Formatter Bug: All values for `parentheses_positions_in_method_declaration` are ignored, while `...invocation` works

[devers2]

Problem

There appears to be a persistent bug in the Java formatter where the setting org.eclipse.jdt.core.formatter.parentheses_positions_in_method_declaration is completely ignored.

It seems that all possible values for this setting (e.g., separate_lines, preserve_positions, etc.) have no effect on formatting method declarations.

This is especially problematic for the separate_lines_if_wrapped value, which is a common and useful setting for code readability. Even when parameters are wrapped onto new lines, the closing parenthesis ) is still incorrectly forced onto the same line as the last parameter, directly violating the ...if_wrapped rule.

Key Evidence of Bug: Inconsistency with ...invocation

The strongest evidence of this bug is the clear inconsistency with its parallel setting, which works perfectly.

• Works: org.eclipse.jdt.core.formatter.parentheses_positions_in_method_invocation

  • This setting correctly applies all its values (like separate_lines_if_wrapped) to method calls.

• Fails: org.eclipse.jdt.core.formatter.parentheses_positions_in_method_declaration

  • This setting fails to apply the exact same logic to method declarations.

This inconsistency proves the bug is isolated specifically to the ..._in_method_declaration implementation.

Related Anomaly (Workaround Glitch)

We also observed a related anomaly that further highlights the bug:

If a user manually inserts a blank line (two newlines) before the closing parenthesis in a method declaration (e.g., ...status, \n \n ) {), the formatter preserves this structure.

This shows the formatter is not applying its own rules consistently. It fails to enforce the parentheses_positions... rule, but it can be "tricked" into the correct layout by this manual formatting glitch (likely due to join_wrapped_lines="false" behavior).

Conclusion

This is a bug in the Red Hat Java extension's formatter. The configuration settings for method declarations and invocations should behave identically, but they do not.

[fbricon]

Can you please attach a sample project with code and settings, reproducing this behavior?

[StrikeForceZero]

idk if it will work in vscode but try org.eclipse.jdt.core.formatter.parentheses_positions_in_method_delcaration

I was running into a similar problem when setting up Spotless (eclipse) in intellij for a Java project I'm working on and this same problem was driving me crazy.

turns out LLM's are good for something as it pointed out the typo in the official documentation:

https://help.eclipse.org/latest/topic/org.eclipse.jdt.doc.isv/reference/api/org/eclipse/jdt/core/formatter/DefaultCodeFormatterConstants.html#FORMATTER_PARENTHESES_POSITIONS_IN_METHOD_DECLARATION

Note that there is a typo (delcaration vs. declaration) in the option name. It has to remain, as fixing it would be a breaking change.

[fbricon]

nice catch! indeed, vscode-java would reuse the same typo'ed org.eclipse.jdt.core.formatter.parentheses_positions_in_method_delcaration