This fixes an issue where the CSS viewport was too large on high-DPI displays
because it was set to the window size in device pixels, instead of px. This
patch ensures that the window size is converted from device pixels to px
before being passed to script/layout code.
The Window trait now exposes the window size in both device pixels and
density-independent screen coordinates, with clearer method names.
This is cherry-picked from https://github.com/mozilla/servo/pull/1688:
* Initial spec-incompliant implementation of default click action for anchor elements.
* Add documentation; gut the new document URL loading method
and move it all into the new Window method.
* Add test for default event prevention.
Original developer: Josh Matthews <josh@joshmatthews.net>
Right now, the load is kicked off inside the parser glue. This is unfortunate
for several reasons:
1) we'd like to replace the current parser (libhubbub) by our own parser,
written in Rust, so code intertwined with the parser will have to be
rewritten;
2) it is impossible to support dynamically (i.e. from script) created iframes
in this way;
3) the code flow around loading subdocuments is complicated needlessly.
This commit adds the constellation channel (on which the message to actually
load the document is sent) as a field on the Page, to allow HTMLIFrameElement
to access it.
In rewriting the code, support for dynamically created iframes is added, and
a task failure is avoided when the value of the src attribute can not be
parsed.
Fix#2383
These changes introduce `FooEvent::new_uninitialized()` constructor. This constructor use to create `FooEvent` without calling `FooEvent::InitFooEvent`.
#2383 says integrating `FooEvent::new()` and `InitFooEvent`, but we need to preserve simple `FooEvent::new_uninitialized()` for `document.createEvent()`.
@jdm r?
This is a cleaner way to expose those functions, and makes it possible to
remove a significant amount code in rust-mozjs.
The assert() function is no longer exposed, as it was unused and not very
useful.
The long-term plan for SpiderMonkey is to eliminate JSContexts by merging
(most of) it into JSRuntime, so to future-proof our code, we should avoid
creating multiple JSContexts for the same JSRuntime.
However, this implies we'll have to use the same JSContext for objects in
different compartments, so we need to enter compartments. This is done by
using the with_compartment function.
