Re: [PATCH] mktree: do not check type of remote objects

From: Junio C Hamano <gitster@pobox.com>
Date: Jun 21, 2022, 16:51 UTC
Message-ID: <xmqqwndagdu3.fsf@gitster.g>
In-Reply-To: <748f39a9-65aa-2110-cf92-7ddf81b5f507@roku.com>

Richard Oliver <roliver@roku.com> writes:

Show 6 quoted lines

> As a side note, do you think we need to re-work some uses of the word
> 'missing' in the documentation? Some uses of the word, such as in
> mktree, predate the concept of promisor remotes. The partial-clone.txt
> documentation differentiates between missing "due to a partial clone
> or fetch" and missing "due to repository corruption".  Would making
> such a distinction elsewhere be useful?

I do agree with that direction. Back when the world was simpler, there were just "we have the object" and "we ought to have the object but we do not see it", the latter of which was a clear repository corruption. There was no third choice.

With the lazy/partial clone stuff, it is not a corrupted repository anymore when certain objects that are reachable during traversal is missing (this is slightly different from a shallow clone in that a shallow history makes the traversal stop early, so at least objects in a shallow clone that are reachable during traversal must exist).

So, yes, I think the distinction between "missing but shouldn't be" and "missing by design" is a good thing to keep in mind.

Thanks.