stillwater-sc / universal / 29741065847

85%

master: 84%

Build Type

push

github

Committed by web-flow

Commit Message

feat(ereal): add fdim, modf, rint, nearbyint (parity with efloat) (#1170)

Adds the four <cmath> functions efloat had but ereal lacked, on ereal's own
expansion arithmetic:

- fdim (minmax.hpp): x > y ? x - y : +0; NaN propagates.
- modf (fractional.hpp): returns the fractional part (sign of x) and stores the
  integer part (trunc, toward zero) in *iptr.
- rint / nearbyint (truncate.hpp): round to nearest, halfway to EVEN (IEEE
  default) -- distinct from ereal's existing round() which is halfway away from
  zero. ereal has no dynamic rounding mode or FE_INEXACT flag, so nearbyint is
  identical to rint.

Verified against std:: on gcc and clang, including round-half-to-even ties
(2.5->2, 3.5->4, -2.5->-2, -3.5->-4) and modf int+frac == x. Test cases added to
the existing minmax/fractional/truncate suites; mathlib.hpp location notes
updated. cppcheck-clean.

Resolves #1165
Relates to #582

Coverage Stats

23 of 27 new or added lines in 3 files covered. (85.19%)

20 existing lines in 2 files now uncovered.

49789 of 58742 relevant lines covered (84.76%)

7022115.39 hits per line