[Pitch] Require "infix" for infix operator function declarations

jtbandes · July 12, 2016, 4:03am

Currently, "infix" is not required/allowed on an operator function
definition, but "prefix" and "postfix" are:

    prefix operator ^^ {} // valid
    postfix operator ^^ {} // valid
    infix operator ^^ {} // valid

    prefix func ^^(operand: Int) {} // valid
    postfix func ^^(operand: Int) {} // valid
    infix func ^^(lhs: Int, rhs: Int) {} // *error: *'infix' modifier is
not required or allowed on func declarations
    func ^^(lhs: Int, rhs: Int) {} // valid

It seems like this was removed because it can be inferred from the number
of arguments (
https://github.com/apple/swift/commit/3ad9c58c18f0331444114e2eae3e772e702c326f\).
But IMO the inconsistency from other operator function decls/defs is
jarring.

How does everyone feel about reinstating the "infix" modifier on functoins?
(It was removed before the open-source release and the advent of
swift-evolution, so I thought it'd be worth a public review.)

Jacob

saagarjha · July 12, 2016, 4:37am

I wouldn’t go as far as to require it, but having it for optional use “for symmetry" seems fine to me.

···

On Jul 11, 2016, at 21:03, Jacob Bandes-Storch via swift-evolution <swift-evolution@swift.org> wrote:

Currently, "infix" is not required/allowed on an operator function definition, but "prefix" and "postfix" are:

    prefix operator ^^ {} // valid
    postfix operator ^^ {} // valid
    infix operator ^^ {} // valid

    prefix func ^^(operand: Int) {} // valid
    postfix func ^^(operand: Int) {} // valid
    infix func ^^(lhs: Int, rhs: Int) {} // error: 'infix' modifier is not required or allowed on func declarations
    func ^^(lhs: Int, rhs: Int) {} // valid

It seems like this was removed because it can be inferred from the number of arguments (https://github.com/apple/swift/commit/3ad9c58c18f0331444114e2eae3e772e702c326f\). But IMO the inconsistency from other operator function decls/defs is jarring.

How does everyone feel about reinstating the "infix" modifier on functoins? (It was removed before the open-source release and the advent of swift-evolution, so I thought it'd be worth a public review.)

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

G_B · July 12, 2016, 5:52am

+1

Personally this one wouldn’t bother me to require, but It seems the Swift way is to not require things that can be inferred by context so I’m ok with just allowing it.

···

On Jul 11, 2016, at 9:03 PM, Jacob Bandes-Storch via swift-evolution <swift-evolution@swift.org> wrote:

Currently, "infix" is not required/allowed on an operator function definition, but "prefix" and "postfix" are:

    prefix operator ^^ {} // valid
    postfix operator ^^ {} // valid
    infix operator ^^ {} // valid

    prefix func ^^(operand: Int) {} // valid
    postfix func ^^(operand: Int) {} // valid
    infix func ^^(lhs: Int, rhs: Int) {} // error: 'infix' modifier is not required or allowed on func declarations
    func ^^(lhs: Int, rhs: Int) {} // valid

It seems like this was removed because it can be inferred from the number of arguments (https://github.com/apple/swift/commit/3ad9c58c18f0331444114e2eae3e772e702c326f\). But IMO the inconsistency from other operator function decls/defs is jarring.

How does everyone feel about reinstating the "infix" modifier on functoins? (It was removed before the open-source release and the advent of swift-evolution, so I thought it'd be worth a public review.)

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

Anton_Zhilin · July 12, 2016, 8:22am

Why don't we go the other direction? Example:

func ^^(prefix: Int) {}
func ^^(postfix: Int) {}
func ^^(lhs: Int, rhs: Int) {}

We will make 'prefix', 'postfix' and perhaps 'lhs' and 'rhs' required
external parameter names for operator functions.

···

2016-07-12 7:03 GMT+03:00 Jacob Bandes-Storch via swift-evolution < swift-evolution@swift.org>:

Currently, "infix" is not required/allowed on an operator function
definition, but "prefix" and "postfix" are:

    prefix operator ^^ {} // valid
    postfix operator ^^ {} // valid
    infix operator ^^ {} // valid

    prefix func ^^(operand: Int) {} // valid
    postfix func ^^(operand: Int) {} // valid
    infix func ^^(lhs: Int, rhs: Int) {} // *error: *'infix' modifier
is not required or allowed on func declarations
    func ^^(lhs: Int, rhs: Int) {} // valid

It seems like this was removed because it can be inferred from the number
of arguments (
https://github.com/apple/swift/commit/3ad9c58c18f0331444114e2eae3e772e702c326f\).
But IMO the inconsistency from other operator function decls/defs is
jarring.

How does everyone feel about reinstating the "infix" modifier on
functoins? (It was removed before the open-source release and the advent of
swift-evolution, so I thought it'd be worth a public review.)

Jacob

xwu · July 12, 2016, 8:51am

But is there anything actually wrong with the way it is today?

···

On Tue, Jul 12, 2016 at 3:22 AM, Anton Zhilin via swift-evolution < swift-evolution@swift.org> wrote:

Why don't we go the other direction? Example:

func ^^(prefix: Int) {}
func ^^(postfix: Int) {}
func ^^(lhs: Int, rhs: Int) {}

We will make 'prefix', 'postfix' and perhaps 'lhs' and 'rhs' required
external parameter names for operator functions.

2016-07-12 7:03 GMT+03:00 Jacob Bandes-Storch via swift-evolution <
swift-evolution@swift.org>:

Currently, "infix" is not required/allowed on an operator function
definition, but "prefix" and "postfix" are:

    prefix operator ^^ {} // valid
    postfix operator ^^ {} // valid
    infix operator ^^ {} // valid

    prefix func ^^(operand: Int) {} // valid
    postfix func ^^(operand: Int) {} // valid
    infix func ^^(lhs: Int, rhs: Int) {} // *error: *'infix' modifier
is not required or allowed on func declarations
    func ^^(lhs: Int, rhs: Int) {} // valid

It seems like this was removed because it can be inferred from the number
of arguments (
https://github.com/apple/swift/commit/3ad9c58c18f0331444114e2eae3e772e702c326f\).
But IMO the inconsistency from other operator function decls/defs is
jarring.

How does everyone feel about reinstating the "infix" modifier on
functoins? (It was removed before the open-source release and the advent of
swift-evolution, so I thought it'd be worth a public review.)

Jacob

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