From: Junio C Hamano <gitster@pobox.com>
Date: Thu, 22 May 2025 18:44:43 GMT
Subject: Re: [PATCH v2 2/2] midx: stop repeatedly looking up nonexistent packfiles
Message-ID: <xmqqy0uo5vtw.fsf@gitster.g>
In-Reply-To: <20250522165924.GA1613@coredump.intra.peff.net>

Jeff King <peff@peff.net> writes:

> But back in the caller...
>
>> 	p = m->packs[pack_int_id - m->num_packs_in_base];
>> ...
>
> Our pack_int_id is the global one, so it needs to be adjusted. But this
> pack pointer we access is the same one that was filled (or not) by
> prepare_midx_pack(). So it cannot be NULL or the magic "fails" value,
> because prepare_midx_pack() returned 0.
>
> So I think this code is fine.

Ahh, I missed that call to midx_for_pack() in prepare_midx_pack()
that modifies pack_int_id variable.  Of course, it did not help that
the implementation detail of adjusting by m->num_packs_in_base which
is done in midx_for_pack() is not abstracted out and the caller needs
to do the same as above.  That confused me.

Thanks for an explanation.