You may want to disable the int optimization (on by default; off for indy).  I 
think then primitives would get converted to wrappers and MOP would be 
available consistently.


________________________________
From: mg <[email protected]>
Sent: Monday, August 20, 2018 7:17 AM
To: [email protected]
Subject: Re: Inconsistent overriding of Interger methods

Without having looked at the implementation: Wouldn't "performance" be a 
plausible explanation why these cases can't be moped up... ?

-------- Ursprüngliche Nachricht --------
Von: Paul King <[email protected]>
Datum: 20.08.18 02:59 (GMT+01:00)
An: [email protected]
Betreff: Re: Inconsistent overriding of Interger methods

It appears to be most (all?) primitive/wrapper types not just Integer.

On Sun, Aug 19, 2018 at 11:00 PM ocs@ocs <[email protected]<mailto:[email protected]>> 
wrote:
Ladies and gentlemen,

the debate of null-propagation led me to bumping into a wildly inconsistent 
behaviour when one overrides Integer methods through the metaclass.

To me, this looks like a bug; even if this mess happens to be an intended 
behaviour, it is pretty weird (in this case, is it documented anywhere?)

===
768 /tmp> <q.groovy
java.lang.Integer.metaClass.byteValue={ -> "OK" }
java.lang.Integer.metaClass.plus={ o -> "OK" }
println "byteValue works: ${1.byteValue()}"
println "plus does not: ${1+2}"
println "not even: ${1.plus(2)}"
ArrayList.metaClass.plus={ o -> "OK" }
println "Elsewhere plus works: ${[]+1}"
768 /tmp> /usr/local/groovy-3.0.0-alpha-3/bin/groovy q
WARNING: Using incubator modules: jdk.incubator.httpclient
byteValue works: OK
plus does not: 3
not even: 3
Elsewhere plus works: OK
769 /tmp>
===

Thanks and all the best,
OC

Reply via email to