Re: [PATCH v1] Replace {u}int_fast{16|32} with {u}int32_t

[Noah Goldstein <goldstein.w.n@gmail.com>]

On Mon, Apr 11, 2022 at 1:07 PM Paul Eggert <eggert@cs.ucla.edu> wrote:
>
> On 4/11/22 09:58, Noah Goldstein via Libc-alpha wrote:
> > -   uint_fast16_t stack_flags = DEFAULT_STACK_PERMS;
> > +   uint32_t stack_flags = DEFAULT_STACK_PERMS;
>
> Doesn't this lose width info - or worse, add misleading info? For this
> particular case it looks like we need 32 bits, when we don't.
>
> If we want to lose info, it'd be simpler to use plain 'int' and
> 'unsigned int' for situations like these. That's less misleading.

Assume you only meant replacing the {u}int_fast16_t with unsigned int
if so fixed in v2.
>
> If we don't want to lose info, it'd be better to define private types
> like glibc_uint_fast16_t and use these private types to address
> efficiency concerns.
>
> I prefer losing the width info here, as this will cause less
> bikeshedding in the future. But if we want to keep width info, surely we
> should not make it more misleading.