Turns out gn has 3 types of sources:
inputs (non-source files)
sources (source files)
public (source files for APIs)

I've now taught missing_from_gn about inputs
brettw says I can ignore public for now as it's unlikely
anyone is using that.

R=brettw@chromium.org

Review URL: https://codereview.chromium.org/675283002

Review URL: https://codereview.chromium.org/676093002

Specs: Fix scripts to run at </script> not <script>
Specs: Update parsing to delay on pending imports before running
scripts or firing 'load'
Specs: Fire 'load' at Application or Module rather than Document
Specs: Imply there's going to be a start tag token observer callback
so that you can implement prefetching for custom elements

R=abarth@chromium.org

Review URL: https://codereview.chromium.org/677533004

We want to re-use the code for draining Mojo data pipes in more places that
just the MojoLoader.

R=eseidel@chromium.org

Review URL: https://codereview.chromium.org/650903006

Most of the files listed here can just be removed
from the repository.

R=abarth@chromium.org

Review URL: https://codereview.chromium.org/679783002

There's only one parser: the HTMLDocumentParser. No need for the complex class
hierarchy.

R=eseidel@chromium.org

Review URL: https://codereview.chromium.org/680583003

https://codereview.chromium.org/673073002/

R=abarth@chromium.org

Review URL: https://codereview.chromium.org/654393004

This LICENSE file is redundant with the one in the top-level directory.

R=ojan@chromium.org

Review URL: https://codereview.chromium.org/674013003

R=ojan@chromium.org

Review URL: https://codereview.chromium.org/679583002

We only allow overflow scrolling. The frame isn't special.
This is a first step in making that happen. There's a lot of
code to remove after this patch, but this gets rid of
ScrollView and a bunch of frame-level scrolling code.

Had to add in a FrameWidget class so that Scrollbar.cpp had
a way of getting to FrameView::removeChild without pulling
a core class into platform. This might go away when we rip
out the Widget tree if we made it so that FrameView didn't
keep a list of Scrollbar instances.

Modified scrollbar.html to use overflow scrolling instead of
frame level scrolling. Once we get rid of the split between
Document and documentElement, we'll be able to make the root
element in the page scrollable as well (i.e. any child of the
Document).

R=abarth@chromium.org

Review URL: https://codereview.chromium.org/646273006

R=ojan@chromium.org

Review URL: https://codereview.chromium.org/677573002

Review URL: https://codereview.chromium.org/642903004

R=ojan@chromium.org

Review URL: https://codereview.chromium.org/640143004

I accidentally broke border parsing when
removing CSS color keyword hacks.

While debugging this I also ran into a
race condition in the debugger, which
is fixed here.

Add --gdb argument to skydb

Makes it trivial to drop into gdb having launched
mojo_shell/sky with the right arguments.

R=ojan@chromium.org

Review URL: https://codereview.chromium.org/673073002

R=abarth@chromium.org

Review URL: https://codereview.chromium.org/657393004

The stack now just has ContainerNodes.

R=eseidel@chromium.org

Review URL: https://codereview.chromium.org/666913005

R=abarth@chromium.org

Review URL: https://codereview.chromium.org/673993002

R=abarth@chromium.org

Review URL: https://codereview.chromium.org/676933002

Makes it trivial to drop into gdb having launched
mojo_shell/sky with the right arguments.

R=abarth@chromium.org

Review URL: https://codereview.chromium.org/672383002

R=abarth@chromium.org

Review URL: https://codereview.chromium.org/676923002

This makes it much easier to hack on test cases.

R=abarth@chromium.org

Review URL: https://codereview.chromium.org/654543004

R=eseidel@chromium.org

Review URL: https://codereview.chromium.org/673963002

There's no need to simulate the tree builder.  It doesn't have any state.

R=eseidel@chromium.org

Review URL: https://codereview.chromium.org/678533003

Now that we don't have checkpoints, we can just use SegmentedString directly.

R=eseidel@chromium.org

Review URL: https://codereview.chromium.org/676763003

R=abarth@chromium.org

Review URL: https://codereview.chromium.org/671173006

R=abarth@chromium.org

Review URL: https://codereview.chromium.org/675873002

I had to also register for the text/plain mime-type
which turned out to be harder than expected.
Mostly due to my confusion with mojo_shell
using last-argument-wins argument parsing.

R=ianh@google.com

Review URL: https://codereview.chromium.org/672363002

R=esprehn@chromium.org

Review URL: https://codereview.chromium.org/659043004

R=abarth@chromium.org

Review URL: https://codereview.chromium.org/678533002

I gave it life in https://codereview.chromium.org/13896017 and now I'm going to take it away.

R=abarth@chromium.org

Review URL: https://codereview.chromium.org/647323003

We still have WebThread::Task, but WebThread is now gone.

R=esprehn@chromium.org

Review URL: https://codereview.chromium.org/673033002

R=ojan@chromium.org

Review URL: https://codereview.chromium.org/672353002

We now observe the message loop and the mojo message pump directly.

R=ojan@chromium.org, eseidel@chromium.org

Review URL: https://codereview.chromium.org/673783006

Previously, Canvas2DLayerManager was using addTaskObserver to jump the task
queue. After this CL, it goes to the end of the queue like a normal task.

R=ojan@chromium.org

Review URL: https://codereview.chromium.org/673783005

R=abarth@chromium.org

Review URL: https://codereview.chromium.org/645513006

R=abarth@chromium.org

Review URL: https://codereview.chromium.org/672323002