Re: [PATCH 13/16] prune: keep objects reachable from recent objects

[Jeff King <peff@peff.net>]

On Fri, Oct 03, 2014 at 02:47:57PM -0700, Junio C Hamano wrote:

> With this patch applied, the system will not prune unreachable old
> objects that are reachable from a recent object (the recent object
> itself may or may not be reachable but that does not make any
> difference).  And that is sufficient to ensure the integrity of the
> repository even if you allow new objects to be created reusing any
> of these unreachable objects that are left behind by prune, because
> the reachability check done during prune (with this patch applied)
> makes sure any object left in the repository can safely be used as a
> starting point of connectivity traversal.

Your use of "safely" in the last sentence here made me think.

In a repository that has had this patch from the beginning, it should be
safe to traverse the unreachable but unpruned objects, because the
property of the repository we are trying to guarantee means that we will
have all of the referents.

But in a repository created by current git (or one where you have been
fooling around with hash-object), we might not. Our secondary traversal
for recent objects may fail because we are already missing some of the
referents, and prepare_revision_walk (or traverse_commit_list) will
barf. That's a bad thing.

I think we need to tell that second traversal that it's OK to skip
missing objects that are reachable from recent objects. It's not ideal,
but there's nothing better we can do.  Eventually those objects will
fall off the back of the expiration window, but we should not be holding
up prunes (which, after all, are the thing that expires them :) ) in the
meantime.

I think it would be enough to just set revs->ignore_missing_links for
the secondary traversal.

-Peff