GeoJSONSeq write support

[ttomasz]

Doc page says that GeoJSONSeq GDAL driver is supported for both reading and writing.

When I try to COPY with the following parameters: WITH (FORMAT GDAL, DRIVER 'GeoJSONSeq') I'm getting:

IO Error: GDAL Error (1): PROJ: proj_create: unrecognized format / unknown name

error.

DuckDB 0.9.1

[Maxxen]

Hi! Thanks for reporting this issue. This looks to be a regression introduced in #128, the GeoJSONSeq driver does not seem to handle an empty string as the "default" CRS identifier and instead errors when it tries to parse it, probably because it always tried to transform to wgs84 as that is mandated by the geojson standard. I should fix it so that we don't set the target SRS at all unless its given by the user.

To work around it for now you could simply specify a target srs. As I mentioned above valid geosjon requires WGS84, so you probably want to write something like:

COPY (SELECT ST_GeomFromText('POINT (1 1)')) 
TO 'test.json' 
WITH (FORMAT GDAL, DRIVER 'GeoJSONSeq', SRS 'EPSG:4326');

[zczc2390]

Seems there is a regression with the GDAL MVT driver on DuckDB 0.9.1 as well.
Try
COPY (SELECT ST_GeomFromText('POINT (1 1)')) TO 'test.mvt' WITH (FORMAT GDAL, DRIVER 'MVT');
or
COPY (SELECT ST_GeomFromText('POINT (1 1)')) TO 'test.mvt' WITH (FORMAT GDAL, DRIVER 'GeoJSONSeq', SRS 'EPSG:4326');
will get below error
Error: IO Error: Could not read from file "/tmp/test.mvt.temp.db": Bad file descriptor

The first command can work fine with DuckDB 0.8.1.

And is there any way to set the Dataset creation options for GDAL MVT driver in DuckDB? That would be much helpful to work with GDAL MVT driver

[Maxxen]

Hello @zczc2390! I think I have fixed the Bad file descriptor error in #145 but unfortunately the MVT driver is still broken since it spawns its own threads that are unaware of the thread local DuckDB filesystem context. Im going to need some more time to think about how to best solve it but hopefully I can provide a workaround soon.

For what its worth you should be able to pass any dataset creation options through the DATASET_CREATION_OPTIONS parameter, taking a sequence of KEY=VALUE varchar arguments.