Closed Bug 899936 Opened 6 years ago Closed 6 years ago
Monkey: fix faulty floating point logic in Extract Numeric Literal
The current integer-overflow handling in ExtractNumericLiteral casts doubles to int64 when the double can be inf, -inf, or outside the int64 range. There are unit tests that would have caught this but all the platforms on tbpl apparently cast these extremal doubles to out-of-range int64 values so the correct result is produced (NumLit::OutOfRangeInt). However, qemu and Marty's pandaboard expose the bug. Thanks to marty for helping to track this down.
Attachment #783657 - Flags: review?(mrosenberg)
Attachment #783657 - Flags: review?(mrosenberg) → review+
Status: ASSIGNED → RESOLVED
Closed: 6 years ago
Resolution: --- → FIXED
Target Milestone: --- → mozilla25
You need to log in before you can comment on or make changes to this bug.