alternative implementation for SPI pins

[burrbull]

No description provided.

[therealprof]

Looks good.

[burrbull]

What is better?

let spi = dp.SPI2.spi(
        (spi::NoSck, spi::NoMiso, gpioa.pa10),
...

or

let spi = dp.SPI2.spi(
        (gpio::NoPin, gpio::NoPin, gpioa.pa10),
...

or implement both?

[therealprof]

let spi = dp.SPI2.spi(
        (spi::NoSck, spi::NoMiso, gpioa.pa10),

Is better. It expresses intent and thus prevents misconfiguration.

[burrbull]

Ok.

But misconfiguration is not possible in second variant too.

[therealprof]

But misconfiguration is not possible in second variant too.

Well, misconfiguration in the sense of the user accidentally opting out of the wrong thing. If you read NoPin you know that you're not using a pin but which one. If you only allow NoMiso then everyone can see on a first glance that you're not using a physical MISO pin.

[TheZoq2]

I agree that NoMiso/NoMosi is better than NoPin.

I would almost like to change it to SpiPins{miso, mosi, sck} for even more clarity. I find myself always having to look up the ordering of the pins when using SPI. That is more verbose though, so we should probably allow both.

Perhaps we could add `Impl

[TheZoq2]

Looks good in its current state, we can probably do the thing I mentioned in the previous comment later.

Edit: Oh yea, it's missing a changelog entry, could you add that? @burrbull

[burrbull]

In last commit I've implemented Pins trait for different order.
Now you can pass (sck, miso, mosi) or (mosi, miso, sck) or (mosi, NoMiso, NoSck), etc.

[TheZoq2]

Looks good, thanks!