On Fri, 21 Jan 2022 18:39:02 GMT, Jan Lahoda <jlah...@openjdk.org> wrote:
> I believe this is an artifact of limited optimization of the InnerClasses > attribute: if the InnerClasses attribute refers to a class that is not part > of `ct.sym`, then it is stripped, but if the class is part of ct.sym, the > code does not check if it is used from the headers. So some InnerClass > entries may sadly be produced for classes that were only used inside method > bodies. Possibly an improvement for the future to do more aggressive > cleaning. (In general, recording more/unnecessary classes in the InnerClasses > shouldn't be harmful besides wasting space, I believe.) Good to know; thanks Jan. ------------- PR: https://git.openjdk.java.net/jdk/pull/7165