Nemo157

Tailscale already has the client-side feature for this, someone needs to investigate exactly how it is represented by the server and add it to the details provided by Headscale, there...

The most aggressive attack that @pietroalbini mentioned was: * We build some crate `foo` under control of the attacker * This has a build script that generates a `bar_macro.rlib` and...

Oh, I forgot one indirection in the first step, the crate `foo` depends on another crate `foo2` under control of the attacker. The `foo2` build script will be run during...

Without caching: if `foo2` writes `libbar_macro.rlib` into the target directory, that will be wiped before we build the docs for `bar`, so there's no way for one build-tree to affect...

We have a hard-limit of max 1 day of caching because of updating the compiler. There are probably plenty of cases where a few related crates are published throughout the...

We can get the actual publisher of each release from the crates.io API (even for crates with multiple owners, if they're publishing a workspace at once I'd assume a single...

Maybe with two strategies, one with `--cwd .`, and the other without, so it will fallback to using `atuin`'s global search instead of then going to zsh's history.

Ah yeah, I should post what I did, using multiple strategies for the two cases: ```zsh # Add --cwd flag to have auto-workspace-detection active _zsh_autosuggest_strategy_atuin_auto() { suggestion=$(atuin search --cwd ....

> I'm happy for us to open in normal mode when invoked from normal, and vice-versa for insert. This doesn't match my intuition for how the modes should start, I...

One related issue I just found, the `show_preview` option appears to be using "bright black" as the foreground text which in my color scheme is the same as the default...