1.29.0 Release Planning

[rami3l]

@rust-lang/rustup is welcomed to propose outstanding issues that would be urgent enough to address before cutting the upcoming patch release.
Other issues on the current milestone will be postponed to the milestone representing the next release.

Outstanding Issues/PRs

• illumos: rustup 1.28.{0,1,2} are miscompiled #4420
• Rustup 1.28.2 uses a different default profile during automatic installation #4337
• ci(macos): run x64 workflows with Rosetta 2 #4428
• dist: make installation asynchronous #4605
• download: statically bundle relevant trust anchors #4631

Release Process

• dist: bump rustup version to v1.29.0 #4513
• dist(rustup-init/sh): update commit shasum in help string #4637
• docs(dev-guide): mention snapshot updating in release process #4640

Note

It could be a good opportunity to test out rust-lang/promote-release#84, but it would not be a dealbreaker for cutting this release.

[djc]

I'd like to go to 1.29.0 and drop curl and native-tls support. I have one more change I'd like to make before that.

Might be good to also write up a list of (notable) changes that we have on master right now to get a handle on what the release would look like.

[rami3l]

I'd like to go to 1.29.0 and drop curl and native-tls support.

@djc That sounds... risky since there are still remaining problems in #3790, and I would originally expect OpenSSL to last for at least a few more patches.

This doesn't mean we cannot do 1.29.0 here and now; it's just that it requires more attention into justifying/advertising the change and the possible consequences, and I expect that would take longer than a normal patch release... combined with #4420 I'm afraid that some people might never get a working 1.28 release and now they are forced to face 1.29 🙈

[djc]

@djc That sounds... risky since there are still remaining problems in #3790, and I would originally expect OpenSSL to last for at least a few more patches.

It's not obvious to me what problems/remaining risks you are referring to. As I understand it, the change to rustls by default has been pretty smooth and has not led to a lot of issues, even with the explicit warnings in 1.28.2. The one issue I'm aware of is #4325, which is arguably a pretty specific environment problem, and we now have better error handling for that via rustls-platform-verifier 0.6 and I'm working on a fix that includes the roots required for verifying static.rust-lang.org certificates as part of the rustup binary.

[rami3l]

It's not obvious to me what problems/remaining risks you are referring to. As I understand it, the change to rustls by default has been pretty smooth and has not led to a lot of issues, even with the explicit warnings in 1.28.2. The one issue I'm aware of is #4325, which is arguably a pretty specific environment problem, and we now have better error handling for that via rustls-platform-verifier 0.6 and I'm working on a fix that includes the roots required for verifying static.rust-lang.org certificates as part of the rustup binary.

@djc I'm also aware of #4233 and #4305 (IIRC what has landed was not really a fix but a mitigation?), although they are indeed not particularly easy to reproduce.

[rami3l]

Might be good to also write up a list of (notable) changes that we have on master right now to get a handle on what the release would look like.

@djc After another look, I agree that that would be the best way to decide whether we are going with a patch or a minor. I'll do that shortly :D

[rami3l]

@djc I am really sorry for the unexpected delay, but here is the draft PR: #4513.

[rami3l]

@djc @ChrisDenton I think we are pretty close to cutting a new beta; I'd love to see #4471 in it before doing so though. Do you have other remaining concerns?

[djc]

I'm working on a fix that includes the roots required for verifying static.rust-lang.org certificates as part of the rustup binary.

Still want to get this in.

I also worry a bit about the flickering I observed in the MultiProgress when testing #4531. That might be an upstream issue (maybe console-rs/indicatif#708) but might lead to a bunch of issues if we release it.

[djc]

I also worry a bit about the flickering I observed in the MultiProgress when testing #4531. That might be an upstream issue (maybe console-rs/indicatif#708) but might lead to a bunch of issues if we release it.

This is fixed in

• Fix refresh rate for progress bars #4602

[rami3l]

@djc I think we can continue with the release process once all concerns in #4423 (comment) has been addressed. Is your remaining bandwidth okay for the "static.rust-lang.org certificates" PR, or did I miss something here?