rustdoc: various cross-crate reexport fixes
Fixes for various smaller cross-crate reexport issues.
The PR is split into several commits for easier review. Will be squashed after approval.
Most notable changes:
* We finally render late-bound lifetimes in the generic parameter list of cross-crate functions & methods.
Previously, we would display the re-export of `pub fn f<'s>(x: &'s str) {}` as `pub fn f(x: &'s str)`
* We now render unnamed parameters of cross-crate functions and function pointers as underscores
since that's exactly what we do for local definitions, too. Mentioned as a bug in #44306.
* From now on, the rendering of cross-crate trait-object types is more correct:
* `for<>` parameter lists (for higher-ranked lifetimes) are now shown
* the return type of `Fn{,Mut,Once}` trait bounds is now displayed
Regarding the last list item, here is a diff for visualization (before vs. after):
```patch
- dyn FnOnce(&'any str) + 'static
+ dyn for<'any> FnOnce(&'any str) -> bool + 'static
```
The redundant `+ 'static` will be removed in a follow-up PR that will hide trait-object lifetime-bounds if they coincide with [their default](https://doc.rust-lang.org/reference/lifetime-elision.html#default-trait-object-lifetimes) (see [Zulip discussion](https://rust-lang.zulipchat.com/#narrow/stream/266220-rustdoc/topic/clean_middle_ty.3A.20I.20need.20to.20add.20a.20parameter/near/307143097)). `FIXME(fmease)`s were added.
``@rustbot`` label A-cross-crate-reexports
r? ``@GuillaumeGomez``
rustdoc: add hash to filename of toolchain files
All static files used by rustdoc are now stored in static.files/ and their filenames include a hash of their contents. Their filenames no longer include the contents of the --resource-suffix flag. This clarifies caching semantics. Anything in static.files can use Cache-Control: immutable because any updates will show up as a new URL.
Invocation-specific files like crates-NN.js, search-index-NN.js, and sidebar-items-NN.js still get the resource suffix.
This has a useful side effect: once toolchain files aren't affected by resource suffix, it will become possible for docs.rs to include crate version in the resource suffix. That should fix a caching issue with `/latest/` URLs: https://github.com/rust-lang/docs.rs/issues/1593. My goal is that it should be safe to serve all rustdoc JS, CSS, and fonts with infinite caching headers, even when new versions of a crate are uploaded in the same place as old versions.
The --disable-minification flag is removed because it would vary the output of static files based on invocation flags. Instead, for rustdoc development purposes it's preferable to symlink static files to a non-minified copy for quick iteration.
Example listing:
```
$ cd build/x86_64-unknown-linux-gnu/doc/ && find . | egrep 'js$|css$' | egrep -v 'sidebar-items|implementors' | sort
./crates1.65.0.js
./rust.css
./search-index1.65.0.js
./source-files1.65.0.js
./static.files/ayu-2bfd0af01c176fd5.css
./static.files/dark-95d11b5416841799.css
./static.files/light-c83a97e93a11f15a.css
./static.files/main-efc63f77fb116394.js
./static.files/normalize-76eba96aa4d2e634.css
./static.files/noscript-5bf457055038775c.css
./static.files/rustdoc-7a422337900fa894.css
./static.files/scrape-examples-3dd10048bcead3a4.js
./static.files/search-47f3c289722672cf.js
./static.files/settings-17b08337296ac774.js
./static.files/settings-3f95eacb845293c0.css
./static.files/source-script-215e9db86679192e.js
./static.files/storage-26d846fcae82ff09.js
```
Fixes#98413
rustdoc: Do not add external traits to the crate in `register_res`
It's not clear why it was done, and apparently it's no longer necessary now.
Such additions are unpredictable for early doc link resolution and would force us to collect all doc links from all external traits.
Fixes https://github.com/rust-lang/rust/issues/103463
Remove generation of tuple struct fields in the search index
This comes from [this discussion](https://github.com/rust-lang/rust/pull/103710) as they're not very useful.
r? `@notriddle`
It's not clear why it was done, and apparently it's no longer necessary now.
Such additions are unpredictable for early doc link resolution and would force us to collect all doc links from all external traits.
All static files used by rustdoc are now stored in static.files/ and
include a hash of their contents. They no longer include the contents of
the --resource-suffix flag. This clarifies caching semantics. Anything
in static.files can use Cache-Control: immutable because any updates
will show up as a new URL.
Invocation-specific files like crates-NN.js, search-index-NN.js,
and sidebar-items-NN.js still get the resource suffix.
The --disable-minification flag is removed because it would vary the
output of static files based on invocation flags. Instead, for
rustdoc development purposes it's preferable to symlink static files
to a non-minified copy for quick iteration.
Fix unwanted merge of inline doc comments for impl blocks
Fixes https://github.com/rust-lang/rust/issues/102909.
We need this merge mechanism for inlined items but it's completely unwanted for impl blocks (at least the doc comments are, not the other attributes) since we want to keep what `cfg()` is put on the `pub use` or other attributes.
r? ``@notriddle``
rustdoc: don't mark Box<T> as Iterator, Read, etc
Because Box<T> has pass-through implementations, rustdoc was giving it the "Notable Traits" treatment for Iterator, Read, Write, and Future, even when the type of T was unspecified.
Pin had the same problem, but just for Future.
Fixes#100320
Because Box<T> has pass-through implementations, rustdoc was giving it the
"Notable Traits" treatment for Iterator, Read, Write, and Future, even when the
type of T was unspecified.
Pin had the same problem, but just for Future.
rustdoc: do not filter out cross-crate `Self: Sized` bounds
All type parameters **except `Self`** are implicitly `Sized` ([via](https://doc.rust-lang.org/nightly/std/marker/trait.Sized.html)). Previously, we disregarded the exception of `Self` and omitted cross-crate `Sized` bounds of *any* type parameter *including* `Self` when rendering.
From now on, we *do* render cross-crate `Self: Sized` bounds.
Most notably, in `std` we now finally properly render the `Sized` bound of the `Clone` trait as well as the `Self: Sized` bound on `Iterator::map`.
Fixes#24183.
``@rustbot`` label T-rustdoc A-cross-crate-reexports
r? rustdoc
Delay ambiguity span bug in normalize query iff not rustdoc
Oli and I decided that the compiler debt of adding another usage of `tcx.sess.opts.actually_rustdoc` is fine, because we don't really want to add more complexity to the normalize query, and moving rustdoc to use fulfill normalization (`fully_normalize`, i.e. not use the normalize query) is unnecessary overhead given that it's skipping binders and stuff.
r? oli-obk
Fixes#102827Fixes#103181
rustdoc: render more cross-crate HRTBs properly
Follow-up to #102439.
Render the `for<>` parameter lists of cross-crate higher-rank trait bounds (in where-clauses and in `impl Trait`).
I've added a new field `bound_params` to `clean::WherePredicate::EqPredicate` (mirroring its sibling variant `BoundPredicate`). However, I had to box the existing fields since `EqPredicate` used to be the largest variant (128 bytes on 64-bit systems) and it would only have gotten bigger).
Not sure if you like that approach. As an alternative, I could pass the uncleaned `ty::Predicate` alongside the cleaned `WherePredicate` to the various re-sugaring methods (similar to what `clean::AutoTraitFinder::param_env_to_generics` does).
I haven't yet added the HTML & JSON rendering code for the newly added `bound_params` field since I am waiting for your opinion. Those two rendering code paths should actually be unreachable in practice given we re-sugar all(?) equality predicates to associated type bindings (and arbitrary equality predicates are not part of the Rust surface language at the time of this writing).
If you agree with storing `bound_params` in `EqPredicate`, I think I can use it to greatly simplify the `clean::auto_trait` module (by also using `simplify::merge_bounds`). Maybe I can do that in any case though.
`@rustbot` label T-rustdoc A-cross-crate-reexports
r? `@GuillaumeGomez`
rustdoc: re-sugar more cross-crate trait bounds
Previously, we would only ever re-sugar cross-crate predicates like `Type: Trait, <Type as Trait>::Name == Rhs` to `Type: Trait<Name = Rhs>` if the `Type` was a generic parameter like `Self` or `T`. With this PR, `Type` can be any type.
Most notably, this means that we now re-sugar predicates involving associated types (where `Type` is of the form `Self::Name`) which are then picked up by the pre-existing logic that re-sugars them into bounds. As a result of that, the associated type `IntoIter` of `std`'s `IntoIterator` trait (re-exported from `core`) is no longer rendered as:
```rust
type IntoIter: Iterator
where
<Self::IntoIter as Iterator>::Item == Self::Item;
```
but as one would expect: `type IntoIter: Iterator<Item = Self::Item>;`.
Cross-crate closure bounds like `F: Fn(i32) -> bool` are now also rendered properly (previously, the return type (`Self::Output`) would not be rendered and we would show the underlying equality predicate).
Fixes#77763.
Fixes#84579.
Fixes#102142.
`@rustbot` label T-rustdoc A-cross-crate-reexports
r? rustdoc
