Am 04.12.19 um 15:50 schrieb Oleg Endo: > On Wed, 2019-12-04 at 15:41 +0100, Klaus Doldinger wrote: >> >> Am 04.12.19 um 15:33 schrieb Oleg Endo: >>> On Wed, 2019-12-04 at 14:50 +0100, Klaus Doldinger wrote: >>>> Hello, >>>> >>>> does anybody see a possibility to perform a compiletime-check to >>>> verify >>>> that a bit of a bit-field is at an exact position of the >>>> underlying type? >>>> >>>> I would see a possible using the forecoming std::bit_cast, but >>>> what can >>>> one do with c++17? >>>> >>> >>> Perhaps have a look at this approach >>> https://blog.codef00.com/2014/12/06/portable-bitfields-using-c11/ >> >> Well, I don't think this is correct, since reading/writing from a >> non-active member of a union leads to UB. >> So, I think the blog-post ist simply wrong. > > Right, it's UB ... although "most compilers implement it as non- > standard extension". So I wouldn't say the blog post is wrong per se, > it's just failing to mention a tiny detail :) Because it's UB it is geenerally unusable. If you want to use this as a constexpr compile-time-check, the compiler will reject it, since this is UB. So, sadly not possible.
