Closed amichaia closed 3 years ago
Hi @amichaia,
thanks a lot for the report! I'll take a look and let you know when it's fixed.
Hi @amichaia,
I pushed a new branch fix_31
which should fix the bug. Can you please check whether it works for you?
Great thanks! i'll give it a try and update you ASAP :)
I'm working with Amichai. This bug is fixed. But I found a new one. Will open a new issue.
Hi,
i tried to write 0 records with pandas like this: import pandas as pd d = {'col1': [], 'col2': [] df = pd.DataFrame(data=d) df.to_parquet('/test_data/data.parquet1_debug', compression=None)
when reading back with pandas, all works well and it says that its empty,
however, postgres crush.
i have looked inside the fdw code specifically in the populate_slot() function and saw that the slot container stays uninitialized and later is marked as valid by ExecStoreVirtualTuple() that is called from SingleFileExecutionState::next().
i did a local fix for myself that works for now, but i really dont know if what i did is ok. (i basically return false from populate_slot() if the slot attributes were not initialized )
Can you guys please take a look. i think that reproducing is really easy here.