Better StorageLive / StorageDead placement for constants.

[eddyb]

Fixes problems in miri (see rust-lang/miri#324 (comment)) caused by the new scope rules in #43932.
What I've tried to do here is always have a StorageLive but no StorageDead for 'static slots.
It might not work perfectly in all cases, but it should unblock miri.

r? @nikomatsakis cc @oli-obk

[nikomatsakis]

r=me if we add some mir-opt tests -- but I think it'd be nicer to convert the queries to return a tuple, and get rid of this silent business. That seems awfully branch-heavy and error-prone to me. So if you wanted to do that, I'd be happier =)

[eddyb]

@nikomatsakis It looks like I can't use mir-opt, as it looks for a subset, so the tests succeed even if I remove all the StorageDead from the expected MIR output - I only noticed this because EndRegion wasn't required and I thought this was special-cased for EndRegion.

[oli-obk]

Side note: once I get miri into the test suite, we'll have tests for this.

[nikomatsakis]

OMG this is so much better. I am disappointed about the tests though -- I think we have to fix that test harness, that "subset" property seems to bite everyone.

[nikomatsakis]

@bors r+

[bors]

📌 Commit 02ec0ae has been approved by nikomatsakis

[RalfJung]

I think we have to fix that test harness, that "subset" property seems to bite everyone.

Yes please :D

[bors]

⌛ Testing commit 02ec0ae with merge 23ade23...

[bors]

☀️ Test successful - status-appveyor, status-travis
Approved by: nikomatsakis
Pushing 23ade23 to master...