Rustc-dev-guide: Missing link in ice-breaker/about.md

[CreepySkeleton]

I think [claim the issue] is supposed to lead somewhere, but it doesn't

[amanjeev]

Can you please elaborate? Are you a member you wants to have the issue assigned yourself?

[mark-i-m]

@amanjeev I think they mean that the [claim this issue] text in https://github.com/rust-lang/rustc-guide/blob/master/src/ice-breaker/about.md is likely supposed to be a link but was not checked.

@Michael-F-Bryan is there any way to make the link checker check these sorts of links?

@spastorino @nikomatsakis What was this supposed to link to?

[CreepySkeleton]

@mark-i-m

I think they mean that the [claim this issue] text in https://github.com/rust-lang/rustc-guide/blob/master/src/ice-breaker/about.md is likely supposed to be a link but was not checked.

Yes, exactly.

@Michael-F-Bryan is there any way to make the link checker check these sorts of links?

Yes, I'm about to send such a PR.

[spastorino]

I guess that should have been https://github.com/rust-lang/triagebot/wiki/Assignment or just saying claim this issue by commenting with @rustbot claim. Unsure what exactly the intention if @nikomatsakis was though.

[CreepySkeleton]

Hm, it looks like there's checking via mdbook-linkcheck but it doesn't check for situations like this.

[ehuss]

is there any way to make the link checker check these sorts of links?

It is possible, but difficult. pulldown-cmark has a broken link callback, but it doesn't work very well (it is global, and does not tell you where the link is). I've experimented with this in mdBook, and found a ton of false positives, because [foo] is valid markdown, and is intentionally allowed. It might have to be a pedantic-style check, due to the high false positive risk.

Fixing the callback is tracked in https://github.com/raphlinus/pulldown-cmark/issues/165 and https://github.com/raphlinus/pulldown-cmark/issues/373. It is also a blocker for updating pulldown-cmark in rustdoc, so it would be nice to get some movement on it. Someone just needs to propose a good API that could work with pulldown-cmark's architecture.

There's a similar issue of broken references are not validated (https://github.com/raphlinus/pulldown-cmark/issues/217).

[CreepySkeleton]

OK, then we might set up a convention not to use any [link] links and replace them with [link][]. What do you think?

[ehuss]

I don't see how a collapsed reference link would change anything? I think it would also be difficult (and tedious) to enforce conventions.

What I would encourage is for people to use "View file" link in GitHub PR reviews and read the rendered output to make sure any markdown rendering issues are caught. GitHub's renderer is usually close enough to mdBook's.

[mark-i-m]

Yes, I think for now, the easiest way to do this would be to enforce it in the review process.

[Michael-F-Bryan]

@mark-i-m I've added initial support for this in mdbook-linkcheck. At the moment pulldown-cmark doesn't provide the byte index of the broken link, so I've implemented a small best-effort hack using string searching.

This sort of incomplete link is valid markdown, so I'm only emitting a warning:

I'll probably create a fail-on-warnings flag for use in CI before making another release.

[mark-i-m]

@Michael-F-Bryan Awesome! Is that on the linkcheck 0.4.0? Is there a setting to enable it?

[mark-i-m]

I'll probably create a fail-on-warnings flag for use in CI before making another release.

That would be awesome.

[spastorino]

Just asked @nikomatsakis and indeed he aimed to link to https://github.com/rust-lang/triagebot/wiki/Assignment/. @CreepySkeleton would you like to provide a PR?.

[CreepySkeleton]

I don't see how a collapsed reference link would change anything? I think it would also be difficult (and tedious) to enforce conventions.

The intention here is to separate [regular text enclosed with square braces] and [collapsed link][] use cases. I believe it would also be easier to automatically check such collapsed links, @Michael-F-Bryan any thoughts?

[spastorino]

After #480 this is fixed. Closing this issue. Can you open a new issue or a PR when you have a solution for the link checking idea?.