Re: [RFC PATCH] git push: Push nothing if no refspecs are given or configured

[Finn Arne Gangstad <finnag@pvv.org>]

On Fri, Mar 06, 2009 at 01:07:56PM +0100, Johannes Schindelin wrote:
> [...]
> 
> Speaking of which, Steffen (who cannot reply right now, since he is AFK 
> for a while) had a patch to install "remote.<branch>.push = HEAD" with 
> clone and remote.  Would that be better?

Are you referring to the patch he suggested in October 2007? I'm
reading the October/November archives now, and it seems these things
have been discussed before... Ok so re-reading the discussion from
October/November 2007, it seems that we are having exactly the same
discussion again! I did not investigate far enough back in time before
firing up this thread again evidently.

Reading the threads almost makes me sad. Steffen had a lot of great
patches and ideas (implementing --current, adding tests, ...), but it
was somehow dropped. After having read through this it would feel
wrong somehow to even post patches on this topic, since I would
effectively just redo what he already did.

The only difference between his approach and what I am suggesting
seems to be that he mostly wanted "git push" to do various useful
tasks depending on configuration (the remote.<branch>.push stuff),
while I mostly want "git push" to do nothing by default. We both agree
that the current "git push" default behavior is very bad for our
workflow (many developers against one or more shared repos).

- Finn Arne