expose needs_drop under mem

[Gankra]

Note that I purposefully didn't update the intrinsic's documentation, because I think it makes sense for it be more... "honest" about its semantics.

[rust-highfive]

r? @brson

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

[Gankra]

Fixed the whitespace errors (had auto-whitespace-trimming disabled since Swift people got mad when you fixed that).

The unstable book error thing is uh, new to me.

[aturon]

cc @rust-lang/libs @rust-lang/lang

[alexcrichton]

👍

[eddyb]

So there's some backstory here: basically this was never implemented in the compiler in a principled way, to account for associated types. I've made the Copy check recursive recently but I don't believe that's relevant here.
What's left is to normalize associated type projections in the recursion and/or make an auto trait and rely on the trait system to handle the structural recursion involved here. cc @nikomatsakis

[gnzlbg]

@eddyb Is this a bug in the current implementation, and if so, is there an issue tracking it? If not, could you maybe expand on why needs_drop cannot be accurate?

[bors]

☔ The latest upstream changes (presumably #41847) made this pull request unmergeable. Please resolve the merge conflicts.

[alexcrichton]

ping @gankro looks like this needs a rebase? Looks like there's not a whole lot of opinions here, and we don't have a huge amount of ceremony about unstable apis, so I can r+ when you've rebased

[Mark-Simulacrum]

@aturon You nominated but didn't label with any teams, I've added both that you pinged (T-lang and T-libs).

[Gankra]

Rebased.

[Mark-Simulacrum]

@bors r=alexcrichton

[bors]

📌 Commit 1f01b09 has been approved by alexcrichton

[bors]

⌛ Testing commit 1f01b09 with merge 7ac844f...