Re: question about git and remote repositories

[Michele Ballabio <barra_cuda@katamail.com>]

On Tuesday 08 April 2008, Ittay Dror wrote:
> If I want to start working on a project that uses git (and I want to use 
> git also), I first need to clone the project's repo. That is very slow 
> (using the git:// protocol). Are there alternatives? For example, as far 
> as I understand, if all I want is to get the latest commit in HEAD, 
> branch from it and start working, then all git really needs to have is 
> all objects referenced from that commit (commit, trees, blobs) and 
> that's it, right? (as long as I don't expect to see full log of past 
> operations)

In git terms, that is called a "shallow clone". For example:

git clone --depth 5 git://site.com/repo.git

would clone all the heads of repo.git along with their 5 most
recent commits.

You can make history deeper with

git-fetch --depth 20 git://site.com/repo.git branch

See the glossary, technical/shallow.txt, git-clone and git-fetch
man pages.