Glen Fernandes

It defines BOOST_STATIC_ASSERT if it doesn't exist, but it would better if it did not do this: https://github.com/boostorg/json/blob/master/include/boost/json/detail/config.hpp#L93-L95 Instead it could unconditionally use `BOOST_JSON_STATIC_ASSERT` which in non-standalone mode might make...

@bebuch is actually right here, and MSVC is doing the wrong thing by warning. That the specialization for void doesn't exist does not mean that allocator isn't still a useful...

No, `std::allocator` is still usable in C++20 even though the specialization is removed from the standard. It is still a valid proto-allocator today. You can confirm this with other members...

As an example, the following program will compile in C++11 up to C++20: https://gcc.godbolt.org/z/YCbPD3 ``` #include template struct S { S(const A& a = A()) : a_(a) { } typename...

Furthermore, even Microsoft realized that the `allocator` deprecation warning was wrong and it was removed in future versions of MSVC: https://developercommunity.visualstudio.com/content/problem/500588/boost-asio-reports-stdallocator-is-deprecated-in-c.html > Thanks for filing this issue! Yes, the deprecation...

I'm all for having every possible C++ implementation be warning free, even if that means working around implementation defects. That said, as you already know, ASIO is a special case...

@Kojoley yes, I have complained about this before. @bassoy Is uBLAS effectively unmaintained now? Do we need to find a new maintainer? If not, what is preventing a merge to...

There is no active maintainer, but I'm happy to review this (and eventually merge it when it is suitable).

You might want to talk to @breese who has implemented https://github.com/breese/trial.circular

See also the e-mail thread on the Boost mailing list: https://lists.boost.org/Archives/boost/2019/09/246925.php