Looks like some tests are failing after merging in main branch. I will look at this.

I personally think raising is the most informative behavior, but open to thoughts from other core team members

It is worth mentioning that Series.clip also works in the same way as numpy.clip. I think it is rather logical result, in first step we are clipping lower numbers, in second step upper numbers. We end up with all numbers equal to upper.

Another similar example from pandas api, pandas.date_range - when we pass start > end there is no exception, but empty DatetimeIndex.

Another similar example from pandas api, pandas.date_range - when we pass start > end there is no exception, but empty DatetimeIndex.

pd.date_range makes sense to be more flexible in this sense because the freq can be negative allowing start > end to produce a non-empty date range

Ok, so I guess we also need to change Series.clip method to raise exception.

I am not sure which exception it should raise, some existing pandas exception? I guess ValueError might be appropriate.

I have started implementation and I think there is too much to make validation for scalars, lists and DataFrame arguments.

So if you insist on doing this, I can close this PR and unassign from issue.

I would suggest closing this PR now until there is some consensus on the main issue on how to handle this scenario

Thanks for the PR, but I think this scenario needs a little more discussion before moving forward so closing