ring icon indicating copy to clipboard operation
ring copied to clipboard

New Release

Open georglauterbach opened this issue 7 months ago • 12 comments

It'd be super-helpful to see a new minor version if possible to include the latest changes. In particular, I'd be delighted to see the changes from the associated PR of #2454.

georglauterbach avatar May 06 '25 16:05 georglauterbach

This is blocked on me doing the pre-release testing. Unfortunately, some of the infrastructure I have for this, especially aarch64-windows-msvc machines, are not available to me currently. I filed #2526 to track adding it to GitHub Actions so we don't get blocked on that based on my location.

briansmith avatar May 07 '25 20:05 briansmith

May I assist you in that? What exactly needs to be done? I have a background with GH Actions.

georglauterbach avatar May 08 '25 09:05 georglauterbach

I started the work for aarch64-pc-windows-msvc in https://github.com/briansmith/ring/pull/2529.

Unfortunately, it seems there are multiple compatibility-breaking changes in Rust 1.87 and Rust Nightly that are blocking this.

I addressed one of them in https://github.com/briansmith/ring/pull/2530.

Another one seems to be https://github.com/rust-lang/rust/issues/136088. I think this is caused by the recently-added doctest-cross-compile feature (previously, doctests were ignored when --target was given to cargo test but now they are not). Probably we need a PR to skip doctests for the test-features job for the aarch64-linux-musl target.

Another one, again probably attributable to the doctest-xcompile feature, is that many of the coverage jobs fail. Probably we should just disable all the doctests in the coverage jobs since we don't want to be relying on doctests for code coverage anyway.

briansmith avatar May 20 '25 20:05 briansmith

All the backward-incompatible changes have been dealt with and the aarch64-pc-windows-msvc testing has been added to CI. I hope to do the release soon.

briansmith avatar May 21 '25 17:05 briansmith

Awesome!

georglauterbach avatar May 21 '25 17:05 georglauterbach

Now blocked on Issue #2536 where a test is failing on aarch64-pc-windows-msvc with Rust 1.66 consistently in CI.

briansmith avatar May 21 '25 22:05 briansmith

With #2536 being removed from the 0.17.15 milestone, #2435 and #2444 seem to be the only blockers for the release left. Is this correct?

georglauterbach avatar Jun 04 '25 14:06 georglauterbach

Any updates on this?

georglauterbach avatar Jun 24 '25 13:06 georglauterbach

I've started the process of merging the latest BoringSSL changes into ring. As things have mostly diverged, the merge mostly involves doing analogous things with different code than BoringSSL. So, for example, we have PR #2593 and PR #2594 with quite non-obvious changes that could use extra eyes and (especially in the case of #2593) benchmarking.

After I do the BoringSSL merge then I'll do the re-benchmarking and then do the release.

briansmith avatar Jul 02 '25 23:07 briansmith

The mentioned PRs are merged, so what's the status on this? Can we put the remaining PRs into 17.16? We are really waiting on the improvements to the build process..

georglauterbach avatar Jul 29 '25 14:07 georglauterbach

Is there anything the community can do to help get a release out? I've switched to pulling in ring from git for now, but it takes a solid 30-70 seconds for cargo to clone the repo, depending on network conditions 😂

mcronce avatar Sep 10 '25 21:09 mcronce

Can the community do anything to help speed up the release of 0.17.15? Can the associated PRs be postponed into 0.17.16? I've been seeing

       Dirty ring v0.17.14: dependency info changed
   Compiling ring v0.17.14

for months now, and it really slows down compilations.

georglauterbach avatar Oct 01 '25 13:10 georglauterbach

Instead of merging manually, wouldn't it be easier to include BoringSSL as a git submodule?

reneleonhardt avatar Dec 12 '25 08:12 reneleonhardt