RFR: 8370976: Review the behavioral changes of core reflection descriptor parsing migration

[Roger Riggs]

On Fri, 31 Oct 2025 00:00:02 GMT, Chen Liang <liach at openjdk.org> wrote:

> There were a few other holes in the recent migration of BytecodeDescriptor, most notably:
> 
> 1. BytecodeDescriptor is missing checks for `.`, `[` characters, leading, trailing, consecutive `/`, or empty name.
> 2. EnclosingMethod is only validated by hotspot to carry either field or method type. We still need to check for field types
> 
> I have written up the behavioral changes in the CSR. In addition, I have added a few more tests to ensure the failure case behaviors of the migrated use sites.

Looks ok; it would be good to have a second reviewer.

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

Marked as reviewed by rriggs (Reviewer).

PR Review: https://git.openjdk.org/jdk/pull/28079#pullrequestreview-3473919128