Re: [PATCH v4 01/10] doc: define unambiguous type mappings across C and Rust

From: Junio C Hamano <gitster@pobox.com>
Date: Nov 15, 2025, 20:03 UTC
Message-ID: <xmqqo6p3dpw4.fsf@gitster.g>
In-Reply-To: <CALnO6CA-6waRpkqzLxR+f2yzwfhmf_jvbtEZC7FAFN9NLkqkXg@mail.gmail.com>

"D. Ben Knoble" <ben.knoble@gmail.com> writes:

Show 9 quoted lines

>> Me neither, but I suspect it may mostly use of non-word "signless"
>> that is the issue.
> ...
> Agreed, and I suspect this is roughly the implementation.
>
> My point was that Ezekiel seemed to justify (?) the use of "signless"
> by pointing to those warnings (I personally am on the fence for how to
> treat the combination of facts, but it seems useful to consider that
> char is not easily comparable with integers of various signedness).

Agreed. I think your point matches my suspicion that the use of the non-word "signless" was what Ramsay reacted. The 'char' with the implementation defined signedness is making -Wsign-compare even more quirky than it already is.