vsapsai added a comment. I've updated D110123 <https://reviews.llvm.org/D110123> to be the way I was planning it to be and I was testing locally with it. So far, with this change the speedup over a baseline is ~20%, though measurements aren't super rigorous. I.e., no multiple runs to warm up the caches, no control for other background activity. At least, with this precision can claim there is no slow down in compilation time.
What's interesting, I was able to trigger more diagnostic. Specifically, I got warnings about `length` ambiguity because in NSStatusItem it is CGFloat <https://developer.apple.com/documentation/appkit/nsstatusitem/1529402-length>, while in NSString it is NSUInteger. I also had more diagnostic that's unclear how it is triggered. It can be a project issue or a bug somewhere else, need to check it more carefully. In theory, "set dedupe" approach should be slower than "no external" as we are iterating through shared dependencies. But in practice it isn't, which is puzzling. My current theory is that "no external" also feeds into correctness, so we might be doing more [correct] method overloading checks. Repository: rG LLVM Github Monorepo CHANGES SINCE LAST ACTION https://reviews.llvm.org/D109632/new/ https://reviews.llvm.org/D109632 _______________________________________________ cfe-commits mailing list cfe-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits