[clang] [clang] check deduction consistency when partial ordering function templates (PR #100692)

[Alexander Kornienko via cfe-commits]

alexfh wrote:

Thanks for the clarification! The fact that there's code around that isn't quite following the standard, is not news, but the regular way to approach tightening the compiler is to  provide an opt-out mechanism for some time (one release cycle?) in the shape of a flag or otherwise. Do you think it could be a good option here? Alternatively, do you see a standard-compliant way to rewrite this code?

And did you have a chance to look at the third test case? It would be nice to know if Clang is acting according to the standard there.

https://github.com/llvm/llvm-project/pull/100692