Feature/polling

[TremblingMen]

Added polling support to improve performance.

[coveralls]

Coverage decreased (-4.3%) to 93.691% when pulling c479fe8 on embedded-rust-iml:feature/Polling into 23a2a93 on eldruin:master.

[eldruin]

Interesting. Thank you for this proposal.
A couple of thoughts:

• It could be that in some applications this repeated polling is undesirable, since it blocks the I2C bus. I think a straight-forward solution would be to add a disable_polling method. Could you do that?
• Looking at the Storage::wait method, it seems we could simplify it by implementing the loop directly instead of using the repeat_timeout! and nb::block! with something like:

// I just wrote this down, please excuse small errors.
loop {
    match self.count_down.wait() {
        Err(nb::Error::Other(e)) => { break Err(e) }
        Err(nb::Error::WouldBlock) => {
              if self.eeprom.polling {
                     match self.eeprom.read_byte(0) {
                            Ok(_) => { break Ok(()) } // done
                            Err(Error::I2C(_)) => { } // not ready, repeat
                            Err(e) => { break  Err(e) } // TODO this error probably needs some mapping to the outer error type
                     }
                 }
            }
        Ok(_) => break Ok(()),
    }
}

TODO:

• Remove embedded-timeout-macros if solution above is viable
• Changelog
• Tests
• Docs, README update

[eldruin]

Would you be able to add some tests? I would take care of the rest.

[eldruin]

Thank you for adding the tests! only a small thing left

[eldruin]

Any reason for this change?

[DrTobe]

I guess @eldruin is right and the initial countdown should be zero. The rationale is already given in the comment above.

[eldruin]

Suggested change

embedded-timeout-macros = "0.3"

[guillaume-michel]

Hi,
Can we have a status on this PR, please?
I am will to help for this to be merged.

[eldruin]

I will merge this and do the last fixes myself on it over the next few days.
I will comment here once that is done.