Currently in the writing code (eg ogr_write) we have a final cleanup block where we close the dataset (and check for errors, which was corrected in https://github.com/geopandas/pyogrio/pull/284):
But when an error happens while writing (and this is already bubbled up), we don't close the dataset, see the finally: block in the snippet above (we only clean up the current Feature variable).
When I compare that with the various read paths, it seems we always close the dataset in the finally block, eg in ogr_read:
Currently in the writing code (eg
ogr_write) we have a final cleanup block where we close the dataset (and check for errors, which was corrected in https://github.com/geopandas/pyogrio/pull/284):https://github.com/geopandas/pyogrio/blob/154466ab555f395d24b3d4873e685a841dd81fa6/pyogrio/_io.pyx#L2277-L2295
But when an error happens while writing (and this is already bubbled up), we don't close the dataset, see the
finally:block in the snippet above (we only clean up the current Feature variable).When I compare that with the various read paths, it seems we always close the dataset in the finally block, eg in
ogr_read:https://github.com/geopandas/pyogrio/blob/154466ab555f395d24b3d4873e685a841dd81fa6/pyogrio/_io.pyx#L1245-L1255