[Cfp-interest 2142] scheduling message

[Jim Thomas]

The C/C++ liaison group has been discussing at some length possible conflicts (see below) with the WG21 proposal for extension floating-point types and the new C23 annex for TS-3.  Aaron Ballman, as SC22 chair, suggested a joint 1-hour meeting with CFP and interested WG21 members as a forum to deal with the concerns. After email discussion, Aaron, Rajan, and I decided to propose Wednesday, October 6, at 8 AM PDT / 11 AM EDT  / 3 PM UTC. Would you be able to attend a zoom meeting at that time? Please respond if possible by end of Tuesday, September 21.

- Jim Thomas

Aaron wrote:

> Here's the summary of topics I've pulled together from the various
> reflector discussions:
> 
> * Currently, the new types are exclusively available in the
> _Reserved_name spelling (like "_Bool" was in C99), there is no set of
> "pretty" macros (like "bool" was in <stdbool.h>) -- that means that if
> C++ decides to introduce new pretty names like (std::)float16_t, we
> have no immediately interoperable name (like "float16_t") that would
> be valid C, since there is no macro "#define float16_t _Float_16",
> say.
> 
> * The proposed rules for C and C++ have different arithmetic conversion rules:
>   float + _Float32 -> _Float32
>   float + std::float32 -> float
> 
> * The above point is observable in that you get UB when passing a
> _Float32 vs a float as a vararg parameter due to the promotion rules
> (float promotes to double, _Float32 does not).

> It's possible I've missed some points (there are a few threads going
> with a lot of messages on each of them), but Fred Tydeman and Joseph
> Myers have both weighed in on the thread, which is helpful! ...
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.oakapple.net/pipermail/cfp-interest/attachments/20210920/f539bcbc/attachment.htm>