This is an attempt to balance three problems, each of which would
be violated by a simpler implementation:
- A type alias should show all the `impl` blocks for the target
type, and vice versa, if they're applicable. If nothing was
done, and rustdoc continues to match them up in HIR, this
would not work.
- Copying the target type's docs into its aliases' HTML pages
directly causes far too much redundant HTML text to be generated
when a crate has large numbers of methods and large numbers
of type aliases.
- Using JavaScript exclusively for type alias impl docs would
be a functional regression, and could make some docs very hard
to find for non-JS readers.
- Making sure that only applicable docs are show in the
resulting page requires a type checkers. Do not reimplement
the type checker in JavaScript.
So, to make it work, rustdoc stashes these type-alias-inlined docs
in a JSONP "database-lite". The file is generated in `write_shared.rs`,
included in a `<script>` tag added in `print_item.rs`, and `main.js`
takes care of patching the additional docs into the DOM.
The format of `trait.impl` and `type.impl` JS files are superficially
similar. Each line, except the JSONP wrapper itself, belongs to a crate,
and they are otherwise separate (rustdoc should be idempotent). The
"meat" of the file is HTML strings, so the frontend code is very simple.
Links are relative to the doc root, though, so the frontend needs to fix
that up, and inlined docs can reuse these files.
However, there are a few differences, caused by the sophisticated
features that type aliases have. Consider this crate graph:
```text
---------------------------------
| crate A: struct Foo<T> |
| type Bar = Foo<i32> |
| impl X for Foo<i8> |
| impl Y for Foo<i32> |
---------------------------------
|
----------------------------------
| crate B: type Baz = A::Foo<i8> |
| type Xyy = A::Foo<i8> |
| impl Z for Xyy |
----------------------------------
```
The type.impl/A/struct.Foo.js JS file has a structure kinda like this:
```js
JSONP({
"A": [["impl Y for Foo<i32>", "Y", "A::Bar"]],
"B": [["impl X for Foo<i8>", "X", "B::Baz", "B::Xyy"], ["impl Z for Xyy", "Z", "B::Baz"]],
});
```
When the type.impl file is loaded, only the current crate's docs are
actually used. The main reason to bundle them together is that there's
enough duplication in them for DEFLATE to remove the redundancy.
The contents of a crate are a list of impl blocks, themselves
represented as lists. The first item in the sublist is the HTML block,
the second item is the name of the trait (which goes in the sidebar),
and all others are the names of type aliases that successfully match.
This way:
- There's no need to generate these files for types that have no aliases
in the current crate. If a dependent crate makes a type alias, it'll
take care of generating its own docs.
- There's no need to reimplement parts of the type checker in
JavaScript. The Rust backend does the checking, and includes its
results in the file.
- Docs defined directly on the type alias are dropped directly in the
HTML by `render_assoc_items`, and are accessible without JavaScript.
The JSONP file will not list impl items that are known to be part
of the main HTML file already.
[JSONP]: https://en.wikipedia.org/wiki/JSONP
This is shorter, avoids potential conflicts with a crate
named `implementors`[^1], and will be less confusing when JS
include files are added for type aliases.
[^1]: AFAIK, this couldn't actually cause any problems right now,
but it's simpler just to make it impossible than relying on never
having a file named `trait.Foo.js` in the crate data area.
This commit makes three changes for consistency and readability:
- It shows the sibling navigation on module pages. It's weird
that it didn't work before, and is inconsistent with everything
else (even Crates have sibling navigation with other Crates).
- It hides the "In [parent]" header if it's the same as the
current crate, and if there's no other header between them.
We need to keep it on modules and types, since they have
their own header and data between them, and we don't want
to show siblings under a header implying that they're children.
- It adds a margin to deal with the headers butting directly into
the branding lockup.
It's misleading. The main heading sometimes isn't an fully qualified name at all.
It's also redundant. It's always a child of `div.main-heading`, so just use that.
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.
The exact amount that this reduces the size of an implementors file depends
on whether most of the impls are synthetic or not. For `Send`, it reduces
the file from 128K to 116K, while for `Clone` it went from 64K to 52K.
Now that the "All Crates" dropdown is only rendered on the search results page,
there is no need to load crates.js on most pages. Load it only on crate pages.
Also, add the `defer` attribute so it does not block page rendering.
For sidebar-items.js, move the script tag to `<head>`. Since it already has the
defer attribute it won't block loading. The defer attribute does preserve
ordering between scripts, so instead of the callback on load, it can set a
global variable on load, which is slightly simpler. Also, since it is required
to finish rendering the page, beginning its load earlier is better.
Remove generation and handling of sidebar-vars. Everything there can be computed
with information available in JS via other means.
Remove the "other" wrapper in the sidebar. It was unnecessary.
Remove excess script fields
Previously, search.js relied on the DOM and the `window` object. It can now be
loaded in the absence of the DOM, for instance by Node. The same is true of
search-index.js.
This allows removing a lot of code from src/tools/rustdoc-js/tester.js that
tried to parse search.js and extract specific functions that were needed for
testing.
rustdoc: avoid many `Symbol` to `String` conversions.
Particularly when constructing file paths and fully qualified paths.
This avoids a lot of allocations, speeding things up on almost all
examples.
r? `@GuillaumeGomez`
We had been injecting the list of themes and the rustdoc version into
main.js by rewriting it at doc generation time. By avoiding this
rewrite, we can make it easier to edit main.js without regenerating all
the docs.
Added a more convenient accessor for rustdoc-vars.
Changed storage.js to not rely on resourcesSuffix. It could in theory
use rustdoc-vars, but because rustdoc-vars is at the end of the HTML,
it's not available when storage.js runs (very early in page load).
We carry around a list of stylesheets that can carry two different types
of thing:
1. Internal stylesheets specific to a page type (only for settings)
2. Themes
In this change I move the link generation for settings.css into
settings(), so Context.style_files is reserved just for themes.
We had two places where we extracted a base theme name from a list of
StylePaths. I consolidated that code to be a method on StylePath.
I moved generation of link tags for stylesheets into the page.html
template. With that change, I made the template responsible for special
handling of light.css (making it the default theme) and of the other
themes (marking them disabled). That allowed getting rid of the
`disabled` field on StylePath.
This change has two advantages:
1. It makes the possible states clearer, and it makes it impossible to
construct invalid states, such as a blanket impl that is also an auto
trait impl.
2. It shrinks the size of `Impl` a bit, since now there is only one
field, rather than two.
rustdoc: Compute some fields of `clean::Crate` on-demand to reduce size
`clean::Crate` is frequently moved by-value -- for example, in `DocFolder`
implementations -- so reducing its size should improve performance.
This PR reduces the size of `clean::Crate` from 168 bytes to 104 bytes.
r? `@jyn514`
