| commit | 751f4d3c32cbd11f0bfd44c5facf6c62173310bf | [log] [tgz] |
|---|---|---|
| author | Jeff Brown <jeffbrown@google.com> | Tue Jan 26 15:51:01 2016 -0800 |
| committer | Jeff Brown <jeffbrown@google.com> | Tue Jan 26 15:51:01 2016 -0800 |
| tree | 29b0f7c913476543015f919f4e4ecb590cb61759 | |
| parent | 55f4d308e5e8352bc5f41242bd8b19db431bc774 [diff] |
Initial checkin of the new Mozart compositor. The Mozart compositor is designed to support feed forward rendering of scene graphs using scene version codes and node combinators to determine synchronization behaviors. A unique characteristic of Mozart is its ability to allow clients to specify how to resolve situations where content is missing or not ready by supplying alternative choices for what should be rendered. The goal is to avoid stalls by allowing more flexibility in specifying how scene state updates are coordinated. The scene graph consists of nodes of various types which represent primitive operations such as filling a rectangle, drawing an image, or embedding another scene. The children of each node are composed according to the node's combinator rule. The default MERGE combinator rule simply draws all children sequentially and refuses to draw anying if any of the children are blocked. Other rules, such as FALLBACK, allow for alternative choices to be made by the compositor at snapshot time based on what is available. In the simplest cases, a client might provide substitute content to be used in case the primary content is not yet ready. In more elaborate cases, a client might describe how an older version of the primary content can be adapted to fit current demands, perhaps scaling or cropping it as required. The compositor itself is designed to be relatively unopinionated about the contents that it is compositing. Applications publish their scenes and Mozart composites them, that's it. High-level functionality such as maintaining view embedding relationships or input dispatch are exclusively handled by other components. Mozart also offers: - A relatively easy to use client interface. - Vsync based scheduling. - Hit testing (only partially implemented in this patch). - Support for clients which use separate threads for event processing and rendering. - Multi-display support (only partially implemented in this patch). The rasterizer is currently implemented using Skia and Ganesh and runs on a separate thread from the main engine. This initial patch covers most of the essentials but there's plenty more testing, optimization, and elaboration to do later. This patch is followed by several others which serve to update the view manager to use the new compositor, add various helpers, update examples, and so on. BUG= R=abarth@google.com Review URL: https://codereview.chromium.org/1552963002 .
Mojo is an effort to extract a common platform out of Chrome's renderer and plugin processes that can support multiple types of sandboxed content, such as HTML, Pepper, or NaCl.
The instructions below only need to be done once. Note that a simple “git clone” command is not sufficient to build the source code because this repo uses the gclient command from depot_tools to manage most third party dependencies.
$ fetch mojo # append --target_os=android to include Android build support. $ cd src # Or install-build-deps-android.sh if you plan to build for Android. $ ./build/install-build-deps.sh $ mojo/tools/mojob.py gn
The fetch mojo command does the following:
install-build-deps.sh installs any packages needed to build, then mojo/tools/mojob.py gn runs gn args and configures the build directory, out/Debug.
If the fetch command fails, you will need to delete the src directory and start over.
If you configured your set-up for Linux and now wish to build for Android, edit the .gclient file in your root Mojo directory (the parent directory to src.) and add this line at the end of the file:
target_os = [u'android',u'linux']
Bring in Android-specific build dependencies:
$ build/install-build-deps-android.sh
Pull down all of the packages with this command:
$ gclient sync
You can update your checkout like this. The order is important. You must do the git pull first because gclient sync is dependent on the current revision.
# Fetch changes from upstream and rebase the current branch on top $ git pull --rebase # Update all modules as directed by the DEPS file $ gclient sync
You do not need to rerun gn gen out/Debug - ninja does so automatically each time you build. You might need to rerun mojo/tools/mojob.py gn if the GN flags have changed.
Build Mojo for Linux by running:
$ ninja -C out/Debug -j 10
You can also use the mojob.py script for building. This script automatically calls ninja and sets -j to an appropriate value based on whether Goma (see the section on Goma below) is present. You cannot specify a target name with this script.
mojo/tools/mojob.py gn mojo/tools/mojob.py build
Run a demo:
out/Debug/mojo_shell mojo:spinning_cube
Run the tests:
mojo/tools/mojob.py test
Run the benchmarks:
mojo/devtools/common/mojo_benchmark mojo/tools/data/benchmarks
Create a release build:
mojo/tools/mojob.py gn --release mojo/tools/mojob.py build --release mojo/tools/mojob.py test --release
To build for Android, first make sure that your checkout is configured to build for Android. After that you can use the mojob script as follows:
$ mojo/tools/mojob.py gn --android $ mojo/tools/mojob.py build --android
The result will be in out/android_Debug. If you see javac compile errors, make sure you have an up-to-date JDK
If you're a Googler, you can use Goma, a distributed compiler service for open-source projects such as Chrome and Android. If Goma is installed in the default location (~/goma), it will work out-of-the-box with the mojob.py gn, mojob.py build workflow described above.
You can also manually add:
use_goma = true
at the end of the file opened through:
$ gn args out/Debug
After you close the editor gn gen out/Debug will run automatically. Now you can dramatically increase the number of parallel tasks:
$ ninja -C out/Debug -j 1000
Official builds for android generate a signed Mojo Shell intended for distribution. You normally should not need to produce one. If you have any questions, reach out to etiennej@chromium.org.
Devtools mojo_run is a universal shell runner abstracting away the differences between running on Linux and Android.
Having built Mojo as described above, a demo app can be run as follows:
mojo/devtools/common/mojo_run https://core.mojoapps.io/spinning_cube.mojo # Linux mojo/devtools/common/mojo_run https://core.mojoapps.io/spinning_cube.mojo --android # Android
Whenever mojo_run is run, a development server is set up according to the config file. The server runs on your machine, serving the locally built apps, but appears to the shell under the https://core.mojoapps.io host.
You can ignore the config file and skip spawning the local server (for example, in order to use apps at the actual https://core.mojoapps.io web host) by passing --no-config-file to mojo_run.
Some applications can be run directly from the source tree. The development server serves the src directory, allowing to refer to these apps. For instance, this command serves a dart Mojo app from the source at examples/dart/device_info/main.dart:
mojo/devtools/common/mojo_run https://core.mojoapps.io/examples/dart/device_info/lib/main.dart [--android]
Some applications are meant to be run embedded in a window manager. To run these, you can pass the app url using the --embed flag. This will run the window manager and pass the given url to it:
mojo/devtools/common/mojo_run --embed https://core.mojoapps.io/ganesh_app.mojo [--android]
By default, mojo_run uses mojo:kiosk_wm as the window manager. You can pass a different window manager url using the --window-manager flag to override this.
For additional information on mojo_run refer to the built-in help and the documentation. You can also request more information on what the tool is doing for you by passing the --verbose flag.
Devtools mojo_debug allows you to interactively inspect a running shell, collect performance traces and attach a gdb debugger.
For additional information refer to the built-in help and the documentation.
For the Android tooling to work, you will need to have adb in your PATH. For that, you can either run:
source build/android/envsetup.sh
each time you open a fresh terminal, or add something like:
export PATH="$PATH":$MOJO_DIR/src/third_party/android_tools/sdk/platform-tools
to your ~/.bashrc file, $MOJO_DIR being a path to your Mojo checkout.
The device has to be running Android 5.0 (Lollipop) or newer.
Many features useful for development (ie. streaming of the shell stdout when running shell on the device) will not work unless the device is rooted and running a userdebug build. For Googlers, follow the instructions at this link.
If you wish to, you can also run the Linux Mojo shell directly with no wrappers:
./out/Debug/mojo_shell out/Debug/spinning_cube.mojo
With git you should make all your changes in a local branch. Once your change is committed, you can delete this branch.
Create a local branch named “mywork” and make changes to it.
cd src git new-branch mywork vi ...
Commit your change locally. (this doesn't commit your change to the SVN or Git server)
git commit -a
Fix your source code formatting.
$ git cl format
Upload your change for review.
$ git cl upload
Respond to review comments.
See Contributing code for more detailed git instructions, including how to update your CL when you get review comments. There's a short tutorial that might be helpful to try before making your first change: C++ in Chromium 101.
To land a change after receiving LGTM:
$ git cl land
Our waterfall continuously builds and tests the code. Don't break the build!
One of the bots, Mojo Linux Perf runs a suite of benchmarks and uploads the results to the performance dashboard. You can browse the results here, putting mojo_benchmarks as the “test suite”.
Automated alerts about performance regressions are sent to mojo-perf-alerts@chromium.org.
For examples of interesting sets of graphs see: