On Fri, 1 Nov 2024 13:58:01 GMT, Chen Liang <li...@openjdk.org> wrote:
> An oversight in the patch for > [JDK-8338544](https://bugs.openjdk.org/browse/JDK-8338544) #20665 caused > `ClassDesc` to incorrectly accept empty strings as binary names. This patch > fixes that problem, and unifies the checks for package and class nominal > descriptors. Please describe where the bug was caused and also record the evaluation explaining the cause in the issue. Here is one part I understand for your reference. One of the causes is that the refactoring done by JDK-8338544 converts the explicit check of the return value from `ConstantUtils::skipOverFieldSignature` called by `ClassOrInterfaceDescImpl::ofValidated` to `assert`. The refactoring was hard to review. So `validateBinaryClassName` only validates the binary name. Prior to JDK-8338544, the implementation expects more validations performed from `ClassDesc::ofDescriptor`. I think it requires a thorough understanding of what being validated before JDK-8338544 to see if anything else is missing. I am not sure what validation of package name is regressed. Please also explain. ------------- PR Comment: https://git.openjdk.org/jdk/pull/21830#issuecomment-2452218010
