Fixes#93428
This fix is a response to a couple of special cases related to the
`module_id`, which is eventually used for trait candidates:
* The module id is always set to the current crate, when checking `crate::`.
Normally, the set of in-scope traits would be set in `load_links_in_attrs`,
but if there are no doc comments, then that loop will never run.
* the module id is set to the parent module, when resolving a module
that is spelled like this:
// Notice how we use an outlined doc comment here!
// [`Test::my_fn`]
mod something {
}
As with the above problem with `crate::`, we need to make sure the
module gets its traits in scope resolved, even if it has no doc comments
of its own.
rustdoc: Pre-calculate traits that are in scope for doc links
This eliminates one more late use of resolver (part of #83761).
At early doc link resolution time we go through parent modules of items from the current crate, reexports of items from other crates, trait items, and impl items collected by `collect-intra-doc-links` pass, determine traits that are in scope in each such module, and put those traits into a map used by later rustdoc passes.
r? `@jyn514`
Store a `Symbol` instead of an `Ident` in `AssocItem`
This is the same idea as #92533, but for `AssocItem` instead
of `VariantDef`/`FieldDef`.
With this change, we no longer have any uses of
`#[stable_hasher(project(...))]`
This is the same idea as #92533, but for `AssocItem` instead
of `VariantDef`/`FieldDef`.
With this change, we no longer have any uses of
`#[stable_hasher(project(...))]`
Before, the trait's associated item would be used. Now, the impl's
associated item is used. The only exception is for impls that use
default values for associated items set by the trait. In that case,
the trait's associated item is still used.
As an example of the old and new behavior, take this code:
trait MyTrait {
type AssocTy;
}
impl MyTrait for String {
type AssocTy = u8;
}
Before, when resolving a link to `String::AssocTy`,
`resolve_associated_trait_item` would return the associated item for
`MyTrait::AssocTy`. Now, it would return the associated item for
`<String as MyTrait>::AssocTy`, as it claims in its docs.
The field is also renamed from `ident` to `name. In most cases,
we don't actually need the `Span`. A new `ident` method is added
to `VariantDef` and `FieldDef`, which constructs the full `Ident`
using `tcx.def_ident_span()`. This method is used in the cases
where we actually need an `Ident`.
This makes incremental compilation properly track changes
to the `Span`, without all of the invalidations caused by storing
a `Span` directly via an `Ident`.
These closures were quite complex and part of a quite complex function.
The fact that they are closures makes mistakes likely when refactoring.
For example, earlier, I meant to use `resolved`, an argument of the
closure, but I instead typed `res`, which captured a local variable and
caused a subtle bug that led to a confusing test failure.
Extracting them as functions makes the code easier to understand and
refactor.
This hack was added in 6ab1f05697.
I don't know what change allowed removing the hack, but that commit
added a test (which I presume covered the hack's behavior), and all
tests are passing with this change. So, I think it should be good.
rustdoc: Remove the intra-doc links side channel
The side channel made the code much more complex and harder to
understand. It was added as a temporary workaround in
0c99d806ea, but it's no longer necessary.
The addition of `UrlFragment` in #92088 was the key to getting rid of
the side channel. The semantic information (rather than the strings that
used to be used for fragments) that is now captured by `UrlFragment` is
enough to obviate the side channel. An additional change had to be made
to `UrlFragment` in this PR to make this possible: it now records
`DefId`s rather than item names.
This PR also consolidates the checks for anchor conflicts into one place.
r? `@Manishearth`
I had the epiphany that now that fragments are "semantic" -- rather than
just strings -- they fill the role that used to be handled by the side
channel. I think I may be able to get rid of the other uses of the side
channel using this technique too.
Rustdoc: resolve associated traits for non-generic primitive types
Fixes#90703
This seems to work:
<img width="457" alt="image" src="https://user-images.githubusercontent.com/2807772/147774059-9556ff96-4519-409e-8ed0-c33ecc436171.png">
I'm just afraid I might have missed some cases / broken previous functionality.
I also have not written tests yet, I will have to take a look to see where tests are and how they are structured, but any help there is also appreciated.
Fix rustdoc::private_doc_tests lint for public re-exported items
Closes#72081
This involves changing the lint to check the access level is exported, rather than public.
The [exported access level](e91ad5fc62/compiler/rustc_middle/src/middle/privacy.rs (L24)) accounts for public items and items accessible to other crates with the help of `pub use` re-exports.
The pattern of re-exporting public items from a private module is usage seen in a number of popular crates.
rustc_metadata: Encode list of all crate's traits into metadata
While working on https://github.com/rust-lang/rust/pull/88679 I noticed that rustdoc is casually doing something quite expensive, something that is used only for error reporting in rustc - collecting all traits from all crates in the dependency tree.
This PR trades some minor extra time spent by metadata encoder in rustc for major gains for rustdoc (and for rustc runs with errors, which execute the `all_traits` query for better diagnostics).
This involves changing the lint to check the access level is exported,
rather than public. The exported access level accounts for public items
and items accessible to other crates with the help of `pub use`
re-exports. The pattern of re-exporting public items from a private
module is usage seen in a number of popular crates.
This is a step in the direction of computing the links more lazily,
which I think will simplify the implementation of intra-doc links.
This will also make it easier to eventually use the actual `Res` for
associated items, enum variants, and fields, rather than their HTML
page's `Res`.
