Miguel Ojeda

I hit this when I wanted to check whether RISC-V compressed instructions were being emitted or not.

@Mingun Agreed, this would be nice to have in `rustdoc` and not just docs.rs (potentially also for hidden ones, not just private). In general, switches like that (or collapsed-by-default elements)...

> Well the compiler could scan at least all objects it knows go into the linking, even if they're written in non-rust. It would not cover all cases, so you...

> Wouldn't it make more sense to do it via a configuration file Possibly (see "Future possibilities"). Since there are already other similar command-line flags (e.g. the playground URL), this...

> 1. One discussion for fixing `html_logo_url` and `html_favicon_url` without changing the nature of the API. (Maybe do what was done with `Error::description` and `Error::cause` and keep them around but...

> However, you _are_ mashing together two issues: No, I am not. I have a need for this use case, which is currently not supported, and I wrote an RFC...

> But, as I said, they're two separate sub-issues that should be solved independently, so that you don't force people who want the fix to accept what they see as...

> Given that relative paths are presumably nearly-always broken in the existing URL attributes, does it seem viable to change the meaning of a relative path in those attributes to...

> OK, Fair. I admit I was getting a bit un-civil in my frustration and, in hindsight, I should have, at minimum, addressed that differently. No worries, all good!

Recent changes to the HTML output (https://github.com/rust-lang/rust/pull/115948) broke our workaround for this. It has made it harder to workaround too, since now the Rust logo is not there for non-Rust...