Const qualification comments

[RalfJung]

I extracted some const-qualif knowledge from @eddyb. This is my attempt to turn that into comments.

Cc @oli-obk@eddyb

[rust-highfive]

r? @zackmdavis

(rust_highfive has picked a reviewer for you, use r? to override)

[zackmdavis]

r? @eddyb

[eddyb]

Nope, that's irrelevant. The problem is promoting e.g. &AtomicUsize and then mutating it.
That is, uses of the promoted references must not be able to distinguish between the promoted and unpromoted case (ignoring pointer addresses).

[RalfJung]

Well that's what I meant.^^

And also the part where any use of a const item must be equivalent to inlining its definition.

[RalfJung]

I have expanded the description a bit, is this better?

[eddyb]

Not just functions though, I think for everything? Promotion still has to do something in const/static, it's just slightly different than at runtime.

[RalfJung]

No idea, I just moved that comment up, I did not write it.^^

[RalfJung]

There's a comment in the file though saying

 // No need to do anything in constants and statics, as everything is "constant" anyway
// so promotion would be useless.
if self.mode != Mode::Static && self.mode != Mode::Const {

So it does seem to not do promotion the same way for statics and consts?

[eddyb]

This seems oddly specific - this is the original sole check, so comparing it to something newer seems off.
You could move the comment one line above and make it more about "any qualifications, except HasMutInterior (see above), disqualify from promotion".

[RalfJung]

I don't think which check came first historically has any bearing here. Just because it grew that way doesn't mean that's the best way to explain it.

The other check is IMO clearly "more principled" by looking at the place where the side-effect actually occurs.

[RalfJung]

I have adjusted the wording a bit.

[Centril]

Some proof-reading :)

[bors]

☔ The latest upstream changes (presumably #61548) made this pull request unmergeable. Please resolve the merge conflicts.

[bors]

☔ The latest upstream changes (presumably #61653) made this pull request unmergeable. Please resolve the merge conflicts.

[RalfJung]

This is bitrotting. @eddyb could you have a look at my edits?

[eddyb]

@bors r+

[bors]

📌 Commit 38c7f3e has been approved by eddyb

[bors]

☔ The latest upstream changes (presumably #61506) made this pull request unmergeable. Please resolve the merge conflicts.

[RalfJung]

Rebased.

@bors r=eddyb

[bors]

📌 Commit 0edf46f has been approved by eddyb

[bors]

⌛ Testing commit 0edf46f with merge 0efa0a6a57bfe2ed3e05e02850f3c18d8747b767...

[bors]

💥 Test timed out

[RalfJung]

@bors retry

[bors]

⌛ Testing commit 0edf46f with merge 9d9c7ad...

[bors]

☀️ Test successful - checks-travis, status-appveyor
Approved by: eddyb
Pushing 9d9c7ad to master...