[
https://issues.apache.org/jira/browse/GROOVY-8200?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16028400#comment-16028400
]
Daniel Hammer commented on GROOVY-8200:
---------------------------------------
@[~blackdrag]; ahh, I see, I had forgotten Groovys support for
operator-overload, this complicates matters a bit.
On one hand I see the use-case for meta-programming support, and on the other
hand I appreciate the simplicity of Groovys truth convention and boolean
operators as-is.
I know which approach I would prefer, but I'm unable to fully comprehend the
consequences of either.
Quite the conundrum :)
> Shorthand |= results in NPE
> ---------------------------
>
> Key: GROOVY-8200
> URL: https://issues.apache.org/jira/browse/GROOVY-8200
> Project: Groovy
> Issue Type: Bug
> Affects Versions: 2.3.11
> Environment: Groovy Version: 2.3.11 JVM: 1.7.0_80 Vendor: Oracle
> Corporation OS: Linux
> java version "1.7.0_80"
> Java(TM) SE Runtime Environment (build 1.7.0_80-b15)
> Java HotSpot(TM) 64-Bit Server VM (build 24.80-b11, mixed mode)
> Reporter: Daniel Hammer
> Assignee: Jochen Theodorou
> Attachments: shorthand_and.groovy, shorthand_or_symptom.groovy,
> shorthand_xor_symptom.groovy
>
>
> I've stumbled across some curious behaviour for the shorthand assignment
> operator in Groovy v2.3.11 (symptom also present in Groovy v2.4.9).
> The following [boolean or logic|^shorthand_or_symptom.groovy], unexpectedly
> throws {{NullPointerException}}.
> {code}boolean x = null
> // short hand unrolled: x = null || x
> x |= null // -> unexpected NullPointerException
> assert !x{code}
> The same approach for [boolean and logic|^shorthand_and.groovy], performs as
> expected.
> {code}boolean x = null
> // short hand unrolled: x = null && x
> x &= null
> assert !x{code}
> It's possible to work around the issue with explicit cast.
> {code}x |= null as boolean{code}
--
This message was sent by Atlassian JIRA
(v6.3.15#6346)
