From: Alex Riesen <raa.lkml@gmail.com>
Date: Fri, 04 May 2007 17:24:50 GMT
Subject: Re: repack: handling of .keep files
Message-ID: <81b0412b0705041024i43d7fc5ah1967d6a6192dc6ee@mail.gmail.com>
In-Reply-To: <7vslacttij.fsf@assigned-by-dhcp.cox.net>

On 5/4/07, Junio C Hamano <junkio@cox.net> wrote:
> "Alex Riesen" <raa.lkml@gmail.com> writes:
>
> > Still, git-log shouldn't crash (nothing should, of course).
>
> Honestly, I think that's borderline.  If you "dd if=/dev/random
> of=/dev/hda", should the kernel keep going, perhaps gracefully
> declining access to the filesystem on that drive?

e2fsck has a test somewhere which randomly corrupts a partition
and then lets the program fix it.
All kind of corruptions happen, we will have to deal with them.
Especially if this crash is so simple to reproduce.

> case of temporary pack I do not think there would be a risk of
> filename collisions, I think it makes sense to use either
> GIT_DIR or GIT_OBJECT_DIRECTORY instead of the working tree.
>
> I do not know pros-and-cons between .git/ and .git/objects/;

These are settable separately, so theoretically you can end
up with .git and .git/objects being on different filesystems.
Atomic rename wont be possible than.