On Tue, Aug 28, 2007 at 01:06:56PM CEST, Johannes Schindelin wrote:
> Hmm. I think you'll have to hack pack-objects, or you'll have to create
> dummy refs from the dangling objects/blobs.
>
> The latter is probably easier, since it just involves a "git fsck
> --lost-found" and putting the found things into a
> "refs/I-dont-wanna-lose-you/" refspace. But it will be certainly more
> expensive.
>
> The upside of this method would be that you have an integrity test for
> free.
That is possible as well, but I like the method I've proposed better
since if there are any objects that _really_ aren't referenced anymore,
they will get pruned.
> Putting all objects into a single database could easily break the current
> packfile size limit, and it will be suboptimal for cloning: chances are
> that objects delta well against objects which are not in the same
> (forked)project, and therefore the server is more likely to get
> non-reusable deltas.
The overhead for fetching over HTTP might be insane.
--
Petr "Pasky" Baudis
Early to rise and early to bed makes a male healthy and wealthy and dead.
-- James Thurber
-
To unsubscribe from this list: send the line "unsubscribe git" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
