Re: [RFC][PATCH] Allow transfer of any valid sha1

From: Eric W. Biederman <ebiederm@xmission.com>
Date: Jun 8, 2006, 09:33 UTC
Message-ID: <m164jc9ekx.fsf@ebiederm.dsl.xmission.com>
In-Reply-To: <7virntsto6.fsf@assigned-by-dhcp.cox.net>

A quick status update.

I think I have clean working version of the sha1 transfer code, I left on vacation before I could send it out so I need to dig it out and make certain everything still applies.

I finally figured out what my problem pulling Andrew's changes were. git-quiltimport remembers what the previous commit was and when I added merging I forgot to update that the variable that stores the previous commit. So since I had the history wrong git-merge was finding the wrong common ancestor, which is an easy way to mess up an automatic merge :)

Show 6 quoted lines

> The last time I talked with Andrew, he is not doing a merge nor
> resolving merge conflicts.  He treats git primarily as a
> patchbomb distribution mechanism, and works on (a rough
> equivalent of) the output of format-patch from merge base
> between his base tree and individual subsystem tree.  After that
> things are normal quilt workflow outside git, whatever it is.

Andrews git import does appear to be a git-pull from an appropriate tree and then a diff of the automatic merge result, so while there doesn't appear to be manual merging there is a little bit of automatic merging going on.

Anyway when I wake up in the morning I should see if I have successfully imported Andres 2.6.17-rc5-mm3 tree. All of that pulling of git trees on demand noticeably slows down the import on my dinky test machine. I'm not certain how much of that a machine that had plenty of memory would see though.

Eric