Closed rob256 closed 3 years ago
Very nice - presumably this fixes quite a longstanding performance blooper?
Yeah. We moved from an IXSCAN+FETCH to a COLLSCAN, and this will allow us to do a straight IXSCAN. I've seen long queries on a collection which has a million documents, and this gives it a really nice speed up.
@rob256 any reason to not merge this?
@rob256 any reason to not merge this?
Hi @bmoscon, I can't see a reason why not! I did re-run the checks but a lot of the tests are failing (similar to the failures on master
).
I can update the changes.md once we're in a position to merge.
hi @bmoscon I'm working on this at Man AHL, and I was hoping to get this PR merged so I can go thru the release process. How do I get permissioned to work on this repo?
@dunckerr - if you work there I'm sure someone can add you to the org. @jamesblackburn can you help here?
This is mainly a reversion on https://github.com/man-group/arctic/pull/520, but we add another index to avoid a FETCH stage which gives a massive speedup.