. I've just started working, and I'm currently
investigating David's SVN exporter. Hopefully, after I've moved into
an apartment, I'll have something concrete to show next week.

-- Ram

Re: [GSoC update] git-remote-svn: Week 1

[David Michael Barr]

Hi Ram,

A quick update. I've begun work on a reference exporter written in perl.
I believe I introduced some regressions during code cleanup.
From memory, the svn-dump-fast-export_0.8 tag has been verified for
10,000 revisions of my repository.
My perl implementation works just fine until it reaches around 11,000
revisions at which point there's a memory explosion.

-- David

On 05/05/2010, at 10:59 PM, Ramkumar Ramachandra wrote:

> Hi,
> 
> I'm sorry that this comes a couple of days late owing to the fact that
> I flew down the Copenhagen just a couple of days ago, and am coping
> with the jetlag. I've just started working, and I'm currently
> investigating David's SVN exporter. Hopefully, after I've moved into
> an apartment, I'll have something concrete to show next week.
> 
> -- Ram

Re: [GSoC update] git-remote-svn: Week 1

[Ramkumar Ramachandra]

Hi David,

> A quick update. I've begun work on a reference exporter written in perl.

Curiously, why?

> I believe I introduced some regressions during code cleanup.
> From memory, the svn-dump-fast-export_0.8 tag has been verified for
> 10,000 revisions of my repository.

Thanks for the heads up. Don't worry about it though- I've started
cleaning up the code. Will send you some patches off the list by
email, or whatever you prefer.

-- Ram

Re: [GSoC update] git-remote-svn: Week 1

[David Michael Barr]

Hi Ram,

My off-list reply was because I was still waking up.

>> A quick update. I've begun work on a reference exporter written in perl.
> 
> Curiously, why?

I was working on a slow but reliable alternate implementation for
validation and regression testing. So far, it seems really hard to write
anything reliable atop Subversion.

>> I believe I introduced some regressions during code cleanup.
>> From memory, the svn-dump-fast-export_0.8 tag has been verified for
>> 10,000 revisions of my repository.
> 
> Thanks for the heads up. Don't worry about it though- I've started
> cleaning up the code. Will send you some patches off the list by
> email, or whatever you prefer.

I know its bad form, but I've done what I promise is the last history
manipulation on my public repository. Now that the code is at a point
that is quite functional and ready to receive contributions, I've signed
off on my own commits and reviewed the history.
Its not yet '1.0' but hopefully Ram's eyeballing of the code will get us
there.

--
David Barr

Re: [GSoC update] git-remote-svn: Week 1

[Ramkumar Ramachandra]

Hi David,

On Mon, May 10, 2010 at 9:08 AM, David Michael Barr
<david.barr@cordelta.com> wrote:
> I was working on a slow but reliable alternate implementation for
> validation and regression testing. So far, it seems really hard to write
> anything reliable atop Subversion.

Okay. I won't really touch that because I'm not very fluent in Perl.

> I know its bad form, but I've done what I promise is the last history
> manipulation on my public repository. Now that the code is at a point
> that is quite functional and ready to receive contributions, I've signed
> off on my own commits and reviewed the history.

On a related note, we might also want to benchmark it against
alternatives and make the benchmarks public for the 1.0 release.

> On that note, do you think I should rewrite my history to include
> 'signed-off' or any other cleanup before I start accepting patches?

Yeah, I don't think we should have a problem with non-FF updates,
atleast until the project reaches 1.0 and becomes more public.

-- Ram