Hi @krakow10, thanks for this pull request. I'm happy to add more digit manipulation, what do you specifically want? I will merge this PR, but first could you add documentation for set_bit by creating a set_bit macro in doc/mod.rs, use the bit macro as an example, then add #[doc = doc::set_bit!(U 256)] to the definition instead of #[doc = doc::bit!(U 256)]. Thanks.

I wrote a documentation like you requested. Can you think of a better way to set the bit? For example, do you think $Digit::from(value) is problematic and should be done differently?

I'm happy to add more digit manipulation, what do you specifically want?

I thought I would come up with more digit manipulation needs, but I didn't run into anything else that inspired a useful method. I will propose more changes later if I think of something useful, thanks.

P.S. I couldn't test this because of the unstable features

Sorry for the very slow response time @krakow10, I've just come back from a holiday. Thanks for adding the documentation, I will merge this now. Sorry about the compilation error when trying to test - annoyingly, the nightly compiler seems to have recently dropped const trait support, which is what was causing that issue. I'm going to make those traits non-const for now. I'll publish 0.12.0 with both these changes either today or tomorrow, hopefully.

I think there may be a slight error in your code actually, as applying & (1 << shift) will zero out other bits - I've modified it slightly so that it's fully correct, by doing

if value {
    *digit |= (1 << shift);
} else {
    *digit &= !(1 << shift);
}

I think it was supposed to be a reverse mask, but I forgot !. Like this:

*digit = *digit & !(1 << shift) | ($Digit::from(value) << shift)

Saves a branch that way.

Ah yeah I see, I think a branch is unavoidable though as $Digit::from(value) uses one.