We run into issues where 2d geometries are being processed by code which expects 3d points (and 3d geometries getting handled by code which expects 2d points).
This results in the exception with message "Parameter must be array of length 3" or "Parameter must be array of length 2", thrown by AbstractPositionSequenceBuilder#add(double...).
After examining the geolatte-geom code I observe the static DECODERS list in class org.geolatte.geom.codec.db.oracle.Decoders. It returns (in effect) singleton decoders for decoding geometries.
Each decoder extends AbstractSDODecoder which contains a private CoordinateReferenceSystem<?> crs which is not thread safe in this situation. While one thread may use this variable another thread may set this variable to a new value.
This results in the aforementioned exception.
Can you verify my findings?
maesenka
commented
3 years ago
We run into issues where 2d geometries are being processed by code which expects 3d points (and 3d geometries getting handled by code which expects 2d points). This results in the exception with message "Parameter must be array of length 3" or "Parameter must be array of length 2", thrown by
AbstractPositionSequenceBuilder#add(double...).Our setup:
After examining the geolatte-geom code I observe the static
DECODERSlist in classorg.geolatte.geom.codec.db.oracle.Decoders. It returns (in effect) singleton decoders for decoding geometries. Each decoderextends AbstractSDODecoderwhich contains aprivate CoordinateReferenceSystem<?> crswhich is not thread safe in this situation. While one thread may use this variable another thread may set this variable to a new value.This results in the aforementioned exception. Can you verify my findings?