On Thursday, 20 March 2025 07:50:24 Pacific Daylight Time Hans Åberg via Std-
Proposals wrote:
> The second part is the ability to implement it on your own. This part is
> tricky, and it is unclear if it can be admitted.

I don't think it can. That's what I am arguing is fragile.

> > But if it's not implemented... *why is it there?*
>
> In the example I gave, various functions having float128_t as arguments and
> return value. Specifically, std::float128_t modf(std::float128_t x,
> std::float128_t* y)
> is in the standard, and if one has the type std::float128_t and calls it as
> expected, one gets a long list of all other type variations of “modf” but
> this one is not mentioned.

If I need the ability to perform modulus on quad-precision floating point types
and I have the code for it, why would I bother overriding the standard
function in the first place? I'd just write my own and use it anyway.

The proposal does not give a means to detect whether a declaration is present
so it doesn't solve the problem for when the implementor didn't declare as
overridable-unimplemented in the first place (though that might be possible
with if constexpr (requires {})). This feature would therefore only be useful
where the functionality is explicitly conditional... but then the Standard can
simply declare a macro explaining the presence of the function. And if I have
a macro, I can create my own wrapper my_modf() that dispatches as necessary
depending on the platform in question.

-- 
Thiago Macieira - thiago (AT) macieira.info - thiago (AT) kde.org
  Principal Engineer - Intel DCAI Platform & System Engineering