Re: [GSoC idea] Resumable clone

From: Duy Nguyen <pclouds@gmail.com>
Date: Mar 3, 2015, 01:02 UTC
Message-ID: <CACsJy8Cue+pxPbddkv1_yMD-SLFO4PjKgdCVX0V7YCn_kUArcQ@mail.gmail.com>
In-Reply-To: <xmqqoaob7yi8.fsf@gitster.dls.corp.google.com>

On Tue, Mar 3, 2015 at 6:47 AM, Junio C Hamano <gitster@pobox.com> wrote:

Show 10 quoted lines

> Koosha Khajehmoogahi <koosha.khajeh@gmail.com> writes:
>
>> Among GSoC 2011 ideas of git [1], it was proposed that a GSoC project
>> could be implementing resumable clone for git. AFAIK, this feature is
>> still missing in git but could be a great idea to be implemented. Does
>> that sound OK to the community?
>>
>> [1]: https://git.wiki.kernel.org/index.php/SoC2011Ideas#Resumable_clone
>
> Depends on how it will be implemented, I guess.

I still want to pursue stable pack approach [1]. It's not a good fit for clone case (cached bundles would be much better). On the other hand it does not need special support like cached bundles and can work with upload-pack-v1. It might fit in gsoc scope. I guess whether it may have a chance to get merged after gsoc is the deciding factor.

> I highly suspect that it would not fly well if it is added as a part
> of the current protocol, as people are actively discussing how to
> revamp it.  And I have a feeling that the v2 protocol would not
> ready for a student to build upon by this summer.

[1] http://thread.gmane.org/gmane.comp.version-control.git/184490/focus=184705 Though what I had in mind (and patches) does not need on-disk caching at all.

-- 
Duy