FxHash is faster than FnvHash and SipHash for simple types up to at
least 64 bytes. The cryptographic guarantees are not needed for any
types changed here because they are simple ids.
This changes the types in script and net crates.
In a future PR we will change the remaining Fnv to be also Fx unless
there is a reason to keep them as Fnv.
Testing: Should not change functionality but unit test and wpt will find
it.
Signed-off-by: Narfinger <Narfinger@users.noreply.github.com>
These changes add compile-time assertions that:
* any type that implements the Serializable/Transferable trait has a
`[Serializable]` or `[Transferable]` annotation in the interface WebIDL
* any WebIDL interface with the `[Serializable]` or `[Transferable]`
annotation implements the corresponding trait
This is useful because it means that WebIDL definitions will be less
confusing if you're trying to figure out whether Servo supports
serializing/transferring a particular interface type. It also makes
fixing #21715 in the future a little bit easier, because the annotations
will remain up to date.
Testing: compile-time only; no point in writing tests for this since it
involves webidl codegen.
---------
Signed-off-by: Josh Matthews <josh@joshmatthews.net>
The structured cloning with transfer list
https://html.spec.whatwg.org/multipage/#structuredserializewithtransfer
throws a "DataCloneError" DOM expection by default if
serialization/transferral
is not possible, but a platform object can throw a custom excepton on
its serialization/transfer steps.
One example is OffscreenCanvas, which can throw
an "InvalidStateError" exception if the context mode is not none on
transfer steps.
https://html.spec.whatwg.org/multipage/#the-offscreencanvas-interface:transfer-steps
Testing: Improvements in the following tests
-
html/canvas/offscreen/manual/the-offscreen-canvas/offscreencanvas.transferrable*
Fixes: #37919
Signed-off-by: Andrei Volykhin <andrei.volykhin@gmail.com>
This reduces the boilerplate necessary for adding new
serializable/transferable interfaces to the structured cloning code. We
always need to root the deserialized objects when performing a read
operation, but we don't actually need the concrete object types in the
majority of cases. By storing a list of rooted JS object values, we can
push generic reflector objects into it, and extract the types we need
(MessagePort) at the very end.
Testing: Existing WPT structured cloning tests will provide coverage.
Signed-off-by: Josh Matthews <josh@joshmatthews.net>
Rather than creating unique types for each pipeline-namespaced index
type (eg. MessagePortId, DomExceptionId, etc.), we can create a generic
common type that uses a marker to prevent type confusion. This change
allows us to reduce the boilerplate code required when implementing
serializable/transferable interfaces, since the structured clone
implementation can rely on the common type.
Testing: Existing WPT tests for serialization and transferring provide
coverage.
---------
Signed-off-by: Josh Matthews <josh@joshmatthews.net>
Accept transfer argument for StructuredCloneData::write
Allow structured clone reads to return a boolean
Add Transferable trait
Add basic skeletons to MessagePort
Implement transfer and transfer-receiving steps on MessagePort
Use transfer and transfer_receive in StructuredClone callbacks
Implement MessageChannel
Freeze the array object for the MessageEvent ports attribute
Implement transfer argument on window.postMessage
Use ReentrantMutex instead for MessagePortInternal
Accept origin as a parameter in dispatch_jsval
Fix BorrowMut crash with pending_port_message
Detach port on closure and check for detached during transfer
Enable webmessaging tests
fix webidl
fix
