> > > > The people who expressed opinions on nuking triggers from orbit (it's > the only way to be sure) have yet to offer up any guidance on how to > proceed from here, and I suspect it's because they're all very busy getting > things ready for v12. I definitely have an interest in working on this for > 13, but I don't feel good about striking out on my own without their input. > > Very interesting thread, but the current patch has been through two > CFs without comments or new patches, so I'm going to mark it "Returned > with feedback". I hope all this discussion will trigger more research > in this space. >
I've noticed that the zedstore efforts ran into the same problem that refactoring triggers has: we cannot determine which columns in a table will be affected by a trigger. so we have to assume that all of them will be. This causes a lot of unnecessary overhead with triggers. If we had a compilation step for triggers (which, ultimately means a compilation step for procedures) which kept a dependency tree of which tables/columns were touched, then we would have that insight. it's true that one dynamic statement or SELECT * would force us right back to keep-everything, but if procedures which did not do such things had performance benefits, that would be an incentive to code them more fastidiously.