Re: [PATCH v3 00/32] object-store: lookup_commit

From: Junio C Hamano <gitster@pobox.com>
Date: Jun 29, 2018, 17:57 UTC
Message-ID: <xmqq8t6xv5kb.fsf@gitster-ct.c.googlers.com>
In-Reply-To: <20180629012222.167426-1-sbeller@google.com>

Stefan Beller <sbeller@google.com> writes:

Show 11 quoted lines

> This continues the elimination of global variables in the object store and
> teaches lookup_commit[_reference] and alike to handle a_repository.
>
> This is also available as
> https://github.com/stefanbeller/git/tree/object-store-lookup-commit
> or applies on top of 02f70d63027 (Merge branch 'sb/object-store-grafts'
> into next, 2018-06-28).
>
> Picking a base for this one is really hard, as nearly all series currently
> cooking or in flight collide with it on one or two lines. (lookup_* is used
> heavily, who would have thought?)

One technique these (not just this) recent efforts seem to be forgetting is to introduce "new" names that take a_repo and then make the existing one a thin wrapper that calls the new one with &the_repo as the argument.

IOW, a new call to lookup_commit_reference(oid) in a function that is happy to access the default repo (and notice taht 99% of our code is happy with using just the currrent repository the user started the git process for) does not need to be updated to call the function with the same name with a different function signature lookup_commit_reference(the_repo, oid). If we name the new one lookup_commit_reference_in_repo(r, oid) and make the callers that already take a repo pointer call it, while leaving the callers that do not even know or care about multiple in-core repo instances alone to keep calling lookup_commit_reference(oid), you do not have to worry too much about colliding. That way, you do not have to step on each others' toes with avoidable semantic merge conflicts and incrementally improve things over time, no?

Thanks.