maybe there's an outside chance that if nothing has #include in it, trying to use a std::vector will cause a compiler error;

it won't, since our MR::vector class derives from std::vector directly, so <vector> will always be included anyway...

maybe one will be able to use vector rather than having to resort to MR::vector.

You already can, these latest commits are probably overly cautious - it should work with just vector<type> already, and you'll find plenty of examples of that elsewhere in the code. As long as there is no using namespace std; anywhere, there shouldn't be any confusion possible. I might remove these MR:: prefixes if they're going to cause confusion...

@Lestropie, I've merged this commit into #961 - I assume you're OK with that...

You already can, these latest commits are probably overly cautious - it should work with just vector already, and you'll find plenty of examples of that elsewhere in the code. As long as there is no using namespace std; anywhere, there shouldn't be any confusion possible. I might remove these MR:: prefixes if they're going to cause confusion...

Cool, I had assumed that the explicit MR::vector namespace addition was necessary for whatever reason in that context. If it's just left over from bug hunting, probably removing them is the best bet.

The using namespace std; point is a good one though; check_memalign currently won't catch that (ideally it'd flag the presence of using namespace std; then check for corresponding vector<> usage rather than MR::vector<> in that file).

I assume you're OK with that...

👍

The using namespace std; point is a good one though; check_memalign currently won't catch that (ideally it'd flag the presence of using namespace std; then check for corresponding vector<> usage rather than MR::vector<> in that file).

Yes, that check is really just a safeguard. To be bullet-proof would require essentially catching the fully-parsed output of the compiler... Not something we can possibly hope to achieve. We could simply catch any use of using namespace std; and fail outright though, I think it's fair to say that this is not encouraged within the codebase for basically this reason...