[
https://issues.apache.org/jira/browse/GROOVY-7711?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15059932#comment-15059932
]
Frank Pavageau commented on GROOVY-7711:
----------------------------------------
I've submitted https://github.com/apache/groovy/pull/213 which fixes this issue
as well as GROOVY-7710 (on which [~githubbot] commented). The fix has been
decoupled from the one for GROOVY-7420.
> Wrong method resolution for an overridden method with varargs and covariant
> return type
> ---------------------------------------------------------------------------------------
>
> Key: GROOVY-7711
> URL: https://issues.apache.org/jira/browse/GROOVY-7711
> Project: Groovy
> Issue Type: Bug
> Components: Static Type Checker
> Affects Versions: 2.4.5
> Reporter: Frank Pavageau
> Assignee: Cédric Champeau
>
> Consider the following code:
> {code}
> class A {}
> class B {
> Object m(Object[] args) {
> new Object()
> }
> }
> class C extends B {
> A m(Object[] args) {
> new A()
> }
> }
> C c = new C()
> A a = c.m()
> {code}
> It has 50% chance of failing the static type check at the last line, because
> the STC can resolve to the base definition (with an {{Object}} return type)
> instead of the override, as it doesn't give precedence to the most specific
> definition in the "varargs with 0 argument" case.
> Like for GROOVY-7710, I have a fix for the bug, but it's also based on my
> pull request for GROOVY-7420 as it's once again in the same method. It's
> actually in the same commit, because once refactored a bit, the 2 missing
> method calls become obvious.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
