feat(install): add skip_std flag to allow skipping rust-std installation for custom targets

[cachebag]

Closes #4442
Revision of #4585

[djc]

So I'm not sure yet how/where the ToolchainSection::skip_std field makes it into the DistOptions?

[cachebag]

@djc

So I'm not sure yet how/where the ToolchainSection::skip_std field makes it into the DistOptions?

Yeah, sorry about that. I think it should be fine now, the flag flows through the chain where it should be applied as options.skip_std = skip_std before being used to skip the component in try_update_from_dist_().

[djc]

(You'll need to rebase and squash before this can be merged.)

[rami3l]

Thanks a lot for your contribution and sorry for having taken so long to respond, but this seems to have indicated some bigger changes to be done beforehand, made pretty visible by the extra large number of parameters... I'll find some time to have another look.

[rami3l]

As I understand it, the original issue #4442 actually describes an inconsistency between rustup toolchain install and rustup target add. Although an extra flag is definitely reasonable to be added, I still think it would be better if the user basically doesn't need to do anything and the installation of rust-std is automatically skipped, because that brings the overall experience of the former closer to the latter which is what we would like to see.

I'm saying this because rustup can trigger toolchain installations in many places, and adding a flag regarding installation is like playing a round of whack-a-mole here.

[cachebag]

As I understand it, the original issue #4442 actually describes an inconsistency between rustup toolchain install and rustup target add. Although an extra flag is definitely reasonable to be added, I still think it would be better if the user basically doesn't need to do anything and the installation of rust-std is automatically skipped, because that brings the overall experience of the former closer to the latter which is what we would like to see.

I'm saying this because rustup can trigger toolchain installations in many places, and adding a flag regarding installation is like playing a round of whack-a-mole here.

I did feel awkward writing this patch. So I presume this change actually propagates a bit deeper throughout the codebase than what my stopgap currently does?

Should I just iterate on this PR or would you like me to open a new one?

[rami3l]

@cachebag Given the analysis in #4442 (comment) I'm not 100% sure of what to do next, but this PR wouldn't make it there anyway.

As such I'm closing it for now, many thanks all the same!