Release embedded-io 0.7

[chrysn]

Contributes-To: #566

This is a Draft PR until

• I've satisfied all the interdependencies between other crates from the same workspace (as they have both version and path, they have to change their dependency version in the same PR),
• I've updated the date to my best guess of the release, and
• I think we won't wait more for any new changes (embedded-io: API stability and 1.0.0 release #566 has been open for a while, so that shouldn't be too long).

[chrysn]

I think this is ready for review; having been announced on the Tuesday WG meeting is probably as much exposure for any more "let's get that in"s as this gets before doing an actual release.

[Oakchris1955]

In my opinion, #648 should be resolved before releasing v0.7

[chrysn]

Those are additions to a non_exhaustive enum, and thus not breaking, and can happen in any release. I'm happy to continue the discussion there, as to avoid undue delay, but from the perspective of crates that need to converge onto an 1.0 that 0.7 should be semver compatible with, I'd rather have 0.7 yesterday, and a 0.7.1 once #648 is solved.

[ColinFinck]

@chrysn What's the status on this? Any blockers left for a 0.7 release?
I've already updated the README in #706 to reflect some of the changes to embedded-io.

[chrysn]

I don't think there's anything other than someone from the team saying "Yes" missing. (I can rebase this any time, but there's probably no point to that until there's confirmation, given there should be a date to include).

[jamesmunns]

Ping @rust-embedded/hal

[mkroening]

The following PRs would be very useful to us:

• feat(async-io): make async Write::flush a required method #696
• feat(io): implement Read, ReadReady, BufRead, Write, and WriteReady for VecDeque<u8> #697

Just to make sure they don't go unnoticed. :)

[chrysn]

#696 is about the async counterpart that isn't going for a last-call-before 1.0 (and is unaffected by the embedded-io release). Reading the issue, I think what async does now is better (and maybe we should provide .flush() here), but also, I don't care a lot (like, I'd like to change it if it's uncontroversial, but not start a week-long discussion).

#697 is not breaking, so it could go into a 0.7.1 as well – nonetheless, looking into it.

[therealprof]

LGTM

[chrysn]

Thanks. Date is adjusted to today, hope this holds until the button is pressed :-)

[ColinFinck]

Ping @rust-embedded/hal again. Only thing that's missing here is someone changing the release date in the Changelog, approving+merging this PR, and doing the cargo publish

[Dirbaio]

We should do the release at the same time for -io and -io-async, otherwise async users are left without being able to use it. EDIT: and -io-adapters too.

This means merging #696 first.

[chrysn]

I don't think that embedded-io-async is on the same stabilization path as embedded-io yet – fortunately it probably needs to be (for if embedded-io 0.7 becomes 1.0 compatibly, there can still be an embedded-io-async that depends on -io 1.0 for where it's needed).

Doing those releases together does certainly make things easier in terms of dependency management. I'll put this on hold briefly until #696 is through, and then update this PR to be a simultaneous release (including changelog) of all three crates.

[chrysn]

With #696 merged, I've updated to release all 3 embedded-io, all with current changelogs.

While not breaking-updte critical, if #697 is ready to merge (I think it is), that could still be taken in; if not, let's go?

[Dirbaio]

🚀

[Dirbaio]

can you add the changelog for #697?

- implement `Read`, `ReadReady`, `BufRead`, `Write`, and `WriteReady` for `VecDeque<u8>`

[chrysn]

Done. Conveniently, this also updates the MSRV of async to 1.81, which would otherwise not be achievable due to the embedded-io dependency anyway.