Error creating table from Pandas DataFrame

[greyscaled]

If I switch to_pandas to to_polars, everything works as expected.

import glaredb
import pandas as pd
import polars as pl

con = glaredb.connect("glaredb://...")

df = con.sql("SELECT * FROM './b.parquet'").to_pandas()

con.execute("create table test as SELECT * FROM df")

External error: Execution error: failed to create table in storage: Schema error: Invalid data type for Delta Lake: Null

---

I did not test if this is specific to Parquet, and so it might be useful to just create a simple df in pandas and try the same thing.

I suggest we add a Python test covering this case.

[universalmind303]

@greyscaled do you have an example of the parquet file you used for this?

[greyscaled]

parquet_file.zip

Had to upload in a zip - GitHub doesn't seem to allow direct uploading of a .parquet file

[universalmind303]

hmm, yeah i'm not seeing the same error, but a different one

exceptions.ExecutionException: Error during planning: Unable to fetch table provider for 'test': Not a Delta table: no log files

[greyscaled]

@universalmind303 ah, this is my bad. The error referenced in the original post is when I attempt to query the table.

The Python script exits without error, but SELECT * FROM ... does. My apologies.