bug#61142: 29.0.60; java-ts-mode - Wrong indentation for the body of multiple lines conditions

[Theodor Thornhill]

On 5 February 2023 11:15:47 CET, Eli Zaretskii <eliz@gnu.org> wrote:
>> From: Theodor Thornhill <theo@thornhill.no>
>> Cc: zjyzhaojiyang@hotmail.com, 61142@debbugs.gnu.org
>> Date: Sun, 05 Feb 2023 10:28:29 +0100
>> 
>> Eli Zaretskii <eliz@gnu.org> writes:
>> 
>> >
>> > The procedure is "git cherry-pick", but I don't think I understand
>> > what exactly is the problem here and what is the solution suggested to
>> > solve it.  Also, why does it need to be on emacs-29.  Can you help me
>> > understand that?
>> >
>> 
>> No need to have the test files on emacs-29.  I could just as well add a
>> test when emacs-29 is synced to master later.  The test-files themselves
>> don't exist on emacs-29.
>
>OK, but still: what exactly is the problem that your patch is trying
>to fix?  I didn't understand that from the original report.
>

Oh right! 

There are some indentation issues where if you have 

void foo(
  String foo) {
  // ...
}

Then the normal code will indent like

void foo(
  String foo) {
    //...
}

Because of the parent-bol anchor. When these paired brackets are nested (and 
some other cases) we have to find the grand-parent, but that caused some other 
issues with ie c-ts-mode where compound_statements can be nested _and_ 
siblings. Yuan made an effort to make this work in c-ts-mode due to the many 
bug reports. He then generalized it into c-ts-common, for use in other modes, 
like java.

This patch adds support for that in java.


>And one comment to the patch:
>
>> @@ -300,6 +302,20 @@ java-ts-mode
>>    (c-ts-common-comment-setup)
>>  
>>    ;; Indent.
>> +  (setq-local c-ts-common-indent-block-type-regexp
>> +              (rx (or "class_body"
>> +                      "\\`array_initializer"
>> +                      "annotation_type_body"
>> +                      "interface_body"
>> +                      "interface_constructor_body"
>> +                      "enum_body"
>> +                      "switch_block"
>> +                      "record_declaration_body"
>> +                      "block")))
>> +  (setq-local c-ts-common-indent-bracketless-type-regexp
>> +              (rx (or "if_statement" "do_statement"
>> +                      "for_statement" "while_statement")))
>> +  (setq-local c-ts-common-indent-offset 'java-ts-mode-indent-offset)
>>    (setq-local treesit-simple-indent-rules java-ts-mode--indent-rules)
>>  
>>    ;; Electric
>
>Why are we adding a c-ts-common-SOMETHING variable to java-ts-mode.el?
>Shouldn't it be named java-ts-SOMETHING instead?  Or am I missing
>something?

This is because there is some setup involved in the new bracket-counting code 
in c-ts-common.

Theo