[incorrect info] `to_chars` on Clang

[HolyBlackCat]

Which proposal(s) is / are affected?

P0067

Which compiler / toolchain is affected?

Clang

---

Currently both cppstat and cppref are wrong.

Cppstat incorrectly claims that v7 adds FP support, but v7 adds everything other than FP support.

Also the cppstat notes for v14 and v20 basically say the same thing, which is wrong. (v14 says "add float and double", while v20 says "everything is complete except long double" - same thing).

Also (this is where cppref is wrong), FP to_chars() support seem to have added in v16, not v14 (at least godbolt shows that: https://gcc.godbolt.org/z/jxfGvdTcr).

Also cppstat doesn't mention that long double to_chars falls back to double to_chars, probably a good idea to mention this.

[cdervis]

Hi @HolyBlackCat , what exactly is incorrect on cppstat?

We recently verified and updated P0067 support in #53 .

[HolyBlackCat]

I've edited the original post with more details.

[cdervis]

Thanks for clarifying.

I tested to_chars with Clang/libc++ 14, seems to work: https://godbolt.org/z/nr9av8TMe

So to summarize:

Type	to_chars	from_chars
Integral	7	7
FP	14	20
long double	- (falls back to double overload)	- (falls back to double overload)

I can add the note about long double -> double fallback, good to have it.

[HolyBlackCat]

Looks good to me, thanks. 🙂

---

I tested to_chars with Clang/libc++ 14, seems to work

Ahh, libc++ 14 and 15 seem to allow integral but not FP to_chars even with -std=c++14 and earlier. Clang 16 and newer defaults to C++17 and also no longer allows to_chars at all pre-C++17.

[cdervis]

Ahh, libc++ 14 and 15 seem to allow integral but not FP to_chars even with -std=c++14 and earlier. Clang 16 and newer defaults to C++17 and also no longer allows to_chars at all pre-C++17.

Thanks for sharing this bit! I have updated the data. I will improve the site itself sometime today so it correctly displays tables in the tooltip.