Re: [PATCH 1/7] chunk-format: introduce `pair_chunk_expect()` helper

From: Linus Arver <linusa@google.com>
Date: Jan 15, 2024, 22:53 UTC
Message-ID: <owly4jfei4xk.fsf@fine.c.googlers.com>
In-Reply-To: <owly8r4qi5zj.fsf@fine.c.googlers.com>

Linus Arver <linusa@google.com> writes:

Show 7 quoted lines

> So in summary there appear to be the following possibilities:
>
> CHUNK_MISSING
> CHUNK_TOO_SMALL
> CHUNK_OK
> CHUNK_TOO_BIG_ALIGNED
> CHUNK_TOO_BIG_MISALIGNED

On second thought, it appears that CHUNK_TOO_SMALL has three cases:

(1) chunk_size < record_size
(2) chunk_size >= record_size, but chunk_size < record_size * record_nr
    and decreasing record_nr can make chunk_size "fit"
(3) chunk_size >= record_size, but chunk_size < record_size * record_nr
    and decreasing record_nr cannot make chunk_size "fit"

where (2) and (3) are just like the *_(MIS)ALIGNED cases above when chunk_size is too big.

My default position is that these additional cases might need to be treated differently, although maybe it's overkill also.

Thanks.