Broken links in String documentation from Deref impl

[alexcrichton]

There are a bunch of links in the String documentation page which are broken. The reason for this is that most links are drawn from the documentation of the str type at /std/primitive.str.html but the String page is located at /std/string/struct.String.html. This mismatch in depth of paths causes a number of broken links, for example on the String::split method.

Not sure of the best way to fix this, but needs a tracking issue.

[durka]

It would be most awesome if there were some way to link to an item in Rustdoc, not just putting in a link URL, but that's perhaps out of scope for now. Maybe just some basic variable substitution, so you could write [foo]($htmlroot/bar/baz/foo.html)?

[alexcrichton]

Yeah I suspect some basic substitution could work especially because we control the markdown rendering. I'd be a little wary to add API surface area to rustdoc, but it seems minorly well motivated?

[durka]

It seems pretty much required to put in some facility for linking without relative URLs, because Deref copies and pastes things around.

[steveklabnik]

It would be nice if we could follow the work CommonMark is doing here, so we're not totally just inventing a new thing.

http://talk.commonmark.org/t/generic-directives-plugins-syntax/444 seems to be at least one relevant discussion.

[mitaa]

That seems like a nice long-term solution, that'd have to be implemented in hoedown however.

There is an old issue regarding this on their repo: hoedown/hoedown#99

[steveklabnik]

yes, I would like to move from hoedown to pulldown-cmark, personally.

[frewsxcv]

Opened an issue for moving to pulldown-cmark: #38400

[MaloJaffre]

This issue seems to block the testing of compiler docs: ref #46278

[ollie27]

This needs to be reopened. There are still broken links (e.g. #55666) and this issue can't be closed until the whitelist in linkchecker is removed:

rust/src/tools/linkchecker/main.rs

Lines 134 to 148 in 8315b11

	// FIXME(#32129)
	if file.ends_with("std/string/struct.String.html") ||
	file.ends_with("interpret/struct.ImmTy.html") ||
	file.ends_with("symbol/struct.InternedString.html") ||
	file.ends_with("ast/struct.ThinVec.html") ||
	file.ends_with("util/struct.ThinVec.html") ||
	file.ends_with("util/struct.RcSlice.html") ||
	file.ends_with("layout/struct.TyLayout.html") ||
	file.ends_with("humantime/struct.Timestamp.html") ||
	file.ends_with("log/index.html") ||
	file.ends_with("ty/struct.Slice.html") ||
	file.ends_with("ty/enum.Attributes.html") ||
	file.ends_with("ty/struct.SymbolName.html") {
	return None;
	}

[Dylan-DPC-zz]

This is fixed, closing it. Thanks :)

[estebank]

@Dylan-DPC I don't think it was (unless the nightly docs haven't been updated with the fix yet). In the String::split section, if you click on DoubleEndedIterator you end up in a 404.

[ehuss]

It might be useful to have a tracking issue for all of the causes of broken links (maybe reuse this one?), because there are dozens of broken links, and I think some are for different reasons. It'd be nice to have a goal to remove the linkchecker whitelist. I'd guess most of them are waiting for intra-rustdoc-links to handle all the appropriate scenarios? It's not really clear what needs to be done.

[ehuss]

Reopening until the exclusions in the link checker can be removed.

[jyn514]

I think this was fixed in #74485? Should this be closed?

[ehuss]

Considering that all the String links seem to be fixed, I think it is good to close. It looks like there are plenty of other tracking issues for the last remaining broken links (like #63351 and #74481), so I think those are adequately tracked. Thanks everyone for putting all the work into intra-doc-links!