Use thiserror rather than eyre in mijia crate
#83
Conversation
qwandor
force-pushed
the
eyre-ire
branch
3 times, most recently
from
8bc19e2 to
5f3d016
Compare
qwandor
force-pushed
the
mijia-properties
branch
from
d7fed4f to
b7a4d7e
Compare
Error is just a re-export of Report.
There was a problem hiding this comment.
See what this does to the backtraces and then ship it? Would be nice to know whether converting from box dyn Error to thiserror to eyre throws stack frames away. It may be that the backtrace crate registers itself in a magical way that makes everything get preserved, but I'm also okay with losing a few frames of dbus-rs as long as we can still report the place where our thiserror is being constructed.
| if value.len() != 6 { | ||
| bail!("Wrong length {} for comfort level", value.len()); | ||
| } | ||
| pub(crate) fn decode(value: &[u8]) -> Result<ComfortLevel, DecodeError> { |
There was a problem hiding this comment.
(probably not something to touch in this PR): Now that I this is feeling quite a lot like the signature of impl TryFrom<&[u8]> for ComfortLevel and decode() feels like impl TryFrom<ComfortLevel> for [u8, 6]
It may be that converting these all to TryInto will destroy the ergonomics at the call site though. We would certainly need provide a newtype for SystemTime if we went in this direction.
There was a problem hiding this comment.
That would mean that the decoding logic is part of the public API of the crate, which I'm not sure we want. I was considering adding some private traits, but it doesn't really give any advantage, and prevents some of the slight differences in signatures we currently have (e.g. encoding TemperatureUnit can't fail, so doesn't return a Result).
|
Backtraces go up to the point that |
It seems more appropriate for a library.