-
Notifications
You must be signed in to change notification settings - Fork 210
New issue
Have a question about this project? # for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “#”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? # to your account
Embedded error #218
Embedded error #218
Conversation
Signed-off-by: Daniel Egger <daniel@eggers-club.de>
Signed-off-by: Daniel Egger <daniel@eggers-club.de>
Signed-off-by: Daniel Egger <daniel@eggers-club.de>
Signed-off-by: Daniel Egger <daniel@eggers-club.de>
Signed-off-by: Daniel Egger <daniel@eggers-club.de>
Signed-off-by: Daniel Egger <daniel@eggers-club.de>
4c82b9e
to
d70cde9
Compare
@@ -294,7 +294,7 @@ macro_rules! gpio { | |||
} | |||
|
|||
impl<MODE> OutputPin for $PXx<Output<MODE>> { | |||
type Error = Infallible; | |||
type Error = GpioError; |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
For the other errors, I see the advantages, but replacing infallible by some non empty error, that's not great. I personally into_ok (a personal implementation for stable) on these errors, and it would not be possible after this change.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
If we don't make it uniform then we can't use this universally with drivers and applications, this is just a preview of what's to come in embedded-hal
. There're plenty of cases like on a Raspberry Pi or behind GPIO expanders where GPIO is not infallible.
What is into_ok
? And how is it different from .ok()
? I don't think applications will change a lot but there might be a slight size overhead (not sure, really, Infallible
is also not quite ideal). If you have any ideas for improvement, I'd be all ears.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
In which case would you need to know the error? Can you point on real example? I suppose some intermediate driver between embedded_hal and another driver where you have to manage some kind of error?
https://doc.rust-lang.org/core/result/enum.Result.html#method.into_ok and my "waiting to stabilise implementation" https://github.com/TeXitoi/keyseebee/blob/master/firmware/src/main.rs#L32-L42
That's an infallible .ok().unwrap()
. As I know I don't have to manage error, I can remove the complexity of managing them for nothing. And it won't be possible anymore with this modification.
Maybe we can have Error: Into<GpioError>
and Infallible
can implement this conversion. But it will complexify the interface.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
In which case would you need to know the error? Can you point on real example? I suppose some intermediate driver between embedded_hal and another driver where you have to manage some kind of error?
Yes, for example. No, I don't have an example yet but support for fallible GPIOs has been requested a few times and was incorporated into embedded-hal
(aka the digital::v2
debacle).
Maybe we can have Error: Into and Infallible can implement this conversion. But it will complexify the interface.
When we change the Error types in embedded-hal
the bounds will be Into<GpioError> + Clone + Debug
. I don't think we can implement Into<GpioError>
for Infallible
but if you have any idea how to solve that your input would certainly be appreciated over at rust-embedded/embedded-hal#229.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Implementing Into for Infallible is easy: you just have to implement From, that you should implement if possible instead of Into anyway: https://play.rust-lang.org/?version=stable&mode=debug&edition=2018&gist=86a6dabe6ccc820f756c03718d5e4cc4
Will try to look at the linked issue.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
I'll do a PR for implementing From for *Error
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
I'd be careful with doing that for the other Error. Many (most?) of the other peripherals are impossible to implement correctly in an Infallible
way and I'd be wary to add a loophole to avoid a correct implementation out of the gate; we can always add it later if it makes sense.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
If I'm not mistaken, only embedded-error can implement From<Infallible>
for its types, and at least mathematically/logically, this conversion should exist (and there's only one implementation "up to canonical isomorphism"...). So I think the implementation should be added, with warnings about (peripheral) implementation correctness, no?
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
@nickray Yes, adding it to embedded-error
and using Infallible
at least for GPIO is what we were talking about.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
I've made a different proposition: rust-embedded/embedded-hal#229 (comment)
I'll wait a bit the feedbacks before doing a PR to embedded-error, if you don't mind.
No description provided.