Issue 1046: Order of arguments to powr

Authors: Stephen Heumann
Date: 2026-04-22
Submitted against: C23
Status: Open

In C23 7.12.8.7, the synopsis for the powr functions shows that the first parameter is named y and the second is named x:

double powr(double y, double x);

(and similarly for powrf, powrl, powrd32, powrd64, and powrd128).

The description then says that

The powr functions compute x raised to the power y as e^{y log_e x}.

So the powr functions are specified as computing their second argument raised to the power of their first argument. This is the opposite order of arguments from the original specification of powr in TS 18661-4:2015. It is also the opposite order of arguments from pow.

This change to the order of arguments is likely to cause confusion, and I can find no record that the committee did it intentionally, so it may be an editorial error. Moreover, most of the special cases for powr specified in F.10.5.7 do not make sense mathematically if powr actually computes its second argument raised to the power of its first argument.

The standard should be clarified to indicate the actually intended order of arguments for the powr functions. I suggest changing it back to match TS 18661-4:2015.

Suggested correction

In C23 7.12.8.7 paragraph 1 (Synopsis) and B.11, change all the powr functions to have x as the first parameter and y as the second.