Anatoly Myachev

@pyrito for now, this most likely will not be merged, @vnlitvinov is right. By the way, I have an interesting case, which may not be very relevant to this task,...

> maybe we can wrap deployed functions with a function that just imports pandas as pd, so if the user tries to use pd inside an apply function (for example)...

@naren-ponder do you find the behavior strange? It would be more expected if it would be necessary to explicitly pass the tuple to work with the multi-index, like `df.loc[(bar, one)]`....

@naren-ponder In general you are right. But it seemed to me that there was already a precedent when we issued a warning for users that Modin's behavior in such and...

The behavior of pandas in this case is not erroneous, I looked at the [docs](https://pandas.pydata.org/docs/user_guide/advanced.html#basic-indexing-on-axis-with-multiindex). So we definitely need to fix the case. However, the previous question is still relevant.

> @anmyachev could you rebase the PR so we can re-run CI and make sure there are no issues here? done

> I'm curious about the performance implications here. Is there a way you could check quickly @anmyachev ? This is part of the change required for asynchronous execution. The performance...

> We should label things `PERF` if they're _the PR_ which helps performance. If they're building the base they could be `FEAT` or `REFACTOR`, but not `PERF`. Apparently yes, I...

Experimental implementation: https://github.com/modin-project/modin/pull/2626

This feature has not yet been implemented, as there is difficulty in ensuring compatibility with pandas. If compatibility is not needed, then it's possible to use `to_pickle_distributed`.