AaronBallman wrote: > > Was this discussed/reviewed/motivated? There are drawbacks to this approach > > outlined in #72383 > > @iains @jyknight @AaronBallman @Bigcheese > > The motivation is in #72383 and I comment in [#72383 > (comment)](https://github.com/llvm/llvm-project/issues/72383#issuecomment-2275135890) > > This is not reviewed. I wait for several weeks but got no response. And I > think it is good. So I choose to land it.
Thank you for the link! (FWIW, there's no problems with these changes having been landed; @ChuanqiXu9 is the code owner for modules and the PR was up for three weeks without discussion. This is just typical post-commit review feedback.) There's a fair amount of discussion on that thread in opposition to this approach, and a comment thread on an issue is not really visible to many people. I think this warrants an RFC for a broader discussion, so I'd appreciate temporarily reverting this patch. Some thoughts for the RFC discussion: * Given that this is not the default behavior for MSVC, should `clang-cl` behave differently than `clang`? * What are the impacts on other tooling (debugger, 3rd party static analysis, etc)? * Is this the correct default when considering intellectual property security (will people expect their full, original source to be something that can be pulled from these artifacts)? https://github.com/llvm/llvm-project/pull/102444 _______________________________________________ cfe-commits mailing list cfe-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
