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.
We currently store LayerBuffers, because previously NativeSurfaces did
not record their own size. Now we can store NativeSurfaces directly,
which saves a bit of space in the surface cache and allows us to create
LayerBuffers only in the PaintTask.
This also means that instead of sending cached LayerBuffers, the
compositor can just send cached NativeSurfaces to the PaintTask.
an IPC channel instead.
Because this used a boxed trait object to invoke messages across a
process boundary, and boxed trait objects are not supported across IPC,
we spawn a helper thread inside the compositor to perform the marshaling
for us.
Now that NativeDisplay can be shared between the compositor and the
paint task, we can move the LayerBuffer cache to the compositor. This
allows surfaces to be potentially reused between different paint tasks
and will eventually allow OpenGL contexts to be preserved between
instances of GL rasterization.
I've done a bit of job to get this done. Right now readback is still used, but we have a `LayerId` -> `CanvasRenderer` map on the paint task, that we can use to get rid of that.
I'd want review, to see if this is a good approach (I know it's not the initial `CanvasId` -> renderer approach, but it's pretty similar, since a canvas involves a `PaintLayer`).
I had to do a bit of refactoring to avoid cyclic dependencies between canvas and gfx. I'd want you to review them too.
It's mergeable and doesn't break any tests :P
Some of my main concerns:
* Does the canvas render really need to be behind an `Arc<Mutex<T>>`?
* I can't clone a `NativeSurface` right now (that's why the `SendNativeSurface()` msg is unimplemented in the WebGL task). It should be easy to add that to rust-layers, supposing the caller is responsible to mark it as non-leaking, any reason to not do it?
cc @jdm @pcwalton
<!-- Reviewable:start -->
[<img src="https://reviewable.io/review_button.png" height=40 alt="Review on Reviewable"/>](https://reviewable.io/reviews/servo/servo/6083)
<!-- Reviewable:end -->
A rebuild after touching components/profile/mem.rs now takes 48 seconds (and
only rebuilds `profile` and `servo`) which is much lower than it used to be.
In comparison, a rebuild after touching components/profile_traits/mem.rs takes
294 seconds and rebuilds many more crates.
This change also removes some unnecessary crate dependencies in `net` and
`net_traits`.
This is incomplete in several ways:
* It assumes that there's only one constellation (i.e. top level browsing context), ever.
* The session support is very basic indeed (no capabilities)
* Passing channels over channels may not sit well with IPC
* The error handling is mostly missing
- Most of util::memory has been moved into profile::mem, though the
`SizeOf` trait and related things remain in util::memory. The
`SystemMemoryReporter` code is now in a submodule
profile::mem::system_reporter.
- util::time has been moved entirely into profile::time.
This used to conflict with the util crate from the standard library, which
has long since been removed.
The import in layout has not been changed because of a conflict with the
util mod there.
Notes:
* This adds `#![allow(missing_copy_implementations)]` to components/*/lib.rs. I'm not sure how to approach the missing Copy warnings (are there things for which Copy should NOT be implemented, and how can I tell?) so I stuck this in to make life easier when looking through the warnings. I can easily remove this if necessary.
* This leaves the following type of warnings, which I couldn't figure out how to approach (I'll investigate it later if no one else wants to).
```
css/matching.rs:72:23: 72:35 warning: use of deprecated item: Use overloaded core::cmp::PartialEq, #[warn(deprecated)] on by default
css/matching.rs:72 this_as_query.equiv(other)
^~~~~~~~~~~~
css/matching.rs:95:10: 95:49 warning: use of deprecated item: Use overloaded core::cmp::PartialEq, #[warn(deprecated)] on by default
css/matching.rs:95 impl<'a> Equiv<ApplicableDeclarationsCacheEntry> for ApplicableDeclarationsCacheQuery<'a> {
```
It is possible for a PaintTask to start exiting soon after sending new
buffers to the compositor. In that case, the compositor should return
the now unnecessary buffers to the PaintTask so that it can properly
free them.
To accomplish this, the compositor now keeps a hash map of paint task
channels per pipeline id. When a PaintTask exists, the constellation
informs the compositor that it can forget about it. Additionally, the
PaintTask should not wait for any buffers when the engine is doing a
complete shutdown. In that case, the compositor is already halted and
has simply let all buffers leak. We pipe through the shutdown type when
destroying the pipeline to make this decision.
Fixes#2641.
