[Bug tree-optimization/126467] [13/14/15/16/17 Regression] wrong code with folding 0.0 - x
"amonakov at gcc dot gnu.org via Gcc-bugs" <[email protected]>
| Newsgroups | gmane.comp.gcc.bugs |
|---|---|
| Message-ID | <[email protected]/bugzilla/> |
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=126467
Alexander Monakov <amonakov at gcc dot gnu.org> changed:
What |Removed |Added
----------------------------------------------------------------------------
CC| |amonakov at gcc dot gnu.org
--- Comment #4 from Alexander Monakov <amonakov at gcc dot gnu.org> ---
No, the first sentence is in fact correct, x-y is the same as x+(-y), which by
commutativity is the same as (-y)+x. The problem is that positive zero is _not_
the neutral element for addition under default rounding, but negative zero is:
(0.0 + x) is not the same as x when x is -0.0
(-0.0 + x) and (x - 0.0) are the same as x under default rounding.