e64620208f | ||
---|---|---|
.github | ||
example-configs | ||
src | ||
tests | ||
tools/zsh | ||
.gitignore | ||
.prettierignore | ||
.prettierrc.json | ||
LICENSE | ||
README.md | ||
jest.fast.json | ||
jest.slow.json | ||
package.json | ||
yarn.lock |
README.md
Electron Build Tools
This repository contains helper/wrapper scripts to make building Electron easier.
Installation
A handful of prerequisites, such as git, python, and npm, are
required for building Electron itself; these can be found in
Platform Prerequisites. npm
can be used
with build-tools
itself as well, but we've configured it to run
with yarn
, so we also recommend you install it to your system.
From here, you'll need a command-line prompt. On Mac and Linux, this will be a terminal with a shell, e.g. bash or zsh. You can also use these on Windows if you install them, or use built-in tools like Windows' Command Prompt.
Please note that build-tools
(due to nested dependencies) might not work properly in powershell, please use cmd
on Windows for optimum results.
# Install build-tools package globally:
npm i -g @electron/build-tools
Getting the Code and Building Electron
You can run a new Electron build with this command:
# The 'Hello, World!' of build-tools: get and build `main`
# Choose the directory where Electron's source and build files will reside.
# You can specify any path you like; this command defaults to `$PWD/electron`.
# If you're going to use multiple branches, you may want something like:
# `--root=~/electron/branch` (e.g. `~/electron-gn/main`)
e init --root=~/electron --bootstrap testing
That command's going to run for awhile. While you're waiting, grab a cup of hot caffeine and read about what your computer is doing:
Concepts
Electron's build-tools command is named e
. Like nvm and git,
you'll invoke e
with commands and subcommands. See e --help
or e help <cmd>
for many more details.
e
also borrows another inspiration from nvm: having multiple configurations
that you can switch between so that one is the current, active configuration.
Many choices go into an Electron build:
- Which Electron branch
is used (e.g.
main
,13-x-y
) - Which .gn config file is imported (e.g. testing or release)
- Any compile-time options (e.g. Clang's asan or tsan)
e
holds all these variables together in a build configuration. You can
have multiple build configurations and manage them in a way similar to nvm:
nvm | e | Description |
---|---|---|
nvm ls | e show configs | Show the available configurations |
nvm current | e show current | Show which configuration is currently in use |
nvm use <name> | e use <name> | Change which configuration is currently in use |
Getting the source code is a lot more than cloning electron/electron
.
Electron is built on top of Chromium (with Electron patches) and Node
(with more Electron patches). A source tree needs to have all of the
above and for their versions to be in sync with each other. Electron
uses Chromium's Depot Tools and GN for wrangling
and building the code. e
wraps these tools:
Command | Description |
---|---|
e init | Create a new build config and initialize a GN directory |
e sync | Get / update / synchronize source code branches |
e build | Build it! |
e init
e init
initializes a new local development environment for Electron.
To see all potential options for this command, run:
$ e init --help
New build configs are created with e init
. It has several command-line
options to specify the build configuration, e.g. the path to the source
code, compile-time options, and so on. See e init --help
for in-depth
details.
Each build config has a name, chosen by you to use as a mnemonic when
switching between build configs with e use <name>
. This is the name's
only purpose, so choose whatever you find easiest to work with —
whether it's electron
, 6-1-x--testing
, or chocolate-onion-popsicle
.
Each build also needs a root directory. All the source code and built
files will be stored somewhere beneath it. e init
uses $PWD/electron
by default, but you can choose your own with --root=/some/path
. If you
want to make multiple build types of the same branch, you can reuse
an existing root to share it between build configs.
As an example, let's say you're starting from scratch and want both
testing and release builds of the main branch in electron/electron
.
You might do this:
# making 'release' and 'testing' builds from main
$ e init main-testing -i testing --root=~/src/electron
Creating '~/src/electron'
New build config 'main-testing' created
Now using config 'main-testing'
$ e show current
main-testing
$ e init main-release -i release --root=~/src/electron
INFO Root '~/src/electron' already exists.
INFO (OK if you are sharing $root between multiple build configs)
New build config 'main-release' created
Now using config 'main-release'
$ e show configs
* main-release
main-testing
$ e show current
main-release
$ e show root
~/src/electron
$ e use main-testing
Now using config 'main-testing'
$ e show current
main-testing
$ e show root
~/src/electron
As a convenience, e init --bootstrap
will run e sync
and e build
after creating the build config. Let's see what those do:
e sync
To see all potential options for this command, run:
$ e sync --help
e sync
is a wrapper around gclient sync
from Depot Tools.
If you're starting from scratch, this will (slowly) fetch all the source
code. It's also useful after switching Electron branches to synchronize
the rest of the sources to the versions needed by the new Electron branch.
e sync
is usually all you need. Any extra args are passed along to gclient itself.
$ e show current
main-testing
$ e show root
~/src/electron
$ e sync
Running "gclient sync --with_branch_heads --with_tags" in '~/src/electron/src'
[sync output omitted]
To make your output more verbose, you can add an increasing number of -v
s. For example,
# basic verbosity
$ e sync -v
Running "gclient sync --with_branch_heads --with_tags -v" in '~/src/electron/src'
[sync output omitted]
# significant verbosity
$ e sync -vvvv
Running "gclient sync --with_branch_heads --with_tags -vvvv" in '~/src/electron/src'
[sync output omitted]
e build
e build
builds an Electron executable.
To see all potential options for this command, run:
$ e build --help
Once you have the source, the next step is to build it with e build [target]
.
These build targets are supported:
Target | Description |
---|---|
breakpad | Builds the breakpad dump_syms binary |
chromedriver | Builds the chromedriver binary |
electron | Builds the Electron binary (Default) |
electron:dist | Builds the Electron binary and generates a dist zip file |
mksnapshot | Builds the mksnapshot binary |
node:headers | Builds the node headers .tar.gz file |
As with syncing, e build [target]
is usually all you need. Any extra
args are passed along to ninja, so for example e build -v
runs a verbose build.
Using Electron
After you've built Electron, it's time to use it!
Command | Description |
---|---|
e start | Run the Electron build |
e node | Run the Electron build as Node |
e debug | Run the Electron build in a debugger |
e test | Run Electron's spec runner |
As usual, any extra args are passed along to the executable. For example,
e node --version
will print out Electron's node version.
e debug
e debug
runs your local Electron build inside of lldb or gdb.
$ uname
Linux
$ e debug
Reading symbols from /home/yourname/electron/gn/main/src/out/Testing/electron...
(gdb)
$ uname
Darwin
$ e debug
target create "/Users/yourname/electron-gn/src/out/Testing/Electron.app/Contents/MacOS/Electron"
(lldb)
e test
e test
starts the local Electron build's test runner. Any extra args are passed
along to the runner.
To see all potential options for this command, run:
$ e test --help
Example:
# run all tests
e test
# run main process tests
e test --runners=main
Possible extra arguments to pass:
--node
- Run Node.js' own tests with Electron inRUN_AS_NODE
mode.--runners=<main|remote|native>
- The set of tests to run, can be eithermain
,remote
, ornative
.
e show
e show
shows information about the current build config.
To see all potential options for this command, run:
$ e show --help
Command | Description |
---|---|
e show current | The name of the active build config |
e show configs | Lists all build configs |
e show env | Show environment variables injected by the active build config |
e show exe | The path of the built Electron executable |
e show root | The path of the root directory from e init --root . |
e show src [name] | The path of the named (default: electron) source dir |
e show stats | Build statistics |
Example usage:
$ uname
Darwin
$ e show exe
/Users/username/electron-gn-root/src/out/Testing/Electron.app/Contents/MacOS/Electron
$ uname
Linux
$ e show exe
/home/username/electron-gn-root/src/out/Testing/electron
$ e show out
Testing
$ e show src
/home/username/electron-gn-root/src/electron
$ cd `e show src base` && pwd
/home/username/electron-gn-root/src/base
$ ripgrep --t h TakeHeapSnapshot `e show src`
e remove <name>
e remove|rm <name>
removes a build config from the list.
e open <commit | issue | PR>
e open
opens the GitHub page for the specified commit, pull request, or issue.
To see all potential options for this command, run:
$ e open --help
For example, e open 0920d01
will find the commit with an abbreviated
sha1 of 0920d01
, see that it's associated with pull request #23450,
and open https://github.com/electron/electron/pull/23450 in your browser.
Since you can pass in a pull request or issue number as well,
e open 23450
would have the same effect.
e patches [patch-dir]
e patches
exports patches to the specified patch directory in Electron source tree.
To see all potential options for this command, run:
$ e patches --help
Valid patch directories can include:
node
v8
boringssl
chromium
perfetto
icu
Command | Source Directory | Patch Directory |
---|---|---|
e patches node | src/third_party/electron_node |
src/electron/patches/node |
e patches chromium | src |
src/electron/patches/chromium |
e patches boringssl | src/third_party/boringssl/src |
src/electron/patches/boringssl |
e patches v8 | src/v8 |
src/electron/patches/v8 |
e patches perfetto | src/third_party/perfetto |
src/electron/patches/perfetto |
e patches icu | src/third_party/icu |
src/electron/patches/icu |
e sanitize-config <name>
e sanitize-config
updates and/or overwrites an existing config to conform to latest build-tools
updates.
To see all potential options for this command, run:
$ e sanitize-config --help
Sometimes build-tools
will make updates to its config requirements. In these events warnings will be output to console to inform you that build-tools
has temporarily handled the issues. You can make these warnings go away either by manually updating your config files or by running this command to automatically overwrite the existing configs to update formatting.
Common Usage
Building a Specific Electron Version
e init
checks out the HEAD of the main branch. To build against a specific version of Electron, checkout that version with these commands:
# Change working directory to the Electron source directory
cd `e show src`
# Checkout the desired Electron version (in this case, 11.0.0)
git checkout tags/v11.0.0 -b v11.0.0
# Sync dependencies with the current branch
e sync
# Build Electron
e build
Advanced Usage
Per-Session Active Configs
If you want your shell sessions to each have different active configs, try this in your ~/.profile
or ~/.zshrc
or ~/.bashrc
:
export EVM_CURRENT_FILE="$(mktemp --tmpdir evm-current.XXXXXXXX.txt)"
This will create per-shell temporary files in which the active config file can be changed with e use
.
Disabling Automatic Updates
With the default configuration, build-tools will automatically check for updates every 4 hours.
You can enable and disable these automatic updates with the following commands:
e auto-update enable
e auto-update disable
Regardless of whether automatic updates are enabled, you can manually call the following command to immediately trigger an update.
e auto-update check