Describe the bug When I run direnv edit, I see $EDITOR not found. and it opens TextEdit.

To Reproduce Run direnv edit with $EDITOR unset and $VISUAL set (e.g. to vim)

Expected behavior The editor specified by $VISUAL is used to edit .envrc

Environment

• OS: macOS
• Shell: zsh
• Direnv version 2.32.2

Additional context I expect this to be consistent with git config --edit, for example. I can't think of other examples, but I believe only having VISUAL set has worked out for me for other tools and direnv is the first tool to require EDITOR.

Describe the feature

There should be a way to cleanly stop/interrupt QueuePoller from another thread. It should immediately abort pending poll requests, but if the poll callback is executing, it should wait for that to complete before returning.

Use Case

I would to write code like Signal.trap("INT") { poller.stop }, so that I may use Ctrl+C to interrupt a daemon script that is polling SQS with QueuePoller.

Proposed Solution

The QueuePoller could have a method stop that can be invoked from any thread. If a call to poll is awaiting an SQS poll request, the request will be immediately aborted and poll will return normally. If poll is executing the callback when stop is invoked, then it will wait until the callback returns and then poll will return without making further poll requests.

Other Information

No response

Acknowledgements

• [X] I may be able to implement this feature request
• [ ] This feature might incur a breaking change

SDK version used

3.1.0

Environment details (OS name and version, etc.)

macOS

@saudraisn Did you ever figure this out? I'm experiencing a very similar issue.

Should the "ref" and "copy" lints be collapsed? Refs are Copy.

In particular, struct patterns, tuples and arrays can raise some unique formatting concerns.

I think there's a missing case of

let Some(x) = opt else {
    return
}

Also I don't see any examples with long patterns after let. Probably good to add variation to the types of patterns in general.

r? @Manishearth ?

I haven't been contributing for a while now so I think it's time. It's been fun!

Based on #103603

Lang items can be resolved in the AST rather than the HIR. And then any HIR structures referencing LangItem can be replaced with their "normal" couterpart. Specifically QPath::LangItem and GenericBound::LangItemTrait are factored out.

Best reviewed commit-at-a-time.

Closing (for now) since I haven't had time for rust for a long while 😢

It was previously assumed in #85556 that we can't check if an expression has an uninhabited type in typeck. So an unreachable_code case was added to the liveness pass instead. This PR moves that case back into typeck. It just requires adding cycle_delay_bug to the uninhabited query.

Incidentally this leads to some new cases for unreachable_code. But it should simply be consistent with how ! is currently handled.

My original motivation is to remove a blocker to doing some refactoring on liveness.

r? rust-lang/compiler

Unfortunately I haven't had much time for rust lately. @cjgillot PM'ed me and offered to take over here. I'm just going to close this PR.

@BrianLondon, @soc is just quoting earlier comments in this thread where they got into a squabble with another individual, so no need to worry about that. Generally the community is welcoming to different perspectives.