On Fri, 1 May 2026 06:59:43 GMT, Julian Waters <[email protected]> wrote:
> It's ignoring the retain attribute explicitly and failing the compile now, > which is strange, given that in my own tests it never did that. At this point > I feel like the linker flag might be the better option, despite the flag > looking like a hack, all the source code changes we could try seem to be > messier than the flag. Let's remove the retain, it seems not to work with some toolchains. I added the linker flag ; with both the (hacky) linker flag AND the changed constructor the vtable should stay in the libjvm.so as desired. ------------- PR Comment: https://git.openjdk.org/jdk/pull/30771#issuecomment-4370346065
