Re: CM 1.1 git question

[Graham Percival]

On Wed, Feb 18, 2009 at 03:25:59PM +0100, Johannes Schindelin wrote:
> 
> Sorry, I was _way_ too terse (but look at the time I wrote this; and I was 
> still working, not watching TV)...

:)

> What I really meant is: A 'git clone' will fetch too many branches.  So it 
> is not efficient.  But it is simple, so we should recommend 'git clone'
> even if that means fetching too much.

I disagree; they are both exactly the same level of simplicity:

1)  Copy and paste the following lines:
  git clone blah

1)  Copy and paste the following lines:
  mkdir lilypond
  cd lilypond
  git init
  git checkout blah
  git foobazzle bluh


By "simplicity" I mean "amount of mental effort the contributor
must expend".  We have the *exact* commands written for every main
branch.  See
http://www.kainhofer.com/~lilypond/Documentation/devel/contrib-guide/Getting-the-source-code.html


> In the alternative, you could try to illustrate how branches look like.  
...
> But maybe that is all too complicated; I've been a Git for too long, and 
> you are much closer to people scared by distributed SCMs...

Mao, *I'm* still scared of distributed SCMs.  :)


I really think that the current system is good.  If we can reduce
the number of lines in the copy&paste section (with git checkout
project/branch or whatever), then I'm all for it, but the current
system uses a few lines of black magick to produce separate
directories which only download the relevant history to that
directory ("branch").

Cheers,
- Graham