Andrey Semashev

> But since it appears Andrey can't be arsed to press Enter twice Ouch! Yes, I can add newlines, and I probably will if I have to before the next...

So, what's it going to be? Is the scanner going to employ the target-based approach? Because if your plan is to hardcode Filesystem I'd rather add newlines (with a promise...

> So, what's it going to be? Is the scanner going to employ the target-based approach? Because if your plan is to hardcode Filesystem I'd rather add newlines (with a...

If swapping `const` values is now allowed, this would probably mean we need to drop the `is_const` check [here](https://github.com/boostorg/core/blob/5f6fe65eb227375888f7305c73767600ab7f1cf2/include/boost/core/invoke_swap.hpp#L81). But my understanding is that the support for swapping `const`s is...

> We should make it work, either by transcoding to UTF-8, or by using the `codecvt` facet of the output stream. I have something like this in Boost.Log: https://github.com/boostorg/log/blob/8a880720be5a990fdd0f864244bfff1a39a69d83/include/boost/log/utility/formatting_ostream.hpp#L398-L416 https://github.com/boostorg/log/blob/8a880720be5a990fdd0f864244bfff1a39a69d83/src/code_conversion.cpp#L68-L139...

> We can't add it to `io` because Core is not allowed to depend on Io. I don't like that we're adding stuff to Core just because we don't want...

In my case, I will need a checkpoint-like approach anyway since the checker is not called directly by the test code. But a set of macros that accept a custom...

The code in Boost.Context is specific to x86 (and not entirely correct, BTW - `prefetch_range` should align prefetches to cache line boundaries, otherwise it might not prefetch the last cache...

I suppose, the main reason is that `reference_wrapper` is not intended to be a proxy for `T`.

There's also a similar overload for `char*`.