abs for IEEE floating point is slightly wrong.
Evaluating abs(-0) gives the answer -0. This unexpected since it breaks invariants like 1/(abs x) >= 0. So abs(-0) should be 0. This is also the norm for other languages.
Together with this change, signum should also be changed so signum(-0) is -0. This maintains the invariant abs x * signum x == x.
Trac metadata
| Trac field | Value |
|---|---|
| Version | 7.6.3 |
| Type | Bug |
| TypeOfFailure | OtherFailure |
| Priority | low |
| Resolution | Unresolved |
| Component | Compiler |
| Test case | |
| Differential revisions | |
| BlockedBy | |
| Related | |
| Blocking | |
| CC | |
| Operating system | |
| Architecture |