Re: [PATCH 0/4] git --paginate: do not commit pager choice too early

On Sat, Jun 26, 2010 at 02:22:03PM -0500, Jonathan Nieder wrote:

Here’s a fix from the famous setup cleanup series[1].

The problem it fixes is somewhat obscure: ‘git -p foo’ is not paying
attention to the repository-local ‘[core] pager’ configuration when
run from a subdirectory.  But it is a real bug, and the fix gives
an example of how to deal with repository setup and should be safe.

Patch is against master.  There is a small semantic conflict with
jn/grep-open: SIMPLEPAGER should be changed to SIMPLEPAGERTTY in the
prerequisites for the test_default_pager function.  Please let me
know if I should push a merge commit to help resolve that.

I just read through the patches, and it seems like a reasonable fix. So
in that sense, ACK.

But reading the message for patch 4/4, I can't help but wonder if the
right way forward is for the git wrapper to _always_ find the repository
as the very first thing. I.e., make finding the repository a
non-destructive thing (especially don't end in a different directory),
and just do it before we run any other code.

For a few commands like "init" or "clone", we would end up just
ignoring this information, but the worst case should be a little bit of
wasted effort.

And then all these silly setup and timing issues could just go away, as
we would know that GIT_DIR and GIT_WORK_TREE would be set properly.

-Peff