Closed mwtoews closed 2 years ago
Also, there's nothing particularly unique with the example. For something different, see LINESTRING (0 0, 0 5, 5 5, 5 0, 0 0)
with buffer(10).
Doing a few version bisections, I've discovered that this regression was introduced with JTS 1.17.0.
This seems a duplicate of #876
Thanks, it is a duplicate!
Thanks for the detailed bug report. A fix has been found and a patch will be in place soon (in JTS and GEOS).
Testing with JTS 1.18.3 or since GEOS 3.9.x the buffer behavior changed for closed LineStrings. For example:
Buffer(5) is expected:
Buffer(25) is expected:
Buffer(27) has unexpected interior:
Buffer(35) has unexpected interior that intersects part of the input:
Buffer(60) has unexpected interior that contains the input:
This is a regression since older versions, e.g. JTS 1.15 (not sure exactly when) or GEOS 3.8.1, which get expected results for each example.
(Also, for reference this started from a discussion at https://github.com/shapely/shapely/discussions/1405)