defmt impls for time

[burrbull]

It makes sense to have implementations for time crate primitives.

[jonathanpallant]

Maybe, but there's an almost unlimited number of third-party types we could special case.

Also, most embedded projects seem to use https://docs.rs/chrono/0.4.26/chrono/ and not time.

[Urhengulas]

I agree that there is a potentially unbound number. But I also think that either us or them should implement it, so users can use these types easily with defmt. And since both time and chrono are bigger than defmt I guess it makes sense to do it on our side.

Long story short, I'd be okay with accepting a PR to add Format implements for the time crate, hidden behind a feature gate. Same for chrono.

[Urhengulas]

There is chronotope/chrono#1632 to implement defmt::Format for the chrono structs.

[burrbull]

I still think this should be done on defmt side. chrono and time are common used crates implementing basic time types.
Where defmt is "embedded only" logging framework.

[jonathanpallant]

Yeah, but if chrono or time make breaking change bumps, we have to implement the traits for all common versions, which gets messy.

We're just about to commit to never doing a major version bump, so it's cleaner for them to implement our trait. And DJC said he would once we hit 1.0.

[Urhengulas]

Also if we implement it on the chrono or time side we have access to private fields. I don't know if this is necessary in this case but it makes sense in the general case.