feat(ir): allow the creation of lists from columns

[deepyaman]

Is your feature request related to a problem?

I (possibly on behalf of @lostmygithubaccount) am frustrated by writing my_column.to_pyarrow().to_pylist() to get a Python list from an Ibis column.

What is the motivation behind your request?

Discussion on #10084 (comment) during triage. The shorthand my_column.to_pylist() was also proposed, but list(my_column) would be so much more Pythonic.

Describe the solution you'd like

list(my_column)

I don't need list(my_table) or list(some_scalar) to be implemented.

What version of ibis are you running?

dev

What backend(s) are you using, if any?

N/A

Code of Conduct

• I agree to follow this project's Code of Conduct

[deepyaman]

I can take a stab at this (later this week?), as discussed in triage.

[lostmygithubaccount]

can I just express frustration in meetings and have y'all take it from there? 😂 love it

[jcrist]

but list(my_column) would be so much more Pythonic.

I'm concerned about having __iter__ on columns result in (potentially expensive) query execution. We don't support __bool__/__int__/__float__ on scalar values for the same reason. Accidental coercion like this can be a performance footgun for users, resulting in query execution in unintended locations or occurring multiple times. Other lazy systems (dask, polars) also don't do automatic execution in python magic methods for similar reasons.

I'd much rather have a new to_* method that does this. Why not Column.to_list?

[deepyaman]

but list(my_column) would be so much more Pythonic.

I'm concerned about having __iter__ on columns result in (potentially expensive) query execution. We don't support __bool__/__int__/__float__ on scalar values for the same reason. Accidental coercion like this can be a performance footgun for users, resulting in query execution in unintended locations or occurring multiple times. Other lazy systems (dask, polars) also don't do automatic execution in python magic methods for similar reasons.

I'd much rather have a new to_* method that does this. Why not Column.to_list?

Sorry, forgot to respond. Column.to_list seems fine.

[IndexSeek]

I have also found myself using the to_pyarrow().to_pylist() approach, particularly when feeding a Streamlit selectbox or multiselect widget.

Going directly to a Python list from a column would be very convenient with something like to_list.