[lworld] RFR: 8374115: [lworld] test_classPrinter.cpp fails with Valhalla [v2]

[Casper Norrbin]

On Thu, 29 Jan 2026 13:46:10 GMT, Casper Norrbin <cnorrbin at openjdk.org> wrote:

>> Hi everyone,
>> 
>> `test_classPrinter` tests `classPrinter` using regex to verify that the output matches the expected format.  
>> The `print_classes` test includes (among other things) printing the class details for `java.lang.Integer`. 
>> 
>> On mainline, the static fields are printed like this:
>> 
>> Java mirror oop for java/lang/Integer: java.lang.Class 
>> {0x0000000efc0427a0} - klass: 'java/lang/Class' - flags: 
>>  - ---- fields (total size 19 words):
>> ...
>> - ---- static fields (2):
>>  - public static final 'MIN_VALUE' 'I' @136  -2147483648 (0x80000000)
>>  - public static final 'MAX_VALUE' 'I' @140  2147483647 (0x7fffffff)
>> ...
>> 
>> 
>> With the Valhalla changes, the output now includes `value`:
>> 
>> Java mirror oop for java/lang/Integer: java.lang.Class 
>> {0x000000069f040728} - klass: 'java/lang/Class' - flags: 
>>  - ---- fields (total size 19 words):
>> ...
>>  - ---- static fields (2):
>>  - public static final value 'MIN_VALUE' 'I' @136  -2147483648 (0x80000000)
>>  - public static final value 'MAX_VALUE' 'I' @140  2147483647 (0x7fffffff)
>> ...
>> 
>> 
>> When printing fields' access flags, we now also print `identity`/`value` depending on whether a field has identity. Most fields (including `MIN_VALUE`) do not have identity and therefore show `value`. The existing regex did not account for this new print, making the test fail. I have updated it accordingly and the test now passes again.
>> 
>> Testing:
>> - Tier 1
>
> Casper Norrbin has updated the pull request incrementally with two additional commits since the last revision:
> 
>  - alignment
>  - change to valhalla-only print

Thanks for reviewing! tier1 still looks good.

-------------

PR Comment: https://git.openjdk.org/valhalla/pull/1989#issuecomment-3822945239