On Sat, 7 Dec 2024, 16:26 Jonathan Wakely, <cxx_at_[hidden]> wrote:

>
>
> On Sat, 7 Dec 2024, 15:24 Charles R Hogg via Std-Proposals, <
> std-proposals_at_[hidden]> wrote:
>
>> `std::nextafter` and friends provide the next representable value after
>> the first argument, in the direction of the second argument.
>>
>> Often, I find myself wanting specifically the next-higher or next-lower
>> representable value. In these cases, it feels awkward to cook up a number
>> that I know will be higher or lower. Yes, I can use
>> `std::numeric_limits`. I *think* `max` and `lowest` are the right tools
>> for the job. `infinity` is tempting, but I don't see a "negative infinity",
>>
>
> You just use -numeric_limits<T>::infinity()
>
>
> I don't want to raise questions in the reader's mind about negating
>> singular values like this, and `infinity` would be a disastrous choice for
>> integral types, where it returns `0`. (Not that we typically *intentionally
>> use* `nextafter` for integral types, but it may happen in a generic
>> template by accident.)
>>
>
> There is never a reason to use the numeric_limits<T>::infinity() value for
> the T in a generic function.
>

Sorry, that's a bit overly broad! I mean there's never a reason to use it
as the second argument to nextafter, because you always want a double or
long double value, because...


> nextafter takes a double for the 'to' argument so use
> ±numeric_limits<double>::infinity().
>
> nexttoward is the same but takes a long double, so use
> ±numeric_limits<long double>::infinity().
>
> That works even if the first argument is an integral type.
>
> You can also just use the HUGE_VAL and HUGE_VALL macros.
>
>
>
>> Sometimes I wonder: wouldn't it be more intent-based to provide something
>> like `std::next_higher()` and `std::next_lower()`, with a single argument?
>> Has this ever been proposed?
>>
>> Thanks,
>> Chip
>> --
>> Std-Proposals mailing list
>> Std-Proposals_at_[hidden]
>> https://lists.isocpp.org/mailman/listinfo.cgi/std-proposals
>>
>