RFR: 8271820: Implementation of JEP 416: Reimplement Core Reflection with Method Handle [v16]

[Claes Redestad]

On Sun, 14 Nov 2021 19:54:50 GMT, Martin Grigorov <duke at openjdk.java.net> wrote:

>> Mandy Chung has updated the pull request with a new target base due to a merge or a rebase. The pull request now contains 43 commits:
>> 
>>  - fix copyright header and typo
>>  - improve documentation of AccessorUtils
>>  - Merge branch 'master' of https://github.com/openjdk/jdk into reimplement-method-invoke
>>  - Fall back to the VM native reflection support if method handle cannot be created
>>  - fix bug id in test
>>  - Merge
>>  - Merge branch 'master' of https://github.com/openjdk/jdk into reimplement-method-invoke
>>  - Merge branch 'master' of https://github.com/openjdk/jdk into reimplement-method-invoke
>>  - Separate paramFlgas into paramCount and flags fields
>>  - Minor cleanup.  Improve javadoc in CallerSensitiveAdapter
>>  - ... and 33 more: https://git.openjdk.java.net/jdk/compare/9a3e9542...46cb306b
>
> Thank you for the comments!
> There is no need to be harsh though! We just made use of old Reflection APIs. They were working for many years and according to StackOverflow Apache Wicket is not the only one doing this!
> In our case it is done in a test case, so it is easy to rework it. 
> 
>> Are there many Wicket tests trying to modify static field fields?
> 
> No! We have just one failing test with JDK 18 b23.
> 
> In someone else's case it might be needed to "hack" a third party library.
> 
> As a good citizen I'm reporting it back to you!

@martin-g I wasn't making any judgement. To qualify why this is not something I think anyone should do: poking into reflection internals to make a static final field mutable opens you up to situations where some readers still see the old value. An optimizing compiler (JIT or AOT) may treat the value as constant and propagates it. Future JVM features might decide to be even more aggressive, and trust these fields even more completely. While it's unspecified what should happen after overwriting a static final field with reflection, it does work if you're lucky. The one-off use in Wicket for a test seem benign enough. 

Alan: changing `Field.modifiers` still works, but dropping the final modifier is not enough for this to work in the new impl. It won't be hard to adapt to the new world. Users who relies on this today could for example opt-out of the new MH-based impl using `-Djdk.reflect.useDirectMethodHandle=false` and get the old behavior. I've checked using a minimal reproducer I extracted from the Wicket sources that this works.

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

PR: https://git.openjdk.java.net/jdk/pull/5027