[
https://issues.apache.org/jira/browse/GROOVY-11192?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17776896#comment-17776896
]
ASF GitHub Bot commented on GROOVY-11192:
-----------------------------------------
eric-milles commented on PR #1970:
URL: https://github.com/apache/groovy/pull/1970#issuecomment-1769290243
Generics is a very tricky area of the compiler. The diamond inference may
assume that the target type's generics exactly correspond to the constructor
type's generics. I don't think the `>` check is sufficient. You may have a
type like "class C<V,K> implements Map<K,V>".
`inferDiamondType` should be walking from 'Map' to `MapLong` calling
`parameterizeType` each time to fill in the type parameters. So, I would
expect a fix in `inferDiamondType` not `adjustGenerics` if something is amiss.
> Code that causes Groovy Compiler Crash
> --------------------------------------
>
> Key: GROOVY-11192
> URL: https://issues.apache.org/jira/browse/GROOVY-11192
> Project: Groovy
> Issue Type: Bug
> Components: Compiler
> Affects Versions: 4.0.15
> Environment: I am running on MacBook Pro, but I think this is a
> probably regardless of environment. I have been able to repeat this bug for
> Groovy versions from 3.0.10 to 4.0.15 and all in between.
> Reporter: John DeRegnaucourt
> Priority: Major
> Labels: Generics, compiler, crash, parametize, type
> Attachments: MapLong-3.groovy,
> TestCompilerCrashOnTemplateArgCount.groovy
>
>
> When attempting to compile the code below, it causes the Groovy compiler to
> crash with:
> java.lang.ArrayIndexOutOfBoundsException: Index 1 out of bounds for length 1
> at
> org.codehaus.groovy.ast.tools.GenericsUtils.createGenericsSpec(GenericsUtils.java:494)
> at
> org.codehaus.groovy.ast.tools.GenericsUtils.createGenericsSpec(GenericsUtils.java:480)
> at
> org.codehaus.groovy.ast.tools.GenericsUtils.parameterizeType(GenericsUtils.java:293)
>
> The crash is caused because the GenericType parameter is missing on line 9 in
> TestCompilerCrashOnTemplateArgCount:
> private Map<Long, String> map = new MapLong<>() // causes compiler crash
> ...but
> private Map<Long, String> map2 = new MapLong<String>() // doest not cause
> compiler crash
>
> The compiler is expecting one argument, but a no argument array was
> allocated. When it attempts to access the array at element [0], the compiler
> hits ArrayOutOfBoundsException.
>
> This is super easy to repeat with the two tiny source files attached.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
