Clean up and streamline snapshot data structures
These commits clean up the snapshot structures a bit, so they are more consistent with each other and with the `ena` crate.
They also remove the `OpenSnapshot` and `CommittedSnapshot` entries in the undo log, just like I did for the `ena` crate in https://github.com/rust-lang-nursery/ena/pull/14. This PR in combination with that `ena` PR reduces instruction counts by up to 6% on benchmarks.
r? @nikomatsakis. Note that this isn't quite ready for landing, because the `ena` dependency in the first commit needs to be updated once https://github.com/rust-lang-nursery/ena/pull/14 lands. But otherwise it should be good.
Instead, we talk about:
- creating the "next" universe
- universes "extending" one another
- and `u1.can_name(u2)`, meaning that `u1` contains all names from `u2`
The normalization of type-outlives predicates can depend on misc.
environment predicates, but not the other way around. Inferred lifetime
bounds can propagate type-outlives bounds far and wide, so their
normalization needs to work well.
Fixes#54467
This seemed like a good way to kick the tires on the
elided-lifetimes-in-paths lint (#52069)—seems to work! This was also
pretty tedious—it sure would be nice if `cargo fix` worked on this
codebase (#53896)!
Remove LazyBTreeMap.
It was introduced in #50240 to avoid an allocation when creating a new
BTreeMap, which gave some speed-ups. But then #50352 made that the
default behaviour for BTreeMap, so LazyBTreeMap is no longer necessary.
This leads to a lot of simplifications, as most code doesn't actually need to know about the specific lifetime/type data; rather, it's concerned with properties like name, index and def_id.
It was introduced in #50240 to avoid an allocation when creating a new
BTreeMap, which gave some speed-ups. But then #50352 made that the
default behaviour for BTreeMap, so LazyBTreeMap is no longer necessary.
This gives each type inference variable a notion of universe but doesn't
do anything with it. We can always get the "current universe" from
infer_ctxt. This relies on the property of type variables that they can
never interact with siblings.
