Aaron Ballman

> I don't see how this could introduce any potential security issue. https://security.stackexchange.com/questions/161799/why-does-checksec-sh-highlight-rpath-and-runpath-as-security-issues > If someone doesn't trust our compiler or the location they install our compiler/runtime, they can't trust...

> > It's not about the location of where they install the our compiler, it's that other things on the system (which the user may or may not have control...

> @AaronBallman if you have time to look at this, it seems that hidden visibility is honored in a strange way by icc here. Just to double-check, is this only...

So ICC gets one behavior, ICX gets another behavior, and Clang has a third behavior? That is a surprise. > That said, to be time efficient, I'm not sure if...

I filed an internal JIRA ticket to see if we can track the version of clang we base an ICX build on to help with this sort of situation in...

Thanks for the extra information, that really helps! > So, while its unfortunate I'm doing clang tests on a different Linux box, I certainly can't find evidence of a clang-specific...

FWIW, Clang's C indexing APIs have a TU-level flag named `CXTranslationUnit_DetailedPreprocessingRecord`. Passing that causes the indexing APIs to behave as if `-Xclang -detailed-preprocessing-record` was passed to the compiler which *might*...

I would be happy to initiate a discussion with EDG to see if they have interest in the idea, if you'd be willing to do the integration. If you'd like,...

FWIW, the way I've worked around this lack (and I suspect others have worked around it similarly) is to use ICC to get a rough idea of how EDG behaves....

We also don't support that in C mode either (so I edited the title). Interestingly, we're very explicitly treating them as rvalues and have for a long while: https://github.com/llvm/llvm-project/commit/7decc9e4ea6c56ffce14d25f542f5eda3eeba97a#diff-03e60ab06f82e27e971dc7c0816e51e2a984da855607560e33c94094ac1622c9R2569 @rjmccall...