David Detweiler

I agree with @elshize for clearly-marked "unsafe" members for `span`. Though we could all hope that compilers are smart enough to optimize-out inlined checks that are proveably unnecessary, the unfortunate...

@johnmcfarlane As it turns out, my comment on MSVC was a bit out-of-date, and it's now smarter about it. `gsl::at` gets reliably optimized out for cases when the index is...

I'm also not the first person to have these concerns: https://github.com/isocpp/CppCoreGuidelines/issues/1561#issuecomment-616486172. There are four scenarios of "intended behaviors" I can think of: 1. `Expects()` is a pure contract annotation (and...

R. 11 as it is leaves a semantic hole. Ideally, we'd just use `unique_ptr(new (buffer) type{});`, but `new` would still be flagged under R.11. There is no alternative syntax for...

Ok, I get that. `std::allocator_traits::construct()` just didn't seem all that clean considering it ties placement syntax to the standard allocator model, and it seems odd to have to call the...

Well, if we mean `std::allocator_traits::construct()`, suppose I wanted to make a logging implementation of it for a custom allocator. If I still wanted to use the default for the actual...