Thijs

> lgtm, tho I'm not sure if `get_order_by_expression` should be in partitioning module. It's used by window functions and an aggregate function, so this seemed like an okayish place to...

> desc + nulls first is wrong What do you mean by wrong? Is this different from what Pandas does, or how we did it in the past, or ......

> > > desc + nulls first is wrong > > > > > > What do you mean by wrong? Is this different from what Pandas does, or how...

Closing this PR. PR https://github.com/objectiv/objectiv-analytics/pull/1089 contains all commits in this branch, with a lot more refinement added

Coincidentally, this PR happens to fix another problem: if the data is `None` then the UUID validator gives a `TypeError` instead of ` ValueError`. and as a result `form.validate()` will...

Added Athena and BigQuery labels, as there are behaviour dfiferences between DB implemenations

After fixing this, remove this `if` from the test for this https://github.com/objectiv/objectiv-analytics/pull/1168#discussion_r962359476

Make sure to enable tests that are now skipped for Athena and BigQuery based on this limiations. Some tests( not complete!): * `bach/tests/functional/bach/test_df_setitem.py::test_set_series_column_name_with_spaces()` for all databases afterwards * see this...

Yes. ~There is still one sub-issue open https://github.com/objectiv/objectiv-analytics/issues/1317, but it's low prio and the main issue can be considered to be solved.~ edit: Actually completely solved

Going to close this issue: The main effort here has been done. All test tests use `get_df_with_test_data()`. Converted tests that are not yet working on all platforms are annotated with...