There seem to be a number of spec issues around this. Shouldn't
findStaticSetter say something about what happens when the field is
final? Same for findSetter? This issue seems to be much bigger than just
a simple bug fix.
A CSR request will need to be filed.
David
-----
On 12/01/2019 5:07 am, Mandy Chung wrote:
This issue requires a spec change and it's a behavioral change.
Field::set on a static final field throws IAE. Although there are few
cases in the world that hack around it, I think the compatibility risk
is low in this change.
I added a suggested fix in the JBS issue.
Mandy
On 1/11/19 10:07 AM, Mandy Chung wrote:
Hi Adam,
This is indeed a bug. I suggest to do the check in the checkAccess
method closes to where it performs the instance final field check.
Please also add a test case.
Mandy
On 1/11/19 6:38 AM, Adam Farley8 wrote:
Hi All,
First; thanks for responding so quickly. :)
Second, it sounds like changing the contents of a final field is
acceptable under the spec, but *not* changing the contents of a static
final field.
The test case uploaded to the bug uses a static final field for the
test,
so I think we can agree that the ability to create a setter for a static
final field is still a bug.
It sounds like the same fix could work if we check for "static" when we
check for "final", with a modified message.
As Field.set (post-setAccessible) fails to set the value of a static
final
field, I think we can agree that unreflectSetter should fail to create a
setter MethodHandle for this same field.
Best Regards
Adam Farley
IBM Runtimes
