Re: [RFC/PATCH 1/2] remote tracking: return the tracking branch for the given branches

From: Santi Béjar <santi@agolina.net>
Date: Jun 18, 2009, 21:31 UTC
Message-ID: <adf1fd3d0906181431o9b50191n29f349a14f9d1be1@mail.gmail.com>
In-Reply-To: <7v63etnydr.fsf@alter.siamese.dyndns.org>

2009/6/18 Junio C Hamano <gitster@pobox.com>:

Show 11 quoted lines

> Santi Béjar <santi@agolina.net> writes:
>
>> One thing it just occurred to me is to return the explicit refspec
>> instead of the tracking branch. So with the default config (after a
>> clone):
>>
>> $ git remote tracking origin refs/heads/master
>> refs/heads/master:refs/remotes/origin/master
>
> I think the output is sensible, instead of saying something like
> 'origin/master'.

OK.

>
> We can and should allow the end users to use abbreviated form as input to
> us (either command line or configuration value) when it is unambiguous,

OK. Moreover I think it also applies to the %(upstream) case I reported some days ago.

Show 21 quoted lines

> but when returning values for use by Porcelains, we should be strict and
> precise to avoid ambiguities.
>
>> this makes a difference in case we want to allow returning all the
>> matching tracking branch and not the first one with more than one
>> branch, as:
>>
>> $ git config remote.origin.fetch --add +refs/heads/*:refs/remote/another/*
>>
>> $ git remote tracking origin refs/heads/master
>> refs/heads/master:refs/remotes/origin/master
>> refs/heads/next:refs/remotes/origin/next
>
> I am not sure what this example is doing.  You asked about refs/heads/master
> but you also talk about refs/heads/next?   I guess you meant to say
>
>    $ git remote tracking origin refs/heads/master
>    refs/heads/master:refs/remotes/origin/master
>    refs/heads/master:refs/remotes/another/master
>
> instead, judging from the next example?

Yes.

Santi