Brave Developer tips
Here are some tips for developing Brave. Add yours here as well!
brave-core follows the Chromium style guide. In short:
- variable_names
- member_variable_names_
- FunctionNames()
- kEnumValues
- MACRO_NAMES
- ClassNames
For each function used in a .cc file, add the necessary .h file, but don’t add any additional .h files. If a .h file is needed to avoid an incorrect re-definition in a chromium_src override file, then add a comment to that effect. https://google.github.io/styleguide/cppguide.html#Include_What_You_Use
Minimize the length of #define definitions by factoring out functions or methods wherever possible.
Don’t squash commits after reviews. That way your review can see intermediate fixes. Only squash before merging.
Building Android works great on Linux and Windows with WSL2 from the command line. I commonly use this to create the binaries:
npm run create_dist -- Release --channel=nightly --target_os=android --target_arch=arm --target_android_output_format=apk
This may require environment variables being set if the signing step is done (BRAVE_ANDROID_KEYSTORE_PATH, BRAVE_ANDROID_KEY_PASSWORD, BRAVE_ANDROID_KEYSTORE_PASSWORD, BRAVE_ANDROID_KEYSTORE_NAME). However, you should still be able to get an APK without the signing taking place.
Once you have the APK, you can easily sideload it onto your phone using an app like AirDroid. This will require that you have the developer mode enabled on your phone.
Building and running Android seems to only work on Linux (not macOS) and so far only on x86 (not x64). Use Android Studio > Device Manager > Create Device. When creating a device, choose an x86 image (not an x86_64).
Then use
npm run build -- --target_os=android --target_arch=x86 --target_android_output_format=apk
adb install ./src/out/android_Component_x86/apks/Bravex86.apk
Alternatively, after building the APK, you can just drag and drop the APK to the running emulator as well.
You should not be using InProcessBrowserTest for anything that is (or will be) available on Android. That should include just about everything. See https://bravesoftware.slack.com/archives/C03MMJ4CHEZ/p1671052432003879?thread_ts=1671050489.910559&cid=C03MMJ4CHEZ for more information.
To run a browser test, you can use
npm run test -- brave_browser_tests --filter="*BraveScreenFarblingBrowserTest*"
to run an Android test, instead use
npm run test -- brave_unit_tests --target_os=android --target_arch=x86 --manual_android_test_device
npm run test -- brave_browser_tests --target_os=android --target_arch=x86 --manual_android_test_device
On Linux, we run upstream tests as well in CI which might sometimes fail and need replicating:
npm run test -- unit_tests --filter=whatever
First try updating everything:
git checkout master # in src/brave
git pull origin master # in src/brave
git pull origin master # in src/
# then sync and build, run the following in the top-level directory (above src/)
npm run sync && cd src/brave && npm run build
# once this works, try rebasing your feature branch on master and checking it out and building
Brave employees can check for the binaries created by PR builder in the #browser-artifacts-bot Slack channel.
Alternatively, you can go into the Jenkins job for the platform (linked on the PR) and then check at the end of the job, the URL created where the binaries are uploaded to. Via the Jenkins interface, you can also re-queue the build and set release channel and signing options. This is super helpful for testing scenarios requiring a certain channel or scenarios requiring signed binaries.
For Windows installs, there are command line options you can pass to the generated installer which will execute the system install. For more information about that, please check the Brave omaha wiki page. These same options can be passed to the brave_installer.exe that is created if you were to run npm run create_dist.