[Cfp-interest 2263] Re: terminology issues

[Vincent Lefevre]

On 2021-11-09 11:39:17 +0100, Vincent Lefevre wrote:
> IIRC, this has already been mentioned, but 5.2.4.2.2p3 (and p4)
> defines the term "floating-point number", which is not used in a
> consistent way.
> 
> For instance, 7.12.7.3 in the latest C2x draft N2731 specifies the
> fabs function, saying
> 
>   The fabs functions compute the absolute value of a floating-point
>   number x.

More importantly, for FLT_MAX, etc., in 5.2.4.2.2p19,

  maximum representable finite floating-point number; if that number
  is normalized, its value is (1 − b^(−p)) b^(e_max)

should be changed to something like

  maximum representable finite number; if that number is a normalized
  floating-point number, its value is (1 − b^(−p)) b^(e_max)

and in p20,

  minimum positive floating-point number

should be changed to something like

  minimum positive number

(in case there would be tiny numbers that do not match the FP model,
and this would follow the idea in the footnote "the value is intended
to be a positive number no greater than the minimum normalized
positive number for the type.", i.e. it doesn't imply that the
FP model needs to be followed).

-- 
Vincent Lefèvre <vincent at vinc17.net> - Web: <https://www.vinc17.net/>
100% accessible validated (X)HTML - Blog: <https://www.vinc17.net/blog/>
Work: CR INRIA - computer arithmetic / AriC project (LIP, ENS-Lyon)