GeometryType(geom) triggered an exception about sedona HOT 6 OPEN

@ruanqizhen can you show me the full stacktrace? The one that is after Caused by: java.lang.NullPointerException ?

from sedona.

@ruanqizhen can you show me the full stacktrace? The one that is after Caused by: java.lang.NullPointerException ?

This is the entire stacktrace it returned. Caused by: java.lang.NullPointerException is the last line.

from sedona.

@ruanqizhen I think the GeometryType might have indeterministic behavior over invalid geometries. We need to further investigate this issue.

Can you run ST_IsValid on your geometry column? Is there any invalid geometries? If you remove those and run GeometryType again, do you still have the same problem?

from sedona.

@ruanqizhen I think the GeometryType might have indeterministic behavior over invalid geometries. We need to further investigate this issue.

Can you run ST_IsValid on your geometry column? Is there any invalid geometries? If you remove those and run GeometryType again, do you still have the same problem?

It is not likely caused by the invalid geometry, because I've called the ST_MakeValid() for all of the geometries.

from sedona.

@ruanqizhen ST_MakeValid might not work for all cases. Any way to fix geometries is ST_Buffer(geom, 0).

If both of the method still cannot fix the issue, I think you can stick to ST_GeometryType. If you don't want the ST_ in the result, you can easily write a PySpark UDF to split the string by _ and only keep the second half.

from sedona.

@ruanqizhen ST_MakeValid might not work for all cases. Any way to fix geometries is ST_Buffer(geom, 0).

If both of the method still cannot fix the issue, I think you can stick to ST_GeometryType. If you don't want the ST_ in the result, you can easily write a PySpark UDF to split the string by _ and only keep the second half.

I'm using ST_GeometryType now. But I'm wondering if there is a way to find out which row caused the problem. Is there a way that I can "try", or skip the error, to let the query continue process other rows, so that I can then check which row was skipped.

from sedona.