getting source with git

[Graham Percival]

Ok, it's time for our yearly "how should people get started with
git" debate.  There's three contenders: git clone, git clone
--depth=1, and the complicated "getting everything on the main
branch" thing that's in the CG currently.

After the init, I see the following sizes of the .git dir:
1) git clone: 101M
2) git clone -depth=1: 27M
3) main branch only: 67M

This isn't a perfect measure of the size/speed of download, so any
improvements to these figures are welcome.

My initial preference is for #2, since I've used it and had no
difficulty making pushes.  However, various sites/email archives
on the web warn that such pushes are not reliable; it depends on
the parents and other commits and stuff.  I guess I just lucked
out in being able to push the stuff I did.  Since git is already
so much of a pain, I'd rather not rely on luck for new
contributors.
OTOH, new contributors won't have push ability, and I didn't find
any warnings about not being able to git format-patch.

A few people talked about browsing the history, which surprised
me.  Whenever I want to look at history, I use the web git
interface.  But evidently other people don't share my pathological
hatred of git, so I guess that "availability of history" is a
non-insignificant factor.


Taking the above into consideration, I suggest that we don't
change anything: CG 1 will only discuss method #3 (git remote
add).  67 megs is still much larger than I'd _like_ to ask
potential typo fixers to download, but this seems unavoidable.  :(

On the plus side, with the integration of web, we only have two
branches that normal contributors will need to track: master and
translations.

Any complaints?

Cheers,
- Graham