-
Notifications
You must be signed in to change notification settings - Fork 1.8k
Open
Labels
Description
Missing functionality
I would love to be able to profile a table in a database without sampling down to pandas.
Proposed feature
Add ibis as a new backend.
As shown in their docs, this would open up profiling on a ton of platforms (ClickHouse, DuckDB, Flink, MySQL, Snowflake, etc.).
I have a work in progress here. Its not demo-able yet but the tests are passing!
Would the team be open to such a contribution?
Alternatives considered
One could just use sqlglot to limit dependencies. However, this would lose the DataFrame interface already followed by the pandas and spark backends.
Additional context
I see this issue from a while ago: #1552.
The suggestion I'm making is quite different. As I say above it would look like an additional backend like the spark functionality.
deepyamandeepyaman and noklam