[ 
https://issues.apache.org/jira/browse/GROOVY-10714?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17780030#comment-17780030
 ] 

Eric Milles edited comment on GROOVY-10714 at 1/16/24 10:45 PM:
----------------------------------------------------------------

GROOVY-6189 links to GEP-12, which specifically mentions that types implemented 
by {{Closure}} represent a coercion-free match and so are preferred over 
SAM-type overloads.

For reference, the distance computed between {{Closure}} and several 
interesting types are currently as follows:
||Type||Argument-Parameter Distance||
|Closure|0|
|Cloneable|1|
|Serializable|1|
|GroovyCallable|1|
|Callable|2|
|Runnable|1|
|GroovyObject|1|
|GroovyObjectSupport|4|
|Object|14|
|SAM-type|13
or 12 (if param count helps; GROOVY-9881)
or 15 (if param count hurts; GROOVY-11121)|


was (Author: emilles):
GROOVY-6189 links to GEP-12, which specifically mentions that types implemented 
by {{Closure}} represent a coercion-free match and so are preferred over 
SAM-type overloads.

For reference, the distance computed between {{Closure}} and several 
interesting types are currently as follows:
||Type||Argument-Parameter Distance||
|Closure|0|
|Cloneable|1|
|Serializable|1|
|GroovyCallable|1|
|Callable|2|
|Runnable|1|
|GroovyObject|1|
|GroovyObjectSupport|4|
|Object|14|
|SAM-type|13 or 12 (if param count helps) or
15 (if param count hurts; GROOVY-11121)|

> STC: Callable, Runnable, Serializable overload preference for functional 
> argument (closure, lambda, etc.)
> ---------------------------------------------------------------------------------------------------------
>
>                 Key: GROOVY-10714
>                 URL: https://issues.apache.org/jira/browse/GROOVY-10714
>             Project: Groovy
>          Issue Type: Bug
>          Components: Static compilation
>    Affects Versions: 4.0.4
>            Reporter: Christopher Smith
>            Assignee: Eric Milles
>            Priority: Major
>
> This appears to be similar to GROOVY-9881, but it's specifically in the 
> method-overload procedure. Given a functional value type with two method 
> overloads:
> {code}
> interface Try<T> {
>   Try<T> andThenTry(Consumer<? super T>)
>   Try<T> andThenTry(Runnable)
> }
> {code}
> When this code is invoked from static code, the STC errors out on an 
> ambiguous method reference even if the method type isn't:
> {code}
> // AWS SDK 2 DynamoDbTable
> class DynamoDbTable {
>   void putItem(PutItemRequest<T>)
>   void putItem(Consumer<PutItemRequest.Builder<T>)
>   void putItem(T)
> }
> @CompileStatic
> class MyServiceClass {
>   void doThing() {
>     Try.success(putItemRequest())
>       .andThenTry(table::putItem) // T for Try<T> is PutItemRequest<I>
>   }
> }
> {code}
> produces
> {code}
> [Static type checking] - Reference to method is ambiguous. Cannot choose 
> between [Try<T> Try#andThenTry(Consumer<? super T>), Try<T> 
> Try#andThenTry(Runnable)]
> {code}
> I think this may have something to do with the relaxed SAM matching that is 
> used to bridge ambiguous closure syntax, but when a plain method reference is 
> used, there's no ambiguity available.



--
This message was sent by Atlassian Jira
(v8.20.10#820010)

Reply via email to