Austin Seipp

Oh, I forgot that you *also* need to set the `DENO_KV_ACCESS_TOKEN` environment variable in the context of `deno` for this to work, which is the access token for the endpoint....

Good catch. I think this one can be closed pretty reasonably since the large file error is no longer fatal (and `--ignore-working-copy` didn't really help that much anyway.)

https://github.com/jj-vcs/jj/blob/1194b94e51155f7955be16dfbc8c972f0672be2c/.github/workflows/release.yml#L70 is the culprit if I had to guess. I'll look into testing a fix soon-ish.

> have it built-in knowing that such a thing will be a consequence That's not true, at least for me. #7452 didn't mention this at all, only that it was...

> I mean if literally _four_ out of _five_ pushes in the Discord fail/fizzle out because people are concerned about not wide enough OSC-8 support in terminals, I don't see...

Some other notes about the current situation too: - buck2 is tied to a specific version of prelude a *little* more than it should be right now, but there's a...

Two questions: - If we created our own dynamically linked glibc builds, and published them in a release, would `cargo binstall` use those instead of the upstream `zig` version that's...

This should be working now, and the binaries seem to work. The ARM build is 25% faster or so than x86 too 🥲 (test time is the same, though)

Hrm, that's an interesting sounding feature, but I'm not sure I understand it. Is the idea you just provide both hashes and try to race both of them against each...

Just a nit I realized now and should have said earlier: if we did do this, we'd probably want to add a `buck2 download` command that could download some file...