More efficient is_zero check

[ivanrg99]

rustc is not smart enough to only use or for the generated assembly and adds extra instructions for jumps and compares, to facilitate the short-circuiting check. With a simple bit comparison we achieve the same effect and the compiler
does not generate any branching instructions:

https://godbolt.org/z/6nrhvsxs7

[Tony-Samuels]

Looks like we'll need a clean-up MR first to fix some issues not introduced by this change.

[paupino]

Nice find!

I think this should target the v1 branch as opposed to master. It seems like a stable fix, which can also be applied to the currently unstable master. (I.e, merge/release against v1, then v1 propagates up to master).

[ivanrg99]

I think this should target the v1 branch as opposed to master. It seems like a stable fix, which can also be applied to the currently unstable master. (I.e, merge/release against v1, then v1 propagates up to master).

Should I create a new PR then with v1 as the base that targets v1 and close this one?

[Tony-Samuels]

Should I create a new PR then with v1 as the base that targets v1 and close this one?

I think we'd like this on both. If you leave this one open and create a new one for v1, then I'll make sure this one is merged when we sort out master.

[paupino]

Oh, yeah - it looks like you may need to rebase your change onto v1 first.

Sorry, I've started using master for unstable breaking change cleanups (aka v2). I need to make sure it's clear for any new contributors what to do here - maybe a pull request template could help?

[ivanrg99]

Created #719 using v1 and targeting v1