two-phase-borrows need a specification

[pnkfelix]

Spawned off of #46887 (comment): While our short term solution for two-phase borrow support (#46037) is to make it as conservative as we can (e.g. the still planned #46747), the reality is that its complex and needs a specification.

(It might even be worth the exercise of adding them to the https://github.com/nikomatsakis/borrowck model )

[pnkfelix]

hmm is this "covered" by "adjust documentation" on #44100? Somehow I suspect that the adjustment expected there did not account for how subtle the corner cases here might be.

[pnkfelix]

cc @bobtwinkles who did the two phase borrows rewrite (#48770)

[nikomatsakis]

I'm going to push this to "the final milestone" -- I do think we want to "freeze" a spec for NLL as it is implemented (pending polonius improvements).

[pnkfelix]

Visited for T-compiler triage. I cannot imagine us blocking the release on this (even though it is of course desirable.)

I-nominating for discussion at next WG-compiler-nll meeting

[pnkfelix]

unnominating. We'll get to this when we get to all the NLL-deferred issues, and right now its just clogging up T-compiler process.

[pnkfelix]

Re-triaging for #56754. P-high. Assigning to self.

[pnkfelix]

(it would be good to take care of this in tandem with resolving the question of #56254 )

[pnkfelix]

subtask of #54129

[pnkfelix]

triage: should we realistically continue keeping this at P-high? (I've been absent but the fact that no one prioritized work-stealing it while I was gone for three weeks makes me wonder if it is in fact a P-medium issue...)

[pnkfelix]

triage: downgrading from P-high to P-medium.