Nick Crews
Nick Crews
@deepyaman Do you have any thoughts here?
@cpcloud Great suggestions, I accepted them, thanks!
The two fixup commits I had to add here (because I thought we errored when doing bogus comparisons, but we don't actually error) to adjust the docstrings inspired me to...
See https://github.com/ibis-project/ibis/pull/11311, where I tried to implement this. I'm not sure which version we should abandon and which we should pursue, but in general I think - We should error...
Also related, I'm not sure if this unsupported error should get pushed down in sqlglot: https://github.com/tobymao/sqlglot/issues/6376 and then possibly surfaced in ibis automatically: https://github.com/ibis-project/ibis/pull/11772
@mesejo I just updated the test, let me know if that is acceptable to you! I think using the handcrafted data will make it more clear what we are testing....
@mesejo any idea why [druid is failing](https://github.com/ibis-project/ibis/actions/runs/18603271186/job/53046631046?pr=11693) on my version? On [your version, the commit](https://github.com/ibis-project/ibis/pull/11693/commits/d357c42ef187d2dc81846e2cf172b0745a8f3f3b) right before, [it passed](https://github.com/ibis-project/ibis/actions/runs/18600205352/job/53036532796?pr=11693).
I'm guessing because in your version you only operated over a single column, where in mine we are ordering by a different column than we are aggregating over? Probably just...
huh, good detective work, thanks for cleaning up my mess. That is weird, since that seems like such basic usage of memtable, and memtable doesn't appear to be broken otherwise...