Re: Additional remote on a local clone. Where do the objects go?

From: Tay Ray Chuan <rctay89@gmail.com>
Date: Sep 4, 2010, 14:15 UTC
Message-ID: <AANLkTik=qeoq67KZa8Bj7Ynrt8O8ija6yeH6_+e4=gOZ@mail.gmail.com>
In-Reply-To: <i5len7$dpr$1@dough.gmane.org>

Hi,

(Please avoid dropping emails from the Cc list. Since you're using gmail, you could use "Reply to all".)

On Wed, Sep 1, 2010 at 7:51 PM, Stephen Kelly <steveire@gmail.com> wrote:

> Tay Ray Chuan wrote:
>> Two improvements: shared objects, and configuring git-pull:
>
> I'm confused. I thought the local clone already used shared objects?

You'd have to use "-s/--shared" with git clone; I don't think it's done by default.

(On hindsight, I should have used this instead of setting it up manually by writing to .git/objects/info/alternates.)

> So this will pull from the gitorious remote and store the objects in the
> parent git from where they are shared with this local clone?

No, you'd have to first pull from gitorious in your parent directory, before going into the local repo/branch (eg. qt46) and then pulling again.

Show 5 quoted lines

> Which is the better solution between this and git-new-workdir given that
> these are permanent branches, not temporary, and I'll want to push and pull
> various clones and share the objects. It looks like the git-new-workdir
> option creates symlinks, whereas the local clone creates hard links, and
> this solution creates a redirect of sorts.

I haven't used git-new-workdir comprehensively, so I'll avoid commenting too much, but I'd like to know if you've looked at branches? You can just switch between each and do your qt 4.6- and 4.7-specific work.

I say that, because what you're doing sounds like what a hg user would do to emulate git branches.

-- 
Cheers,
Ray Chuan