Remove rarely true check in char_traits::move

[AreaZR]

While it is faster to just leave when the source and destination are the same, most programmers would not make that mistake, and if they did, it would be a rarity and a reflection of something going wrong in their code in the first place.

There is no reason to do this check if speed and code size are of utmost importance.

[AdamBucior]

There is no reason to do this check if speed and code size are of utmost importance.

This code is only executed at compile time, but I agree that it's unnecessary.

[AreaZR]

On second thought, since this is a compile time constant, I wonder if this is a good idea.

[CaseyCarter]

On second thought, since this is a compile time constant, I wonder if this is a good idea.

My intuition is that "self-assignment" at compile time is as rare as "self-assignment" at runtime and that the guard won't trigger often enough to offset the cost of evaluation. I don't think any of our supported compilers currently implements a branch predictor for constexpr evaluation, so I suspect this is compile-time guard has a larger negative impact than would a similar runtime guard. I would nonetheless be surprised to see a benchmark which demonstrated a statistically significant difference between having and not having the guard, but I'm happy to have fewer lines of code to maintain.

[AreaZR]

True. Maybe when our compiler is better.

[AreaZR]

Alright I squashed everything and rebased, so this should be good to go!

[CaseyCarter]

Alright I squashed everything and rebased, so this should be good to go!

It's not an issue given how small this change is, but for future reference please avoid force-pushing the branch since it interferes with incremental code review. (Merging main into your branch is a good alternative to rebasing if you need to pull new stuff into your branch, and we'll "Squash and merge" anyway when we merge.)

[AreaZR]

Alright, thank you!

[StephanTLavavej]

Thanks for removing this unnecessary code! 😺 🧹 🎉