IDL `record` types can themselves contain types that are not described
anywhere else. An example is in
https://github.com/servo/servo/issues/37038, where the `record` contains
a definition of a union. These inner types must be returned from
`getAllTypes`, otherwise we won't generate code for them.
This PR also adds a few type annotations. I can remove them if
requested, but I think they're helpful.
Testing: Includes a regression test
Fixes: https://github.com/servo/servo/issues/37038
---------
Signed-off-by: Simon Wülker <simon.wuelker@arcor.de>
Before it was only used when converting to a `Record`, using it all the
times allow us to remove two methods.
Plus added a helper method in CodegenRust.py to avoid repeated code.
Testing: a successful build and existing tests should cover the changes.
Fixes: #36410
---------
Signed-off-by: Gae24 <96017547+Gae24@users.noreply.github.com>
These changes make us match Gecko's setup for how Window and non-Window
globals are initialized. Since Window globals are much more common than
Worker globals, using lazy interface definitions can be a useful memory
optimization at the expense of increased complexity for property
lookups.
Also adds the MayResolve hook for all globals, which is an optimization
for the JIT to avoid calling resolve hooks unnecessarily.
Testing: Existing test coverage on global interfaces should suffice.
---------
Signed-off-by: Josh Matthews <josh@joshmatthews.net>
These changes make our implementation of the enumeration hook for
globals [match
Gecko's](https://searchfox.org/mozilla-central/rev/1f65969e57c757146e3e548614b49d3a4168eeb8/dom/base/nsGlobalWindowInner.cpp#3297),
fixing an assertion failure that occurred in the previous
implementation.
Our enumeration hook is supposed to fill a vector with names of
properties on the global object without modifying the global in any way;
instead we were defining all of the missing webidl interfaces. We now do
much less work and crash less.
Testing: New crashtest based on manual testcase.
Fixes: #34686
---------
Signed-off-by: Josh Matthews <josh@joshmatthews.net>
This is needed to implement features like `Response.json` which is a
static helper added to the fetch spec which overlaps with the `json`
instance method `Response` has from `Body`.
Partly based these changes on what Firefox does for this same issue.
(https://searchfox.org/mozilla-central/source/dom/bindings/Codegen.py
and
https://searchfox.org/mozilla-central/source/dom/bindings/Configuration.py
specifically keying `binaryNameFor` on name and `isStatic`).
Testing: I locally updated the Response.webidl to contain the new static
`json` and it compiles.
Signed-off-by: Sebastian C <sebsebmc@gmail.com>
copy generated `include!`d files from script_bindings's OUT_DIR, to
script's OUT_DIR to allow Rust Analyzer to load them.
This is done to bypass limitation of Rust Analyzer:
https://github.com/rust-lang/rust-analyzer/issues/17040
Also build script will now be rerun only when there are actual changes
to concrete bindings due to emitted `cargo::rerun-if-changed` (not for
each change in script crate).
Testing: It compiles so it works, I tested manually and RA now works as
expected (although we need to from type alias to concrete union-types
definitions)
Fixes:
https://servo.zulipchat.com/#narrow/channel/263398-general/topic/rust-analyzer.20failed.20to.20include.20codes.20in.20script_bindings
---------
Signed-off-by: sagudev <16504129+sagudev@users.noreply.github.com>
This is the final step of #1799, where the majority of the generated
code for the JS bindings is now compiled as part of the script_bindings
build step. The remaining pieces in script must live there because they
refer to concrete DOM types; all code in script_bindings is generic over
the
[DomTypes](https://doc.servo.org/script/dom/bindings/codegen/DomTypes/trait.DomTypes.html)
trait.
My testing with incremental builds shows me a 12 second reduction in
build times on my 2024 M4 Macbook Pro when modifying code in the script
crate after these changes. Before this PR those changes took 20 seconds
to rebuild Servo, and now they take 8 seconds.
Testing: Existing WPT tests ensure no regressions.
Fixes: #1799
---------
Signed-off-by: Josh Matthews <josh@joshmatthews.net>
* script: Move HasParent to script_bindings and update imports for InheritTypes.
Signed-off-by: Josh Matthews <josh@joshmatthews.net>
* script: Make principal creation generic over DOM interface.
Signed-off-by: Josh Matthews <josh@joshmatthews.net>
* script: Move a bunch of proxy-related code to script_bindings.
Signed-off-by: Josh Matthews <josh@joshmatthews.net>
* script: Make some proxy-related code generic over the DOM interface.
Signed-off-by: Josh Matthews <josh@joshmatthews.net>
* script: Move DomSlice to script_bindings.
Signed-off-by: Josh Matthews <josh@joshmatthews.net>
* script: Move some utility bindings code to script_bindings.
Signed-off-by: Josh Matthews <josh@joshmatthews.net>
* script: Make enumerating and resolving globals generic over the DOM interface.
Signed-off-by: Josh Matthews <josh@joshmatthews.net>
* script: Make realm helpers generic over the DOM interface.
Signed-off-by: Josh Matthews <josh@joshmatthews.net>
* script: Move implementations on concrete DOM types to concrete bindings.
Signed-off-by: Josh Matthews <josh@joshmatthews.net>
* script: Make additional codegen helpers generic over the DOM interface.
Signed-off-by: Josh Matthews <josh@joshmatthews.net>
* script: Make iterator creation generic over the DOM interface.
Signed-off-by: Josh Matthews <josh@joshmatthews.net>
* script: Make reporting an exception a generic operation.
Signed-off-by: Josh Matthews <josh@joshmatthews.net>
* script: Move AsCCharPtrPtr to script_bindings.
Signed-off-by: Josh Matthews <josh@joshmatthews.net>
* Formatting.
Signed-off-by: Josh Matthews <josh@joshmatthews.net>
* Address clippy warnings.
Signed-off-by: Josh Matthews <josh@joshmatthews.net>
---------
Signed-off-by: Josh Matthews <josh@joshmatthews.net>
* Migrate to 2024 edition
Signed-off-by: Simon Wülker <simon.wuelker@arcor.de>
* Allow unsafe_op_in_unsafe_fn lint
This lint warns by default in the 2024
edition, but is *way* too noisy for servo.
We might enable it in the future, but not now.
Signed-off-by: Simon Wülker <simon.wuelker@arcor.de>
* Compile using the 2024 edition
Signed-off-by: Simon Wülker <simon.wuelker@arcor.de>
---------
Signed-off-by: Simon Wülker <simon.wuelker@arcor.de>
#30151 added support for setlike and maplike declarations in WebIDL, but
the tests only validated if generator code worked with 'DOMString' as
the key type. The support for interface type as members was broken as
`DomRoot<T>` didn't satify the bounds `Eq` and `Hash` needed by the
`Key` and `Value` types in `Setlike` and `Maplike` traits respectively.
In addition, the splitting of bindings into a separate 'script_bindings'
crate had also broken support for this in CodegenRust.py, as the types
used within the definition of `DomTraits` were not referenced using
`Self::`.
This patch fixes the WebIDL code generator by doing a simple string
replacement on the return value of `getRetvalDeclarationForType` so that
the proper `Self::` is used. I'm not not sure if there is a better
approach to this as it seems most logic in CodegenRust.py uses the `D::`
prefix that is expected to be available only when compiling `script`
crate and not `script_bindings`.
This patch also adds the missing trait implementations for `DomRoot` and
ensures that the generated code works for both members of primitive and
interface types by splitting the existing `TestBinding{Map,Set}Like`
interfaces into `TestBinding{Map,Set}LikeWith{Primitive,Interface}`
tests.
Fixes#35542.
Signed-off-by: Mukilan Thiyagarajan <mukilan@igalia.com>
* script: Take away Fallible from new_resolved and new_rejected
Both Promise::new_resolved and new_rejected only return `Ok`. We don't
need them to be fallible. Simply return `Rc<Promise>`, instead of
`Fallible<Rc<Promise>>`. Also, clean up relevant code.
Signed-off-by: Kingsley Yung <kingsley@kkoyung.dev>
* script: pull_algorithm becomes infallible
The method pull_algorithm only returns `Some(Ok(_))`, which means it is
infallible. Clean up the returned type.
Signed-off-by: Kingsley Yung <kingsley@kkoyung.dev>
* script: generic_initialize becomes infallible
The method generic_initialize only returns `Ok(())`, which means it is
infallible. Clean up the returned type.
Signed-off-by: Kingsley Yung <kingsley@kkoyung.dev>
---------
Signed-off-by: Kingsley Yung <kingsley@kkoyung.dev>
* script: Make DOM proxy handlers generic over DOM types.
Signed-off-by: Josh Matthews <josh@joshmatthews.net>
* script: Make finalize and trace hooks generic over DOM types.
Signed-off-by: Josh Matthews <josh@joshmatthews.net>
---------
Signed-off-by: Josh Matthews <josh@joshmatthews.net>
* bindings: Start making generated bindings methods generic over DOM trait.
Signed-off-by: Josh Matthews <josh@joshmatthews.net>
* bindings: Make binding initialization generic over the DOM types trait.
Signed-off-by: Josh Matthews <josh@joshmatthews.net>
* bindings: Start making proxyhandler code generic over DOM types.
Signed-off-by: Josh Matthews <josh@joshmatthews.net>
---------
Signed-off-by: Josh Matthews <josh@joshmatthews.net>
