`@discardableResult` works with `try` and `try!`, but not `try?`

[AnthonyLatsis]

Description

No response

Reproduction

@discardableResult
func foo() throws -> Int {}

func test() {
  try? foo() // warning: result of 'try?' is unused [expression_unused_optional_try]
}

Expected behavior

No warning

Environment

Swift version 5.11-dev (Swift 11ef6e5)

Additional information

No response

[li3zhen1]

Hi, I think I can fix this. but in lib/Sema/TypeCheckStmt.cpp there was a commit from 9 years ago:

Always warn when 'try?' is discarded.
rdar://problem/22195906
Swift SVN r31090

(as as? as! are all treated as non-discardable, and I feel like it's reasonable if try? behaves like them.)

I also found another corner case with discardable computed member reference, not sure if it's a expected behavior:

struct S {
    var s: S {
        @discardableResult
        get {
            return self
        }
    }
}

24 │
25 │ func test() {
26 │     S().s
   │         ╰─ warning: expression of type 'S' is unused
27 │ }
28 │

[AnthonyLatsis]

there was a commit from 9 years ago:

Always warn when 'try?' is discarded.
rdar://problem/22195906
Swift SVN r31090

Apparently, that was supposed to mean "warn even if the call returns Void", perhaps to encourage error handling or explicit discarding. @discardableResult should override these rules.

I also found another corner case with discardable computed member reference, not sure if it's a expected behavior

Is that even a reasonable thing to do? Either way the current behavior is a bug. Could you report it?

---

If you made up your mind to work on an issue (not just take it for maybe later), please give us a clear cue and ask us — you can ask me — to assign you. You can always hand it back if plans change on your side. Otherwise, there is really nothing wrong with someone else showing up and making progress faster, which we know is demotivating.