Replace manual iterator exhaust with for_each(drop)

[clarfon]

This originally added a dedicated method, Iterator::exhaust, and has since been replaced with for_each(drop), which is more idiomatic.

This is just shorthand for for _ in &mut self {} or while let Some(_) = self.next() {}. This states the intent a lot more clearly than the identical code: run the iterator to completion.

At least personally, my eyes tend to gloss over for _ in &mut self {} without fully paying attention to what it does; having a Drop implementation akin to:

for _ in &mut self {}; unsafe { free(self.ptr); }

Is not as clear as:

self.exhaust(); unsafe { free(self.ptr); }

Additionally, I've seen debate over whether while let Some(_) = self.next() {} or for _ in &mut self {} is more clear, whereas self.exhaust() is clearer than both.

[rust-highfive]

r? @TimNN

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

[ollie27]

As I suggested in #45168 I think this should take self by value like all the other methods which consume an iterator.

Also the implementation should probably be written in terms of fold like #45168.

[TimNN]

Re-assigning to @rust-lang/libs:

r? @Kimundi

---

As I suggested in #45168 I think this should take self by value like all the other methods which consume an iterator.

I think that would kind of defeat the purpose of the method proposed here: Not dropping / consuming the iterator seems to be an explicit goal.

If consuming the iterator was the goal either mem::drop(iter) or iter.for_each(|_| ()) should work. In fact, .exhaust() would probably be equivalent to (&mut iter).for_each(|_| ()), I think.

[clarfon]

Note that |_| () could be worded as drop also.

[ExpHP]

And putting it all together with by_ref():

iter.by_ref().for_each(drop);
unsafe { free(self.ptr); }

I rather like the looks of what we can already write.

[Centril]

Is this operation common enough to warrant a dedicated method in Iterator?

[sfackler]

I think I agree that as_ref + for_each sufficiently covers this kind of thing.

[bors]

☔️ The latest upstream changes (presumably #47813) made this pull request unmergeable. Please resolve the merge conflicts.

[pietroalbini]

Ping from triage @Kimundi! This PR (probably) needs a review.

@sfackler so, should this PR be closed?

[clarfon]

(If there is a motion to close I'll open a new PR to replace the instances that I changed to exhaust to by_ref().for_each(drop))

[pietroalbini]

Ping from triage @rust-lang/libs! Can we get a review/decision on this PR?

[Kimundi]

Sorry, seems to have slipped under my radar. I think that I agree with @ExpHP that .by_ref().for_each(drop) is probably the better solution, as it just uses existing functionality.

I personally don't have any strong feelings about wether to add the new method or not, but I know that something like exhaust() method has been proposed multiple times in the past already, and not been added even before we had .for_each(), so I'd tentatively counts that as a argument against it.

[Centril]

@Kimundi Can we make this a matter of documentation then and include text about .by_ref().for_each(drop) somewhere on Iterator?

[scottmcm]

somewhere on Iterator

It feels mure like a "common patterns" document more than the doc comment. Is there a good place for that in one of the books? RBE is on doc.RLO now; maybe there?

[Centril]

cc @steveklabnik ^

[clarfon]

I've replaced instances of the method with for_each(drop) as preferred by everyone here.

[scottmcm]

Ooh, nice that this just works without needing .by_ref() 🙂

[clarfon]

When I saw this comment my immediate thought was to hope that the CI actually passed, because I was pretty sure it would work but not 100%. :p

[pietroalbini]

Ping from triage @Kimundi! If I understood correctly this PR still needs your review.

[pietroalbini]

Ping from triage! Can @Kimundi (or someone else from @rust-lang/libs) review this?

[Kimundi]

@bors r+

[bors]

📌 Commit 5c58eec has been approved by Kimundi

[bors]

⌛️ Testing commit 5c58eec with merge 1ef1563...

[bors]

☀️ Test successful - status-appveyor, status-travis
Approved by: Kimundi
Pushing 1ef1563 to master...