On 1/13/2018 6:53 PM, Peter Dimov via Boost wrote:
Edward Diener wrote:
But when the answer argues that they are conformant because in the particular case a warning is valid for establishing conformance, then I cannot understand it. We might as well say that any compiler which issues warnings whenever a compiler error should occur, simply because it can continue compiling, is a valid C or C++ conformant compiler.
It is. That's what the standard says. Whether a diagnostic starts with the word "error" or the word "warning" is irrelevant for conformance.
So basically if the code continues compiling, even if it produces an incorrect result, the C++ standard says that the compiler is conformant because it issued a warning to the end-user instead of an error. Thanks for notifying me about this. I really did not know that this is how the standard is written. Needless to say if I were on the C++ standard committee I would have voted against this, but that is irrelevant to the way things are. At least I have learned something.