Our utils impl and std do the same thing, so I wonder whether it's necessary to do this. I would've hoped that we can just get the std impl and then remove the polyfill trait some way down the road when the MSRV allows for it.

So you want to leave the warnings until these functions are stabilized?

We could just silence the warnings. Or do you think it's a bad idea?

It feels a bit unnecessary to me to change to fully qualified calls and then switch back to the exact previous code.

I changed it as you wished

Thanks! I folded this into #6775 to make sure we don't accumulate new warnings before bumping.