Stuart Marks on Nostr: nprofile1q…6gyrf > Java is pretty tightly specified! any UB will be in dark ...
nprofile1qy2hwumn8ghj7un9d3shjtnddaehgu3wwp6kyqpqf3nlrmpjja65pzmvlegyl27l0y2m007nncxttdwqndrejajvdzask6gyrf (nprofile…gyrf) > Java is pretty tightly specified! any UB will be in dark corners, I think
Yeah we think so too… but we’re not sure. Behavior of bytecodes is quite tightly specified. There may be some seams between the VM spec, the language, and libraries, as well as some corners of the verifier and classloaders. Of course there are Unsafe and JNI and the foreign function+memory stuff which certainly admit to UB. But we haven’t established a boundary where we can say “there is no UB inside here.”
Yeah we think so too… but we’re not sure. Behavior of bytecodes is quite tightly specified. There may be some seams between the VM spec, the language, and libraries, as well as some corners of the verifier and classloaders. Of course there are Unsafe and JNI and the foreign function+memory stuff which certainly admit to UB. But we haven’t established a boundary where we can say “there is no UB inside here.”