Maintain negative inputs for the Distinct operator#7279
Conversation
|
Hey Moritz sorry for the drive-by review I know this PR is still a draft. Iff these changes are meant for this ticket https://github.com/MaterializeInc/database-issues/issues/2266 then I'm pretty sure you can achieve what its asking for without a feedback edge. |
|
Ruchir, much appreciated! I rewrote the patch to not use a feedback loop but rather compute the negation and concat it further on directly. Some tests seem to pass... |
|
Sorry for the drive by, and I should have said this earlier, but: one bit of fall out that would happen here is that |
|
Maybe one take is that both implementations are valuable, and that we should flesh out the reduce plan to explicitly indicate which one we want (e.g. so that if the planner notices that we would use the arrangement, we can stay with the current implementation, and if it notice that we would not use the arrangement it can go with the new implementation). |
| /// to combine and present results in the appropriate order. If we | ||
| /// were only asked to compute a single aggregation, we can skip | ||
| /// that step and return the arragement provided by computing the aggregation | ||
| /// that step and return the arrangement provided by computing the aggregation |
Signed-off-by: Moritz Hoffmann <mh@materialize.com>
Signed-off-by: Moritz Hoffmann <mh@materialize.com>
|
Closing this PR in favor of #7287. |
3 participants
Signed-off-by: Moritz Hoffmann mh@materialize.com