Nick Crews

Not sure what is causing [the postgres failure](https://github.com/ibis-project/ibis/actions/runs/13428575878/job/37520654230#step:12:1) but I think it's unrelated. Details ``` ________________ test_insert_with_database_specified[postgres] _________________ [gw3] linux -- Python 3.10.16 /home/runner/work/ibis/ibis/.venv/bin/python con_create_database = @pytest.mark.notyet( ["flink"], reason="unclear whether...

As you can see in the updated tests, this change is potentially breaking for users who are trying to catch IbisTypeError. But, it does unify these two errors so they...

@cpcloud this is ready for another review whenever you get the chance. Thank you!

@cpcloud is there anything I can do to reduce the effort to review this? want to chat about this change live in eg zulip?

@cpcloud any way we can get this in as a breaking change for 11.0?

Yeah good idea, I'll do that and report back.

By running this several times: ```python import time import duckdb url_388mb = "https://github.com/NickCrews/apoc-data/releases/download/20250514-111142/income.csv" url_4mb = "https://github.com/NickCrews/apoc-data/releases/download/20250514-111142/campaign_form.csv" def timeit(sql): con = duckdb.connect() con.execute("SET enable_progress_bar = false;") start = time.time() con.sql(sql) end...

I believe this is solved in duckdb 1.3.0 with the [external file cache](https://duckdb.org/2025/05/21/announcing-duckdb-130.html#external-file-cache)

My first thought is that we should try to fix this upstream in duckdb. I would hope that in duckdb, `FROM read_xlsx(all_varchar=true)` and `FROM st_read(open_options = ['HEADERS=AUTO','FIELD_TYPES=STRING'])` should result in...

@omdaniel can you file an issue with duckdb and link to it here? Let's try to solve it there before we start to be hacky on our end.