On Fri, Dec 6, 2024 at 7:57 AM Peter Dimov pdimov@gmail.com wrote:
They don't need a rationale. I'm asking their opinions on whether a C++14 requirement would be fine for them. Not asking them to speculate whether a C++14 requirement wouldn't be fine for some unspecified other people.
When I read "Do people care about C++11 support in hash2" I interpret "people" as "unspecified other people". And when I read "would it be acceptable to impose a minimum requirement of C++14" I understand this to be asking if it is acceptable to impose the C++14 requirement on everyone. My apologies for the misunderstanding.
Thanks
To my knowledge now that RHEL 7 has gone EOL there is no supported OS without a C++14 compatible compiler by default. Naked constexpr will break MSVC 14.0, but that's not a big loss. Anecdotally Math and Multiprecision moved to C++14 a few years ago with 0 complaints. I know for fact those run on pretty special hardware (e.g. Oak Ridge's supercomputers) with their own specialized compilers and the situation is fine. Matt