Bug#182574: libc6: The pow() function gives incorrect results on special cases (nan, inf...)
On Thu, Feb 27, 2003 at 19:59:12 +0900, GOTO Masanori wrote:
> Also look at Appendix F.9.4.4. instead of 7.12.7.4-2.
OK, I didn't know that they were normalized. I had a quick look at
http://www.validlab.com/754R/standards/754.ps
(link from <http://www.validlab.com/754R/>), but I didn't see anything
concerning transcendental functions. Perhaps an addition of IEC 60559
over IEEE 754?
BTW, note that pow(±1, ±inf) returns a NaN, so
pow(1, nan) = 1
pow(1, inf) = 1
pow(1, -inf) = 1
are incorrect too.
Some choices are a bit strange, e.g. pow(x, +inf) for -1 < x < 0. The
standard says +0, but I think that -0 would have been more logical in
the rounding to minus infinity.
--
Vincent Lefèvre <vincent@vinc17.org> - Web: <http://www.vinc17.org/> - 100%
validated (X)HTML - Acorn Risc PC, Yellow Pig 17, Championnat International
des Jeux Mathématiques et Logiques, TETRHEX, etc.
Work: CR INRIA - computer arithmetic / SPACES project at LORIA
Reply to: