Implement EventSource and update test expectations
Partial #8925.
Most of the processing model for `EventSource` are unimplemented because we currently don't have a fetch implementation.
<!-- Reviewable:start -->
[<img src="https://reviewable.io/review_button.png" height=40 alt="Review on Reviewable"/>](https://reviewable.io/reviews/servo/servo/9029)
<!-- Reviewable:end -->
`linear_gradients_lengths_ref.html` use `background-clip: content-box`.
The tests clearly expect the background painting area to have width of
100px, but since the default value of `background-clip` is `border-box`
and the boxes in question have a 1-pixel border the background painting
areas are actually 102px. This commit fixes the tests so that the width
of the background painting area is actually 100px.
webgl: Make a general way to get data from a JS array buffer view
This fixes an invalid length being reported from
`float32_array_to_slice` (which used the byte length), and also to
generalize getting data from a JS array buffer view, to reduce code
duplication.
The pending type safety issues, like where we could send a `UInt16Array`
where we expect a `Float32` one, should be solved by IDL bindings in
some cases, like `uniform[n]fv` or `vertexAttrib[n]fv`, and with extra
checks in others, like in the pending `texImage2D(..., ArrayBufferView)`.
r? @jdm
<!-- Reviewable:start -->
[<img src="https://reviewable.io/review_button.png" height=40 alt="Review on Reviewable"/>](https://reviewable.io/reviews/servo/servo/8970)
<!-- Reviewable:end -->
Partial implementation of HTTP-network Fetch step
I've been working on making a partial implementation of HTTP-network Fetch[1] so that I can begin testing the entire Fetch protocol as implemented in Servo. The empty steps are intentionally so, including the incomplete Step 5, which I started without realizing I don't need to do basic testing.
[1] https://fetch.spec.whatwg.org/#http-network-fetch
<!-- Reviewable:start -->
[<img src="https://reviewable.io/review_button.png" height=40 alt="Review on Reviewable"/>](https://reviewable.io/reviews/servo/servo/9153)
<!-- Reviewable:end -->
Make "mach run" use --android by default when enabled in .servobuild
When `.servobuild` has `android = true`, the `mach run` command should default to running on Android.
r? @larsbergstrom
<!-- Reviewable:start -->
[<img src="https://reviewable.io/review_button.png" height=40 alt="Review on Reviewable"/>](https://reviewable.io/reviews/servo/servo/9150)
<!-- Reviewable:end -->
The argument to `local()` is not a font family, but the name of a
»single font face within a larger family« according to
http://www.w3.org/TR/css3-fonts/#src-desc.
The previous implementation of `FontCache::run` would panic if the argument to
`local()` would not be the name of a font family present in `self.web_families`.
That happened since `FontCacheTask` tried to use the argument to `local()` as a
key for `self.web_families`, although it previously correctly initialized a
value for the `family` field of the `AddWebFont` command.
This test tries to add a web font to the `FontCacheTask`. The added web font
corresponds to the following CSS font definition:
```
@font-face {
font-family: "test family";
src: local(test font face);
}
```
This test fails, since `FontCacheTask` tries to get the value for the key
"test font face" from `self.web_families`, but previously initialized
a value for the key "test family".
This fixes an invalid length being reported from
float32_array_to_slice (which used the byte length), and also to
generalize getting data from a JS array buffer view, to reduce code
duplication.
The pending type safety issues, like where we could send a UInt16Array
where we expect a Float32 one, should be solved by IDL bindings in
some cases, like uniform[n]fv or vertexAttrib[n]fv, and with extra
checks in others, like in the pending texImage2D(..., ArrayBufferView).