I plan to get these done later today also, and they'll follow the tests as a rough example (but obviously cleaner, easier to understand and commented so users know what's going on).
I plan to add warnings about using Cmov directly, as it would act more of an optimisation barrier if a condition is set to x == y.
I may also add warnings about branching in general, and using the ^ operator where the functions provided by CmovEq aren't adequate (which will be extremely unlikely but still probably worth noting). I'll also briefly explain why using == is bad, but I'm no branching expert so that may require some tweaking.
I'll try to adhere to the RustCrypto example format in order to reduce PR review time, and for consistency.
I think I'll try and get #920 fixed up first though.
I plan to get these done later today also, and they'll follow the tests as a rough example (but obviously cleaner, easier to understand and commented so users know what's going on).
I plan to add warnings about using
Cmovdirectly, as it would act more of an optimisation barrier if a condition is set tox == y.I may also add warnings about branching in general, and using the
^operator where the functions provided byCmovEqaren't adequate (which will be extremely unlikely but still probably worth noting). I'll also briefly explain why using==is bad, but I'm no branching expert so that may require some tweaking.I'll try to adhere to the RustCrypto example format in order to reduce PR review time, and for consistency.
I think I'll try and get #920 fixed up first though.