Daniel Silverstone

Switching away from toml to bincode basically made the manifest decode drop to be in the noise. sub 1ms. Not loading the manifest at all would mean we'd need to...

This kind of simplification sounds sensible. If you're interested in non-core contributions toward this then I'm very interested in helping.

So I think `rustup toolchain link` is important to keep; certainly I wouldn't advocate removing it. I do wonder if there might be something even more nice we could do...

> I guess "(paleo-)mutt" refers to the original mutt? Yes, when I wrote up this feature request I was switching from mutt(-patched) in Debian to Neomutt in nixos and encountered...

Sounds like the general overheads on that platform are higher than we expect. If you reduce RUSTUP_UNPACK_RAM does that help? You can set it to something tiny (e.g. 10 bytes)...

Thanks Jon, I've marked this as draft for now because until we're confident the decisions are right, I don't want to risk merging this. I'll try and get to code...

Copy+Delete would be exceedingly slow because the rename stuff is used in our transactional filesystem accessing code. If we had to open+open+{read,write,loop}+close+close rather than rename then our toolchain update process...

We are working on improving the manual pages experience with `rustup` we cannot do as you suggest because that'd lock a singular toolchain into the MANPATH - a toolchain which...

@lzutao Do you fancy having a go at adding a `man_mode.rs` to rustup, i.e. detecting that rustup was invoked as `man` and acting as though `rustup man` was run? You'd...

man *cannot* know where to look because the requisite information is held by rustup. i.e. man cannot examine `rust-toolchain` or the rustup default setting, or directory overrides, etc. in order...