What do people think of a new GH tag for
I-slow issues that are fixed by enabling MIR inlining? Presumably it will make closing old issues easier if/when we do some MIR inlining by default on stable. See https://github.com/rust-lang/rust/issues/68667#issuecomment-620203064
I'd like that! It would also be good to harvest some real world test cases from that to beef up our mir-opt tests.
Jonas Schievink suggested using optimization-specific tags in T-compiler. How does
This sounds like a really good idea! With Wesley we also have made some changes that will for the moment not make it to
mir-opt-lvl=1 because with current codegen, they make performance regressions. Tagging the future issue that represents our change would definitely help keep track of it.