[
https://issues.apache.org/jira/browse/GROOVY-5502?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17457987#comment-17457987
]
Eric Milles commented on GROOVY-5502:
-------------------------------------
This kind of tracking was implemented in Groovy Development Tools for Eclipse
back in 2020. When a branch (if-else in this case) is reached, the assignments
in both are collected and if a variable is assigned in both branches, the
previous type can be discarded. See {{updateVariableSoft}} and
{{bubbleUpdates}}.
https://github.com/groovy/groovy-eclipse/blob/bc05754a9f4f2ae8986fb1ec9dfd78d8ae3bca19/base/org.eclipse.jdt.groovy.core/src/org/eclipse/jdt/groovy/search/TypeInferencingVisitorWithRequestor.java#L1314
https://github.com/groovy/groovy-eclipse/commit/48e9a464d56eebe1b42fe7ae07c2468413e44847
> If/else branch does not always infer the variable type
> ------------------------------------------------------
>
> Key: GROOVY-5502
> URL: https://issues.apache.org/jira/browse/GROOVY-5502
> Project: Groovy
> Issue Type: Improvement
> Components: Static Type Checker
> Reporter: Ariel Morelli Andres
> Priority: Minor
>
> Suppose we have the following code:
> {code}
> class A {
> void m() {
> }
> }
> class B extends A {
> }
> class C extends A {
> }
> @groovy.transform.TypeChecked
> class Test {
> void m() {
> def var = new Object() //If instead I put just "def var" it works as
> spected
> if (true) {
> var = new B()
> }
> else {
> var = new C()
> }
> var.m() //fails here
> }
> }
> {code}
> In this case, after the if/else structure we can infer that var is instanceOf
> A.
> But, the initialization seems to confuse the type inference.
> [Static type checking] - Cannot find matching method java.lang.Object#mA()
> at line: 23, column: 7
> If instead of {code}def var = new Object(){code} we put just {code}def
> var{code}, then the inference works fine.
--
This message was sent by Atlassian Jira
(v8.20.1#820001)
