webgl: Fix texturing
These two tiny changes were making WebGL textures not work.
It was not seen in our texturing test since it only used one texture,
we render to a texture by default, and that texture was bound to
`gl::TEXTURE_2D`.
<!-- Reviewable:start -->
[<img src="https://reviewable.io/review_button.png" height=40 alt="Review on Reviewable"/>](https://reviewable.io/reviews/servo/servo/8788)
<!-- Reviewable:end -->
The validations performed in the script side of this should always
prevent this from happening, and this will allow us to catch bugs like
the previous one.
Fixing unused attributes warnings required updating serde_macros
which required updating to rustc 1.6.0-nightly (a2866e387 2015-11-30)
which required updating some other dependencies.
These two tiny changes were making WebGL textures not work.
It was not seen in our texturing test since we render to a texture by
default, and that texture was bound to `gl::TEXTURE_2D`.
This commit adds angle-based validation and translation to WebGL
shaders.
The changes to the tex_image_2d test is neccessary (it was not valid
GLES 2.0 shader language).
This commit implements WebGL's:
* cullFace
* frontFace
* enable
* disable
* depthMask
* colorMask
* clearDepth
* clearStencil
* depthFunc
* depthRange
* hint
* lineWidth
* pixelStorei
* polygonOffset
* texParameteri
* texParameterf
* texImage2D (partially)
It inlines a lot of OpenGL calls to keep the file
`components/canvas/webgl_paint_task.rs` as small as possible while
keeping readability.
It also improves error detection on previous calls, and sets node damage
on the canvas in the drawing calls.
It adds a `TexImage2D` reftest, even though it's not enabled because:
* WebGL paints the image when it loads (asynchronously), so the reftest doesn't wait for it and it finishes early
* If we change the source for the base64 src of the image it works as expected in non-headless mode, but the test harness locks
To actually make the multiprocess communication work, we'll need to
reroute the task creation to the pipeline or the compositor. But this
works as a first step.
The idea here is to land this before making images and canvas IPC-safe,
because this will shake out bugs relating to the shared memory. There
are currently test timeouts that are preventing multiprocess images and
canvas from landing, and I believe those are due to the inefficiency of
sending large amounts of data in the unoptimized builds we test with. By
moving to shared memory, this should drastically reduce the number of
copies and `serde` serialization.
Under the hood, this uses Mach OOL messages on Mac and temporary
memory-mapped files on Linux.
