efriedma-quic wrote: > > I'd be okay with introducing some form of "weak" builtin, where we model > > the call as a builtin function while only doing some approximate > > verification of the signature. > > Is that feedback _for me_, or [in > response](https://github.com/llvm/llvm-project/pull/231696#issuecomment-6176614178) > to @ojhunt ?
That was in reply to the question from @ojhunt . I wouldn't consider it a blocker, I think. > > I think this patch is missing serialization bits. CC @ChuanqiXu9 > > I don't follow, do you @efriedma-quic or @ChuanqiXu9 have more info? https://github.com/llvm/llvm-project/pull/200994 . But looking more closely, I guess this doesn't actually add any new state that needs to be serialized; the lookup is deferred until the builtin is actually requested, then the result of the lookup is only stored in the builtin declaration itself. https://github.com/llvm/llvm-project/pull/231696 _______________________________________________ cfe-commits mailing list [email protected] https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
