Move Stylesheet loading and ownership from the layout task into HTML elements
Stylesheets for `HTMLLinkElement`s are now loaded by the resource task, triggered by the element in question. Stylesheets are owned by the elements they're associated with, which can be `HTMLStyleElement`, `HTMLLinkElement`, and `HTMLMetaElement` (for `<meta name="viewport">).
Additionally, the quirks mode stylesheet (just as the user and user agent stylesheets a couple of commits ago), is implemented as a lazy static, loaded once per process and shared between all documents.
This all has various nice consequences:
- Stylesheet loading becomes a non-blocking operation.
- Stylesheets are removed when the element they're associated with is removed from the document.
- It'll be possible to implement the CSSOM APIs that require direct access to the stylesheets (i.e., ~ all of them).
- Various subtle correctness issues are fixed.
One piece of interesting follow-up work would be to move parsing of external stylesheets to the resource task, too. Right now, it happens in the link element once loading is complete, so blocks the script task. Moving it to the resource task would probably be fairly straight-forward as it doesn't require access to any external state.
Depends on #7979 because without that loading stylesheets asynchronously breaks lots of content.
<!-- Reviewable:start -->
[<img src="https://reviewable.io/review_button.png" height=40 alt="Review on Reviewable"/>](https://reviewable.io/reviews/servo/servo/8039)
<!-- Reviewable:end -->
Stylesheets for `HTMLLinkElement`s are now loaded by the resource task, triggered by the element in question. Stylesheets are owned by the elements they're associated with, which can be `HTMLStyleElement`, `HTMLLinkElement`, and `HTMLMetaElement` (for `<meta name="viewport">).
Additionally, the quirks mode stylesheet (just as the user and user agent stylesheets a couple of commits ago), is implemented as a lazy static, loaded once per process and shared between all documents.
This all has various nice consequences:
- Stylesheet loading becomes a non-blocking operation.
- Stylesheets are removed when the element they're associated with is removed from the document.
- It'll be possible to implement the CSSOM APIs that require direct access to the stylesheets (i.e., ~ all of them).
- Various subtle correctness issues are fixed.
One piece of interesting follow-up work would be to move parsing of external stylesheets to the resource task, too. Right now, it happens in the link element once loading is complete, so blocks the script task. Moving it to the resource task would probably be fairly straight-forward as it doesn't require access to any external state.
We cache the state of any live HTMLCollection, keeping track of
a) the optional cached length of the collection, and
b) an optional cursor into the collection (a node in the collection plus its index).
The cache is invalidated based on the version number of the node.
We use these caches for speeding up random access to the collection.
When returning coll[i], we search from the cursor, if it exists,
and otherwise search from the front of the collection.
In particular, both a forward for-loop and a backward for-loop
through the collection will now have each access take O(1)
time rather than O(n) time.
This gets 1000x speed-up on the relevant Dromaeo DOM query tests.
This is the strategy we'll need to take for attributes, and so this change
puts us in a position to handle attributes and state the same way.
This does mean that we stop taking care to track the situations where our
state has reverted to the original state, with no net change. I think that's
probably of negligible value though.
This should make it somewhat easier to experiment with alternative
representations in the future. To reduce churn, this commit leaves the String
field public, though.
Also, this will allow us to use the default String type to represent the IDL
USVString type, which explicitly forbids unpaired surrogates, ans as such is
a better match to the Rust String type.
Correct event dispatching for multiple simultaneous touch points
Instead of just converting the mouse into a single "touch" input, Servo can now listen for multi-touch events from Glutin, maintain a list of active touch points, and dispatch events for all of them.
r? @glennw (for the compositor changes) and @jdm (for the DOM changes)
<!-- Reviewable:start -->
[<img src="https://reviewable.io/review_button.png" height=40 alt="Review on Reviewable"/>](https://reviewable.io/reviews/servo/servo/8232)
<!-- Reviewable:end -->
This adds some overhead, but also provides the small performance benefit of
avoiding dirtying in the case where an event state is toggled an even
number of times between reflows.
The main benefit here though is that it sets us up to be smarter about
what we mark as dirty using restyle hints.
