MIR: Add pass that erases all regions right before trans #29886
Conversation
Yes.
Yes. |
|
Actually, thinking more on this, can we move this pass to later? (Vs doing it in a separate PR) It seems like we can easily modify the driver to pass an |
|
r+ if the nit about |
michaelwoerister
force-pushed
the
mir-erase-regions
branch
from
086342a to
0ce26db
Compare
|
I've moved region cleaning to |
|
@bors r+ |
|
📌 Commit 0ce26db has been approved by |
|
⌛ Testing commit 0ce26db with merge 716a00d... |
|
💔 Test failed - auto-mac-64-opt |
michaelwoerister
force-pushed
the
mir-erase-regions
branch
from
0ce26db to
c533902
Compare
|
@bors r=nikomatsakis |
|
📌 Commit c533902 has been approved by |
…akis This change adds a `MirPass` erasing all early-bound regions from MIR, right before storing it in the MIR map. I've added some assertions at neuralgic points in `trans::mir` doing cheap checks whether region have actually been erased. Here are some assumptions that I worked under: - AdtDef references stay untouched. It's the `Substs` accompanying them that need to be handled (e.g. in `AggregateKind::Adt`). - We can't really get rid of late-bound regions at this point because there is no version `BareFnTy` (for example) that comes without one. These still have to be handled on demand in trans. Are this assumptions right? r? @nikomatsakis
This change adds a
MirPasserasing all early-bound regions from MIR, right before storing it in the MIR map. I've added some assertions at neuralgic points intrans::mirdoing cheap checks whether region have actually been erased.Here are some assumptions that I worked under:
Substsaccompanying them that need to be handled (e.g. inAggregateKind::Adt).BareFnTy(for example) that comes without one. These still have to be handled on demand in trans.Are this assumptions right?
r? @nikomatsakis