@cgiffard ugh, premature optimisation is so annoying. There’s been cases where we’ve pulled N+1 fetches into pre-fetching outside the loop. Had to explain to another dev why we don’t always do that, especially when we know the limit on N is tiny.
@cgiffard internally, we’re getting to the point where we need a custom queue (mostly because of the edge-case interactions between certain jobs), but no way are we going to do that without planning and research.
@cgiffard ugh, I’ve got to do that. To make it worse, it’s the core part of our business and the most complicated :( (award interpretation, looks like it was written in a weekend. Is >20k lines in a single class)