I think it’d be better to just have reverse as it’s own wrapper.
Otherwise things get real convoluted real fast
True, but because these wrappers are run-time objects, each wrapper costs a little cpu and memory, so having one wrapper instead of two is beneficial.
It was my understanding these wrappers are in play at a fairly high level, such that especially if they’re built to be inlined or templated (probably a similar effect in performance for those calculations…) they’d only be queried for computation a relatively few number of times, and ofc an inline would avoid the jump and all the overhead that comes with that.
Unfortunately that’s not the case.
The blade wrappers usually have to get involved with the set() function which gets called ~10k times per second, and the way they are written, inlining doesn’t work. Global optimization may work to reduce the calls, but there are no guarantees.