add Clang sanitizer checks in CI #711
Conversation
isaacbrodsky
force-pushed
the
fsanitize-in-ci
branch
from
89db472
to
b6b8638
Compare
isaacbrodsky
force-pushed
the
fsanitize-in-ci
branch
from
c8c53fa
to
58be7a0
Compare
| {0.659966917655, -2.1364398519395}, | ||
| {0.659966917655, -2.1364398519396}}; | ||
| {0.659966917656, -2.1364398519395}, | ||
| {0.659966917657, -2.1364398519396}}; |
There was a problem hiding this comment.
Why did these need to be changed?
There was a problem hiding this comment.
Possibly they didn't need to - I was trying to adjust some test cases to get better coverage although I don't think I really accomplished that.
There was a problem hiding this comment.
My mistake, they do need to be changed because as defined this is a straight line. polygonToCells will reject straight lines now - even if a cell centroid is along the line - because it would have attempted to divide by zero when estimating.
| double a = d * d / fmin(3.0, fabs((p1.lng - p2.lng) / (p1.lat - p2.lat))); | ||
| double lngDiff = p1.lng - p2.lng; | ||
| double latDiff = p1.lat - p2.lat; | ||
| if (lngDiff == 0 || latDiff == 0) { |
There was a problem hiding this comment.
For my info, how are NAN and Infinity values handled here? Do we need to check for something like isFinite as well?
There was a problem hiding this comment.
isfinite is checked below
|
Does this supersede #710? |
No, that PR appears to be addressing a different issue in the ratio calculation, it doesn't address the divide by zero problem. |
No description provided.