On Wednesday, April 30th, 2025 at 11:11 AM, Jan Schultke <janschultke@googlemail.com> wrote:

I don't think it makes much sense to use _BitInt for this, even if it was in the standard, which it isn't.

We will reference C23 soon, so it is.

You wouldn't need to parse in this function; the two byte spans would contain the little-endian bits of the variable-length integers that make up the integer and fraction. I don't see a compelling reason why you would take in the integer and fraction as an integer type. You're still passing the same data.

Oh, I see. We originate from the same idea.

As a side note, std::span<const std::byte>

could be used to make this clearer.

If we further talk about variable length

bits vs. fixed-length bits (_BitInt(K) or

std::array<std::byte>), I slightly prefer the latter

because when comes to to_chars, there is

some advantage to exchange the data

via return value, in which case span

can't be reused as it defines no storage.

Regardless this is less important than the

point: no parse and no format.

Zhihao Yuan, ID lichray
The best way to predict the future is to invent it.
_______________________________________________