This little program currently compiles fine, but should it?

Jens · December 14, 2015, 6:08pm

func ok() {
    let s = "see"; "Why are there exactly "
    0 == 0 * 42
    "warnings and errors in this code"
    let t = s + "?"
    print(t)
}
ok() // see?

johankj · December 14, 2015, 6:40pm

Because currently only unused variable/constant declarations have
warnings/fix-its attached to them.
There have been some discussion on making non-void functions warn of unused
results
<https://lists.swift.org/pipermail/swift-evolution/Week-of-Mon-20151207/001580.html>
and it’s possible that this would be in style with that.
Perhaps (basically) empty statements should also give a warning?

—Johan

···

On Mon, Dec 14, 2015 at 7:08 PM, Jens Persson via swift-dev < swift-dev@swift.org> wrote:

func ok() {
    let s = "see"; "Why are there exactly "
    0 == 0 * 42
    "warnings and errors in this code"
    let t = s + "?"
    print(t)
}
ok() // see?

_______________________________________________
swift-dev mailing list
swift-dev@swift.org
https://lists.swift.org/mailman/listinfo/swift-dev

Jens · December 14, 2015, 7:04pm

Yes, and I don't know if there is any particular reason why the == operator
for Int doesn't have @warn_unused_result while the one for Doubles has:

func f() { 1 == 2 } // OK
func g() { 1.0 == 2.0 } // Warning: Result of call '==' is unused

···

On Mon, Dec 14, 2015 at 7:40 PM, Johan Jensen <jj@johanjensen.dk> wrote:

Because currently only unused variable/constant declarations have
warnings/fix-its attached to them.
There have been some discussion on making non-void functions warn of
unused results
<https://lists.swift.org/pipermail/swift-evolution/Week-of-Mon-20151207/001580.html>
and it’s possible that this would be in style with that.
Perhaps (basically) empty statements should also give a warning?

—Johan

On Mon, Dec 14, 2015 at 7:08 PM, Jens Persson via swift-dev < > swift-dev@swift.org> wrote:

func ok() {
    let s = "see"; "Why are there exactly "
    0 == 0 * 42
    "warnings and errors in this code"
    let t = s + "?"
    print(t)
}
ok() // see?

_______________________________________________
swift-dev mailing list
swift-dev@swift.org
https://lists.swift.org/mailman/listinfo/swift-dev

--
bitCycle AB | Smedjegatan 12 | 742 32 Östhammar | Sweden

Phone: +46-73-753 24 62
E-mail: jens@bitcycle.com

dabrahams · December 15, 2015, 10:54pm

Yes, and I don't know if there is any particular reason why the == operator for Int doesn't have @warn_unused_result while the one for Doubles has:

func f() { 1 == 2 } // OK
func g() { 1.0 == 2.0 } // Warning: Result of call '==' is unused

That’s an oversight; please file a radar!

Because currently only unused variable/constant declarations have warnings/fix-its attached to them.
There have been some discussion on making non-void functions warn of unused results <https://lists.swift.org/pipermail/swift-evolution/Week-of-Mon-20151207/001580.html> and it’s possible that this would be in style with that.
Perhaps (basically) empty statements should also give a warning?

—Johan

func ok() {
    let s = "see"; "Why are there exactly "
    0 == 0 * 42
    "warnings and errors in this code"
    let t = s + "?"
    print(t)
}
ok() // see?

_______________________________________________
swift-dev mailing list
swift-dev@swift.org <mailto:swift-dev@swift.org>
https://lists.swift.org/mailman/listinfo/swift-dev

--
bitCycle AB | Smedjegatan 12 | 742 32 Östhammar | Sweden
http://www.bitcycle.com/
Phone: +46-73-753 24 62
E-mail: jens@bitcycle.com <mailto:jens@bitcycle.com>

_______________________________________________
swift-dev mailing list
swift-dev@swift.org <mailto:swift-dev@swift.org>
https://lists.swift.org/mailman/listinfo/swift-dev

-Dave

···

On Dec 14, 2015, at 11:04 AM, Jens Persson via swift-dev <swift-dev@swift.org> wrote:
On Mon, Dec 14, 2015 at 7:40 PM, Johan Jensen <jj@johanjensen.dk <mailto:jj@johanjensen.dk>> wrote:
On Mon, Dec 14, 2015 at 7:08 PM, Jens Persson via swift-dev <swift-dev@swift.org <mailto:swift-dev@swift.org>> wrote:

Jens · December 15, 2015, 11:42pm

Ok thanks, and what about the initial topic of this thread, ie stuff like
this:

func f() {
    1; "two"; 3.0
    [4, 5]; 6 * 7
    print("No warnings or errors!")
}
f() // No warnings or errors!

Should I file a bug for that too or is this by design / has this been
discussed somewhere already?

···

On Tue, Dec 15, 2015 at 11:54 PM, Dave Abrahams <dabrahams@apple.com> wrote:

On Dec 14, 2015, at 11:04 AM, Jens Persson via swift-dev < > swift-dev@swift.org> wrote:

Yes, and I don't know if there is any particular reason why the ==
operator for Int doesn't have @warn_unused_result while the one for Doubles
has:

func f() { 1 == 2 } // OK
func g() { 1.0 == 2.0 } // Warning: Result of call '==' is unused

That’s an oversight; please file a radar!

On Mon, Dec 14, 2015 at 7:40 PM, Johan Jensen <jj@johanjensen.dk> wrote:

Because currently only unused variable/constant declarations have
warnings/fix-its attached to them.
There have been some discussion on making non-void functions warn of
unused results
<https://lists.swift.org/pipermail/swift-evolution/Week-of-Mon-20151207/001580.html>
and it’s possible that this would be in style with that.
Perhaps (basically) empty statements should also give a warning?

—Johan

On Mon, Dec 14, 2015 at 7:08 PM, Jens Persson via swift-dev < >> swift-dev@swift.org> wrote:

func ok() {
    let s = "see"; "Why are there exactly "
    0 == 0 * 42
    "warnings and errors in this code"
    let t = s + "?"
    print(t)
}
ok() // see?

_______________________________________________
swift-dev mailing list
swift-dev@swift.org
https://lists.swift.org/mailman/listinfo/swift-dev

--
bitCycle AB | Smedjegatan 12 | 742 32 Östhammar | Sweden
http://www.bitcycle.com/
Phone: +46-73-753 24 62
E-mail: jens@bitcycle.com

_______________________________________________
swift-dev mailing list
swift-dev@swift.org
https://lists.swift.org/mailman/listinfo/swift-dev

-Dave

--
bitCycle AB | Smedjegatan 12 | 742 32 Östhammar | Sweden

Phone: +46-73-753 24 62
E-mail: jens@bitcycle.com

Jens · December 16, 2015, 10:27am

Filed:

github.com/apple/swift

[SR-245] Missing @warn_unused_result attributes in Standard Library

opened 10:25AM - 16 Dec 15 UTC

closed 09:06PM - 04 Aug 16 UTC

jepers

improvement standard library

| | | |------------------|-----------------|… |Previous ID | SR-245 | |Radar | None | |Original Reporter | @jepers | |Type | Improvement | |Status | Closed | |Resolution | Done | <details> <summary>Additional Detail from JIRA</summary> | | | |------------------|-----------------| |Votes | 0 | |Component/s | Standard Library | |Labels | Improvement | |Assignee | None | |Priority | Medium | md5: 8a2d58a566fb9e188b29e87376a9f71e </details> **relates to**: * [SR-778](https://bugs.swift.org/browse/SR-778) Forgetting the in keyword can sometimes lead to compiling code **Issue Description:** The Standard Library currently has almost 400 @warn_unused_result attributes and, as demonstrated here, that is not enough. (I am a fan of the proposal to make this warning default for all non-Void returning funcs.) I generated a list of potential missing @warn_unused result attributes by searching the Standard Library's generated interface for non void public funcs without the attribute, using something like: <a href="t" class="attachment">t</a>\npublic func \[^\n\]<u>\[-\]\[\>\]\[^\n\]</u> From that list I wrote these two functions to show my findings / some low hanging fruit for contributors: ``` none func asExpectedThisFuncHasWarningsOnEveryLine() { 1.0 + 2.0 1.0 - 2.0 1.0 * 2.0 1.0 / 2.0 1.0 % 2.0 1.0 == 2.0 1.0 != 2.0 1.0 < 2.0 1.0 > 2.0 1.0 <= 2.0 1.0 >= 2.0 } func butSurprisinglyThisFuncHasNoWarningsAtAll() { // Missing @warn_unused_result for these Int operators: 1 + 2 1 - 2 1 * 2 1 / 2 1 % 2 1 == 2 1 != 2 1 < 2 1 > 2 1 <= 2 1 >= 2 1 << 2 1 >> 2 1 & 2 1 | 2 1 ^ 2 // And it's the same for UInt, UInt8-64 and Int8-64. // And also for these: class C : NonObjectiveCBase {}; var c = C(); var oc: C? = C() typealias MBP = ManagedBufferPointer<Void, Void> isUniquelyReferenced(&c) isUniquelyReferencedNonObjC(&c) isUniquelyReferencedNonObjC(&oc) MBP(unsafeBufferObject: c) == MBP(unsafeBufferObject: c) zip([1, 2, 3], ["one", "two", "three"]) } ``` EDIT: Also, as suggested by @belkadan in the [swift evolution mailing list](https://lists.swift.org/pipermail/swift-evolution/Week-of-Mon-20160201/008619.html), I'm adding the related issue of missing warning(s) for unused literals, like eg: ``` none func perhapsConfusing() { let s = "Since this function compiles fine. One might " "think that string literals in Swift can be " "written like this. But that is not the case. " "Only the first string literal will be printed." "The rest (including eg this) will be silently " "ignored." 12.34 // As will this. print(s) // Will print: "Since this function compiles fine. One might " } ```

···

On Tue, Dec 15, 2015 at 11:54 PM, Dave Abrahams <dabrahams@apple.com> wrote:

On Dec 14, 2015, at 11:04 AM, Jens Persson via swift-dev < > swift-dev@swift.org> wrote:

Yes, and I don't know if there is any particular reason why the ==
operator for Int doesn't have @warn_unused_result while the one for Doubles
has:

func f() { 1 == 2 } // OK
func g() { 1.0 == 2.0 } // Warning: Result of call '==' is unused

That’s an oversight; please file a radar!

On Mon, Dec 14, 2015 at 7:40 PM, Johan Jensen <jj@johanjensen.dk> wrote:

Because currently only unused variable/constant declarations have
warnings/fix-its attached to them.
There have been some discussion on making non-void functions warn of
unused results
<https://lists.swift.org/pipermail/swift-evolution/Week-of-Mon-20151207/001580.html>
and it’s possible that this would be in style with that.
Perhaps (basically) empty statements should also give a warning?

—Johan

On Mon, Dec 14, 2015 at 7:08 PM, Jens Persson via swift-dev < >> swift-dev@swift.org> wrote:

func ok() {
    let s = "see"; "Why are there exactly "
    0 == 0 * 42
    "warnings and errors in this code"
    let t = s + "?"
    print(t)
}
ok() // see?

_______________________________________________
swift-dev mailing list
swift-dev@swift.org
https://lists.swift.org/mailman/listinfo/swift-dev

--
bitCycle AB | Smedjegatan 12 | 742 32 Östhammar | Sweden
http://www.bitcycle.com/
Phone: +46-73-753 24 62
E-mail: jens@bitcycle.com

_______________________________________________
swift-dev mailing list
swift-dev@swift.org
https://lists.swift.org/mailman/listinfo/swift-dev

-Dave

--
bitCycle AB | Smedjegatan 12 | 742 32 Östhammar | Sweden

Phone: +46-73-753 24 62
E-mail: jens@bitcycle.com

Chris_Lattner · December 16, 2015, 12:40am

Ok thanks, and what about the initial topic of this thread, ie stuff like this:

func f() {
    1; "two"; 3.0
    [4, 5]; 6 * 7
    print("No warnings or errors!")
}
f() // No warnings or errors!

Should I file a bug for that too or is this by design / has this been discussed somewhere already?

There is a discussion on the swift-evolution mailing list about making “@warn_unused_results” semantics the default for all non-void functions. To me, that seems like the cleanest path to address these issues.

-Chris

···

On Dec 15, 2015, at 3:42 PM, Jens Persson via swift-dev <swift-dev@swift.org> wrote:

On Tue, Dec 15, 2015 at 11:54 PM, Dave Abrahams <dabrahams@apple.com <mailto:dabrahams@apple.com>> wrote:

On Dec 14, 2015, at 11:04 AM, Jens Persson via swift-dev <swift-dev@swift.org <mailto:swift-dev@swift.org>> wrote:

Yes, and I don't know if there is any particular reason why the == operator for Int doesn't have @warn_unused_result while the one for Doubles has:

func f() { 1 == 2 } // OK
func g() { 1.0 == 2.0 } // Warning: Result of call '==' is unused

That’s an oversight; please file a radar!

On Mon, Dec 14, 2015 at 7:40 PM, Johan Jensen <jj@johanjensen.dk <mailto:jj@johanjensen.dk>> wrote:
Because currently only unused variable/constant declarations have warnings/fix-its attached to them.
There have been some discussion on making non-void functions warn of unused results <https://lists.swift.org/pipermail/swift-evolution/Week-of-Mon-20151207/001580.html> and it’s possible that this would be in style with that.
Perhaps (basically) empty statements should also give a warning?

—Johan

On Mon, Dec 14, 2015 at 7:08 PM, Jens Persson via swift-dev <swift-dev@swift.org <mailto:swift-dev@swift.org>> wrote:
func ok() {
    let s = "see"; "Why are there exactly "
    0 == 0 * 42
    "warnings and errors in this code"
    let t = s + "?"
    print(t)
}
ok() // see?

_______________________________________________
swift-dev mailing list
swift-dev@swift.org <mailto:swift-dev@swift.org>
https://lists.swift.org/mailman/listinfo/swift-dev

--
bitCycle AB | Smedjegatan 12 | 742 32 Östhammar | Sweden
http://www.bitcycle.com/
Phone: +46-73-753 24 62
E-mail: jens@bitcycle.com <mailto:jens@bitcycle.com>

_______________________________________________
swift-dev mailing list
swift-dev@swift.org <mailto:swift-dev@swift.org>
https://lists.swift.org/mailman/listinfo/swift-dev

-Dave

--
bitCycle AB | Smedjegatan 12 | 742 32 Östhammar | Sweden
http://www.bitcycle.com/
Phone: +46-73-753 24 62
E-mail: jens@bitcycle.com <mailto:jens@bitcycle.com>

_______________________________________________
swift-dev mailing list
swift-dev@swift.org <mailto:swift-dev@swift.org>
https://lists.swift.org/mailman/listinfo/swift-dev