[swift-evolution] [Pitch] Introduce user-defined dynamically "callable" types

[John Holdsworth]

Isn’t there a third way? I wrote a bridge between Swift and Java without having to
change the compiler at all by using a code generator to generate bridging Swift
rather than having a Java importer. The script introspects the classes/packages
you’re interested in using a script and generates code such as the following:

    /// public java.lang.String java.lang.Object.toString()

    private static var toString_MethodID_9: jmethodID?

    open func toString() -> String! {
        var __locals = [jobject]()
        var __args = [jvalue]( repeating: jvalue(), count: 1 )
        let __return = JNIMethod.CallObjectMethod( object: javaObject, methodName: "toString", methodSig: "()Ljava/lang/String;", methodCache: &JavaObject.toString_MethodID_9, args: &__args, locals: &__locals )
        defer { JNI.DeleteLocalRef( __return ) }
        return __return != nil ? String( javaObject: __return ) : nil
    }

This bridging code, along with some supporting classes is compiled conventionally
along with your app. See:

https://github.com/SwiftJava/SwiftJava <https://github.com/SwiftJava/SwiftJava>
https://github.com/SwiftJava/java_swift/blob/master/Sources/JavaObject.swift <https://github.com/SwiftJava/java_swift/blob/3a48760a5528ccb31d998afbfcb09d291268674f/Sources/JavaObject.swift>
https://github.com/SwiftJava/SwiftJava/blob/master/src/genswift.java <https://github.com/SwiftJava/SwiftJava/blob/master/src/genswift.java>

Python’s introspection seems to be reasonably capable and could certainly support
this approach. You’d need some sort of omni-type enum for the arguments and return.

https://stackoverflow.com/questions/196960/can-you-list-the-keyword-arguments-a-python-function-receives <https://stackoverflow.com/questions/196960/can-you-list-the-keyword-arguments-a-python-function-receives>
https://docs.python.org/2/c-api/intro.html <https://docs.python.org/2/c-api/intro.html>

John

> On 11 Nov 2017, at 16:13, David Hart via swift-evolution <swift-evolution at swift.org> wrote:
> 
> 
> On 11 Nov 2017, at 16:02, Joe Groff via swift-evolution <swift-evolution at swift.org <mailto:swift-evolution at swift.org>> wrote:
> 
>> 
>> 
>> On Nov 10, 2017, at 8:35 PM, Chris Lattner via swift-evolution <swift-evolution at swift.org <mailto:swift-evolution at swift.org>> wrote:
>> 
>>> 
>>>> On Nov 10, 2017, at 6:12 PM, Slava Pestov via swift-evolution <swift-evolution at swift.org <mailto:swift-evolution at swift.org>> wrote:
>>>> 
>>>> 
>>>> 
>>>>> On Nov 10, 2017, at 6:10 PM, Matthew Johnson via swift-evolution <swift-evolution at swift.org <mailto:swift-evolution at swift.org>> wrote:
>>>>> 
>>>>> Setting this aside, I’m very curious to hear whether type providers influence your thinking after you’ve had a chance to look into them.  I have always thought they were very cool.
>>>> 
>>>> I’m in favor of solving this problem with something like type providers also. The required compiler changes would be significant but would also clean up the interface between the ClangImporter, Sema and Serialization. If done right it would be a net gain that would benefit all users, instead of just adding YetAnotherCornerCase™ that makes implementation maintainers curse and scream.
>>> 
>>> I find it ironic that you’re talking pejoratively about a feature that has very narrow impact, complaining about how much of an impact on the compiler it would have, and then pine for a hugely invasive feature - one that would cause a ton of code churn, and probably wouldn’t actually be enough to eliminate the special cases in place because of ObjC interop.
>> 
>> You underestimate the impact this would have on function call type checking, but since this is an additive, non-ABI-stability feature, I have trouble considering it a candidate for Swift 5, so I don't think there's a time constraint forcing us to consider the "narrow" vs "huge" dimension. What's the best thing for the language and tools in the long term? This is a feature that influences the semantics of potentially any call site in all Swift code, which we'd have to live with forever if we accepted it now. Opening up the compiler architecture to make custom importers easier to write is a great solution to a ton of problems, including yours I think, without adding complexity to the core language. Experience in .NET land seems to show it's a great technique for integrating dynamic systems with static type systems, without poking unnecessary holes in the static language's type system
> 
> I agree with Joe. I also think that it would be very important to compare both approaches in detail before choosing one, because the last thing we want is to end up with both in the language. And if this proposal is accepted, we might refrain from introducing custom importers later, even if they are the better long term solution.
> 
> I want Swift to continue to shine for a very long time because I enjoy this language and I just want to make sure we don’t jeopardize that by choosing a quick solution without taking in consideration other solutions.
> 
>> -Joe
>> 
>>> -Chris
>>> 
>>> 
>>> _______________________________________________
>>> swift-evolution mailing list
>>> swift-evolution at swift.org <mailto:swift-evolution at swift.org>
>>> https://lists.swift.org/mailman/listinfo/swift-evolution <https://lists.swift.org/mailman/listinfo/swift-evolution>
>> _______________________________________________
>> swift-evolution mailing list
>> swift-evolution at swift.org <mailto:swift-evolution at swift.org>
>> https://lists.swift.org/mailman/listinfo/swift-evolution <https://lists.swift.org/mailman/listinfo/swift-evolution>
> _______________________________________________
> swift-evolution mailing list
> swift-evolution at swift.org <mailto:swift-evolution at swift.org>
> https://lists.swift.org/mailman/listinfo/swift-evolution <https://lists.swift.org/mailman/listinfo/swift-evolution>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.swift.org/pipermail/swift-evolution/attachments/20171111/1c74f7db/attachment.html>