Closed halfabrane closed 6 years ago
Merging #207 into master will increase coverage by
0.16%
. The diff coverage is90.47%
.
@@ Coverage Diff @@
## master #207 +/- ##
==========================================
+ Coverage 90% 90.16% +0.16%
==========================================
Files 52 52
Lines 1790 1789 -1
Branches 129 124 -5
==========================================
+ Hits 1611 1613 +2
+ Misses 179 176 -3
Impacted Files | Coverage Δ | |
---|---|---|
src/main/scala/magellan/BoundingBox.scala | 95.65% <100%> (+2.17%) |
:arrow_up: |
src/main/scala/magellan/Polygon.scala | 87.58% <100%> (-0.17%) |
:arrow_down: |
src/main/scala/magellan/PolyLine.scala | 89.47% <100%> (ø) |
:arrow_up: |
src/main/scala/magellan/Shape.scala | 58.53% <80%> (+6.03%) |
:arrow_up: |
src/main/scala/magellan/catalyst/SpatialJoin.scala | 95.23% <0%> (-2.39%) |
:arrow_down: |
...in/scala/org/apache/spark/sql/types/PointUDT.scala | 100% <0%> (+3.12%) |
:arrow_up: |
Continue to review full report at Codecov.
Legend - Click here to learn more
Δ = absolute <relative> (impact)
,ø = not affected
,? = missing data
Powered by Codecov. Last update a226690...63275a3. Read the comment docs.
Thanks @zebehringer
Previously Polygon#intersects(Line) used to be a strict intersection check With 62910c30, this behavior changed. This breaks BoundingBox#relate which now always returns Intersects even when it should return Within, creating a performance regression
This PR adds tests to BoundingBox#relate which if they existed would have caught the issue It also introduces a strict intersection check it is based off a PR from @zebehringer who first noticed this regression (thanks Zack!)
@zebehringer for review