Re: Hacking on PostgreSQL via GIT

Aidan Van Dyk wrote:
> * Florian G. Pflug <fgp(at)phlo(dot)org> [070430 08:58]:
>
>> It seems as if git pulls all revisions of all files during the pull -
>> which it shouldn't do as far as I understand things - it should only
>> pull those objects referenced by some head, no?
>
> Git pulls full history to a common ancestor on the clone/pull. So the
> first pull on a repo *will* necessarily pull in the full object history.
> So unless you have a recent common ancestor, it will pull lots. Note
> that because git uses crypto hashes to identify objects, my conversion
> and Martin's probably do not have a recent common ancestor (because my
> header munging probably doesn't match Martin's exactly).
Ah, OK - that explains things.

>> The interesting thing is that exactly the same problem occurs with
>> both if your mirrors...
>>
>> Any ideas? Or is this just how things are supposed to work?
>
> Until you have a local repository of it, you'll need to go through the
> full pull/clone. If you're really not interested in history you can
> "truncate" history with the --depth option to git clone. That will give
> you a "shallow repository", which you can use, develop, branch, etc in,
> but won't give you all the history locally.
I'll retry with the "--depth" option - I'm doing development on my powerbook,
and OSX seems to cope badly with lots of little files - the initial unpacking
took hours - literally..

> Also - what version of GIT are you using? I *really* recommend using at
> least 1.5 (1.5.2.X is current stable). Please, do your self a favour,
> and don't use 1.4.4.
I'm using 1.5.0 currently - it was the latest stable release when I began
to experiment with git.

greetings, Florian Pflug