@FiloSottile just a quick reminder in case you weren't aware of this PR. Any chance we can add this? Thank you!

I have no objection to adding back pre-built binaries for darwin/amd64, I didn't even intentionally remove them, I think I patterned the update on top of a different workflow that didn't have them. (Also, I honestly thought most people used Homebrew on macOS!)

However, this PR will only add them to the next release. Adding them to the latest release, while still producing GitHub Actions attestations, is not trivial. If anyone wants to take a stab at that, be my guest. I'm happy to do the Sigsum part manually.

I don't want to tag a v1.3.2 just for this, because that kicks off a bunch of work for a lot of downstream distributions.

Also, I honestly thought most people used Homebrew on macOS

Some people use mise as it's more easier to have a specific version per repository, without dealing with a single version system wide.

I don't want to tag a v1.3.2 just for this, because that kicks off a bunch of work for a lot of downstream distributions.

Understandable and that's fine with me! I'll pin it to 1.3.0 for now and will update my project mise configuration on the next release (if this is merged).

mise ERROR Failed to install aqua:FiloSottile/age@latest: unsupported env: darwin/amd64 (supported: ["linux", "darwin/arm64", "windows"])
mise ERROR Run with --verbose or MISE_VERBOSE=1 for more information

I use mise, and the repos my team work also use mise since we are centralizing these tools across the repos.

I gotta say that I am not enthusiastic about a package manager that doesn't do the basic job of a package manager—packaging and compiling software—and instead drops all the responsibility of building binaries on all of their supported architecture onto the upstream project. Even worse, it doesn't even try to keep track of what version has binaries for what architecture, keeping darwin/amd64 working by installing v1.2.0? Like, what the hell.

age still compiles on and works on darwin/amd64. I believe Debian packages it for even weirder architectures, but they don't make it my problem.

Anyway, I wrote a Workflow Dispatch one-off GitHub Action to backfill the darwin/amd64 binaries without breaking build provenance attestations. https://github.com/FiloSottile/age/actions/runs/21220026585 They work under Rosetta2. I will have to do the Sigsum proofs when I am back at home.

Someone please send a PR to mise.

Thank you for the support, really appreciate.

For what is worth, a lot people discover your package because of mise I do think the maintain intent to give you more work but honestly promotes it as a professional solution to use.

I did a one-time donation, hopefully compensates for the frustration, that is the least I can do at the moment.

Thank you so much.