potential confusion points

[ajfriend]

df.sql(s) might be the intended, but users will probably first try writing df.query(s). but query is already used by pandas...

does that mean it makes sense to prefix with df.pqd.query?

[ajfriend]

There's a risk that a query works "accidentally" without specifying the table/df correspondence because duckdb looks for dataframe names in scope. can we turn off this behavior?

# df.sql('select * from df where hour(ts_utc) = 8') # note!! this works because of duckdb defaults
df.sql('select * from tbl where hour(ts_utc) = 8', tbl=tbl2)