Make Vec::len() and Vec::is_empty() const

[jplatte]

Since these can trivially be const, I applied rustc_const_stable right away. Is this okay?

[rust-highfive]

r? @cramertj

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

[jyn514]

Since these can trivially be const, I applied rustc_const_stable right away. Is this okay?

@jplatte it means it has to go through FCP instead of only having a single reviewer approve. These seem like good candidates for const fn though.

[m-ou-se]

Did you run into a situation where having these as const fn is useful? Or is this only for consistency with [T]s methods?

[jplatte]

I made a bunch of types' parameterless constructors const (for no immediate gain as it turned out, but it worked) and it made sense to me that the methods on these types that check for an all-default / empty instance would also be const, but that didn't work because of is_empty not being const.

[m-ou-se]

(cc @rust-lang/wg-const-eval)

[RalfJung]

I seem to recall this was proposed before and we decided to wait until these methods are actually useful... but @oli-obk will probably remember better.^^

I personally do not think we should stabilize const fn that have no way of being useful. I have no objections to making them unstably const though.

[oli-obk]

Previous iterations of this PR: #66462 and #65341 and some more discussion in #55278 (comment)

We closed all such PRs before, maybe we should leave a comment on these functions informing future constifiers that we want to delay making them const fn until we have rust-lang/const-eval#20 (or until someone comes up with a use case)

[oli-obk]

I personally do not think we should stabilize const fn that have no way of being useful. I have no objections to making them unstably const though.

Oh, sure, we could make them unstably const fn and then mention in the tracking issue that we're blocking it on a use case being presented or const heap becoming that use case

[jplatte]

Oh, sorry for not being very diligent with looking for existing PRs. I did some searching but obviously not enough (maybe it was just issues).