[lworld] RFR: 8294678: [lworld] Update AccessFlags tests to include ACC_IDENTITY and ACC_VALUE [v2]

[Roger Riggs]

On Tue, 4 Oct 2022 20:03:59 GMT, Mandy Chung <mchung at openjdk.org> wrote:

>> Roger Riggs has updated the pull request incrementally with one additional commit since the last revision:
>> 
>>   Update ObjectMethods test
>
> src/java.base/share/classes/java/lang/Class.java line 1522:
> 
>> 1520:         final int mask = ((location == AccessFlag.Location.CLASS) ?
>> 1521:                 getClassAccessFlagsRaw() : getModifiers()) |
>> 1522:                 (isArray() ? Modifier.IDENTITY : 0);
> 
> For an array class `a`, `a.accessFlags()` and `AccessFlag.maskToAccessFlags(a.getModifiers(), Location.CLASS)` won't return the same `AccessFlag` set then.   The access flags for array type is special anyway.    It would be less confusing if `Class::accessFlags()` is consistent with `Class::getModifiers()`.

My goal was to have the Class.accessFlags() be consistent with Class methods `isIdentity()` and `isValue()`.
Unless we prevail in adding ACC_IDENTITY and ACC_VALUE to Class.getModifiers() there will be an inconsistency among the APIs.  `getModifiers()`, though the oldest, has the least flexibility.  The goal of the AccessFlags API was to be more flexible, correct, and extensible.  It is a question of what the end state should be and how to get there.
(The check for isArray() to select the INNER_CLASS location was from the original AccessFlag PR.)

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

PR: https://git.openjdk.org/valhalla/pull/775