DEPS: depot_tools/cipd, recipe_engine/context, recipe_engine/file, recipe_engine/path, recipe_engine/platform, recipe_engine/python, recipe_engine/step, recipe_engine/url
Functions to work with Miniconda python environment.
See http://conda.pydata.org/miniconda.html
— def install(self, version, path):
Downloads Miniconda installer for given version and executes it.
Args: version: version of Miniconda to install, e.g. ‘Miniconda2-3.18.3’. path: prefix to install Miniconda into.
Returns: Instance of CondaEnv, that also optionally acts as context manager that deletes the environment on exit.
DEPS: depot_tools/bot_update, depot_tools/gclient, depot_tools/git, depot_tools/presubmit, recipe_engine/buildbucket, recipe_engine/context, recipe_engine/file, recipe_engine/path, recipe_engine/python, recipe_engine/runtime
Stateless API for using public infra gclient checkout.
— def checkout(self, gclient_config_name, patch_root=None, path=None, internal=False, named_cache=None, **kwargs):
Fetches infra gclient checkout into a given path OR named_cache.
Arguments:
internal argument value. Note: your cr-buildbucket.cfg should specify named_cache for swarming to prioritize bots which actually have this cache populated by prior runs. Otherwise, using named cache isn't particularly useful, unless your pool of builders is very small.Returns: a Checkout object with commands for common actions on infra checkout.
DEPS: depot_tools/cipd, recipe_engine/buildbucket, recipe_engine/context, recipe_engine/json, recipe_engine/python, recipe_engine/runtime, recipe_engine/step
API for building packages defined in infra's public and intenral repos.
Essentially a shim around scripts in https://chromium.googlesource.com/infra/infra.git/+/master/build/ and its internal counterpart.
— def build(self):
Builds packages.
@contextlib.contextmanager
— def context(self, path_to_repo, goos=None, goarch=None):
Sets context building CIPD packages.
Arguments: path_to_repo (path): path infra or infra_internal repo root dir. Expects to find build/build.py inside provided dir. goos, goarch (str): allows for setting GOOS and GOARCH for cross-compiling Go code.
Doesn't support nesting.
— def tags(self, git_repo_url, revision):
Returns tags to be attached to uploaded CIPD packages.
— def test(self, skip_if_cross_compiling=False):
Tests previously built packages integrity.
— def upload(self, tags, step_test_data=None):
Uploads previously built packages.
DEPS: recipe_engine/context, recipe_engine/path, recipe_engine/platform
API for interacting with a provisioned infrastructure system.
@property
— def sys_bin_path(self):
@contextlib.contextmanager
— def system_env(self):
Yields a context modified to operate on system paths.
DEPS: recipe_engine/raw_io, recipe_engine/url
APIs for interacting with omahaproxy.
— def history(self, min_major_version=None, exclude_platforms=None):
@staticmethod
— def split_version(text):
DEPS: build/luci_config, depot_tools/git, depot_tools/git_cl, depot_tools/gsutil, recipe_engine/context, recipe_engine/file, recipe_engine/json, recipe_engine/path, recipe_engine/python, recipe_engine/raw_io, recipe_engine/step, recipe_engine/time
— def roll_projects(self, projects):
Attempts to roll each project from the provided list.
If rolling any of the projects leads to failures, other projects are not affected.
DEPS: depot_tools/git, depot_tools/osx_sdk, depot_tools/windows_sdk, recipe_engine/archive, recipe_engine/buildbucket, recipe_engine/cipd, recipe_engine/context, recipe_engine/file, recipe_engine/json, recipe_engine/path, recipe_engine/platform, recipe_engine/python, recipe_engine/raw_io, recipe_engine/runtime, recipe_engine/step
Allows uniform cross-compiliation, version tracking and archival for third-party software packages (libs+tools) for distribution via CIPD.
The purpose of the Third Party Packages (3pp) recipe/module is to generate CIPD packages of statically-compiled software for distribution in our continuous integration fleets, as well as software distributed to our develepers (e.g. via depot_tools).
Target os and architecture uses the CIPD “${os}-${arch}” (a.k.a. “${platform}”) nomenclature, which is currently defined in terms of Go's GOOS and GOARCH runtime variables (with the unfortunate exception that CIPD uses ‘mac’ instead of ‘darwin’). This is somewhat arbitrary, but has worked well so far for us.
The 3pp module loads package definitions from a folder containing subfolders. Each subfolder defines a single software package to fetch, build and upload. For example, you might have a folder in your repo like this:
my_repo.git/
3pp/ # "root folder"
.vpython # common vpython file for all package scripts
zlib/ # zlib "package folder"
3pp.pb # REQUIRED: the Spec.proto definition for zlib
install.sh # a script to build zlib from source
extra_resource_file
other_package/
3pp.pb # REQUIRED
fetch.py # a script to fetch `other_package` in a custom way
install.sh
install-win.sh # windows-specific build script
...
This defines two packages (zlib, and other_package). The 3pp.pb files have references to the fetch/build scripts, and describe what dependencies the packages have (if any).
NOTE: Only one layer of package folders is supported currently.
Packages are named by the folder that contains their definition file (3pp.pb) and build scripts. It's preferable to have package named after software that it contains. However, sometimes you want multiple major versions of the software to exist side-by-side (e.g. pcre and pcre2, python and python3, etc.). In this case, have two separate package definition folders.
Each package folder contains a package spec (3pp.pb), as well as scripts, patches and/or utility tools to build the software from source.
The spec is a Text Proto document specified by the spec.proto schema.
The spec is broken up into two main sections, “create” and “upload”. The create section allows you to specify how the package software gets created, and allows specifying differences in how it's fetched/built/tested on a per-target basis, and the upload section has some details on how the final result gets uploaded to CIPD.
The 3pp.pb spec begins with a series of create messages, each with details on on how to fetch+build+test the package. Each create message contains a “platform_re” field which works as a regex on the ${platform} value. All matching patterns apply in order, and non-matching patterns are skipped. Each create message is applied with a dict.update for each member message (i.e. [‘source’].update, [‘build’].update, etc.) to build a singular create message for the current target platform. For list values (e.g. ‘tool’, ‘dep’ in the Build message), you can clear them by providing a new empty value (e.g. tool: "")
Once all the create messages are merged (see schema for all keys that can be present), the actual creation takes place.
Note that “source” is REQUIRED in the final merged instruction set. All other messages are optional and have defaults as documented in spec.proto.
The creation process is broken up into 4 different stages:
All scripts described below are invoked with the following environment variables set:
patch_version set for the version we're building (if any patch version was set).Additionally, on cross-compile environments, the $CROSS_TRIPLE environment variable is set to a GCC cross compile target triplet of cpu-vendor-os.
The source is used to fetch the raw sources for assembling the package. In some cases the sources may actually be binary artifacts (e.g. prebuilt windows installers).
The source is unpacked to a checkout directory, possibly in some specified subdirectory. Sources can either produce the actual source files, or they can produce a single archive file (e.g. zip or tarball), which can be unpacked with the ‘unpack_archive’ option. In addition, patches can be applied to the source with the ‘patch_dir’ option (the patches should be in git format-patch format, and will be applied with git apply).
git - This checks out a semver tag in the repo.cipd - This fetches data from a CIPD package.script - Used for “weird” packages which are distributed via e.g. an HTML download page or an API. The script must be able to return the ‘latest’ version of its source, as well as to actually fetch a specified version. Python fetch scripts will be executed with vpython, and so may have a .vpython file (or similar) in the usual manner to pull in dependencies like requests.Additionally the Source message contains a patch_version field to allow symver disambiguation of the built packages when they contain patches or other alterations which need to be versioned. This string will be joined with a ‘.’ to the source version being built when uploading the result to CIPD.
The build message allows you to specify deps, and tools, as well as a script install which contains your logic to transform the source into the result package.
Deps are libraries built for the target ${platform} and are typically used for linking your package.
Tools are binaries built for the host; they‘re things like automake or sed that are used during the configure/make phase of your build, but aren’t linked into the built product. These tools will be available on $PATH (both ‘$tools’ and ‘$tools/bin’ are added to $PATH, because many packages are set up with their binaries at the base of the package, and some are set up with them in a /bin folder)
Installation occurs by invoking the script indicated by the ‘install’ field (with the appropriate interpreter, depending on the file extension) like:
<interpreter> "$install[*]" "$PREFIX" "$DEPS_PREFIX"
Where:
$install[*] are all of the tokens in the ‘install’ field.$PREFIX is the directory which the script should install everything to; this directory will be archived into CIPD verbatim.$DEPS_PREFIX is the path to a prefix directory containing the union of all of your packages' transitive deps. For example, all of the headers of your deps are located at $DEPS_PREFIX/include.tools are in $PATHIf the ‘install’ script is omitted, it is assumed to be ‘install.sh’.
If the ENTIRE build message is omitted, no build takes place. Instead the result of the ‘source’ stage will be packaged.
During the execution of the build phase, the entire ‘root folder’ is copied into the source checkout in the .3pp directory, and the script will be invoked as /path/to/checkout/.3pp/$package_name/$script_name. Because the entire root folder is copied, you can have shared resources (like .vpython files or helper scripts) which are common to all package definitions and located relative to the install script.
Once the build stage is complete, all files in the $PREFIX folder passed to the install script will be zipped into a CIPD package.
It is strongly recommended that if your package is a library or tool with many files that it be packaged in the standard POSIXey PREFIX format (e.g. bin, lib, include, etc.). If your package is a collection of one or more standalone binaries, it's permissible to just have the binaries in the root of the output $PREFIX.
If the build stage is skipped (i.e. the build message is omitted) then the output of the source stage will be packaged instead (this is mostly useful when using a ‘script’ source).
After the package is built it can be optionally tested. The recipe will run your test script in an empty directory with the path to the packaged-but-not-yet-uploaded cipd package file and it can do whatever testing it needs to it (exiting non-zero if something is wrong). You can use the cipd pkg-deploy command to deploy it (or whatever cipd commands you like, though I wouldn't recommend uploading it to CIPD, as the 3pp recipe will do that after the test exits 0).
Additionally, vpython for the tool platform will be guaranteed to be in $PATH.
Once the test comes back positive, the CIPD package will be uploaded to the CIPD server and registered with the prefix indicated in the upload message. The full CIPD package name is constructed as:
<prefix>/<pkg_name>/${platform}
So for example with the prefix infra, the bzip2 package on linux-amd64 would be uploaded to infra/bzip2/linux-amd64 and tagged with the version that was built (e.g. version:1.2.3.patch_version1).
You can also mark the upload as a universal package, which will:
${platform} suffix from the upload namelinux-amd64, regardless of what platform you build the recipe on. This was chosen arbitrarially to ensure that “universal” packages build consistently. You can override this behavior (and bypass the normal docker environment entirely) by setting the no_docker_env flag to true in your Create.Build message.Every package will try to build the latest identifiable semver of its source, or will attempt to build the semver requested as an input property to the 3pp recipe. This semver is also used to tag the uploaded artifacts in CIPD.
Because some of the packages here are used as dependencies for others (e.g. curl and zlib are dependencies of git, and zlib is a dependency of curl), each package used as a dependency for others should specify its version explicitly (currently this is only possible to do with the ‘cipd’ source type). So e.g. zlib and curl specify their source versions, but git and python float at ‘head’, always building the latest tagged version fetched from git.
When building a floating package (e.g. python, git) you may explicitly state the symver that you wish to build as part of the recipe invocation.
The symver of a package (either specified in the package definition, in the recipe properties or discovered while fetching its source code (e.g. latest git tag)) is also used to tag the package when it's uploaded to CIPD (plus the patch_version in the source message).
Third party packages are currently compiled on linux using the ‘infra.tools.dockerbuild’ tool from the infra.git repo. This uses a slightly modified version of the dockcross Docker cross-compile environment. Windows and OS X targets are built using the ‘osx_sdk’ and ‘windows_sdk’ recipe modules, each of which provides a hermetic (native) build toolchain for those platforms.
For linux, we can support all the architectures implied by dockerbuild, including:
If the recipe is run with force_build it will always build all packages indicated (and their dependencies), and will not upload any of them to the central server.
The recipe must always be run with a package_prefix (by assigning to the .package_prefix property on the Support3ppApi). If the recipe is run in experimental mode, ‘experimental/’ will be prepended to this. Additionally, you may specify experimental: true in the Create message for a package, which will have the same effect when running the recipe in production (to allow adding new packages or package/platform combintations experimentally).
As an example of the package definition layout in action, take a look at the 3pp folder in this infra.git repo.
This module uses the following named caches:
3pp_cipd - Caches all downloaded and uploaded CIPD packages. Currently tag lookups are performed every time against the CIPD server, but this will hold the actual package files.osx_sdk - Cache for depot_tools/osx_sdk. Only on Mac.windows_sdk - Cache for depot_tools/windows_sdk. Only on Windows.— def ensure_uploaded(self, packages=(), platform='', force_build=False):
Executes entire {fetch,build,package,verify,upload} pipeline for all the packages listed, targeting the given platform.
Args:
Returns (list[(cipd_pkg, cipd_version)], set[str]) of built CIPD packages and their tagged versions, as well as a list of unsupported packages.
— def initialize(self):
— def load_packages_from_path(self, path):
Loads all package definitions from the given path.
This will parse and intern all the 3pp.pb package definition files so that packages can be identified by their name. For example, if you pass:
path/ pkgname/ 3pp.pb install.sh
This would parse path/pkgname/3pp.pb and register the “pkgname” package.
Args:
Returns a set(str) containing the names of the packages which were loaded.
Raises a DuplicatePackage exception if this function encounters a package whose name is already registered. This could occur if you call load_packages_from_path multiple times, and one of the later calls tries to load a package which was registered under one of the earlier calls.
— def package_prefix(self, experimental=False):
Returns the CIPD package name prefix (str), if any is set.
This will prepend ‘experimental/’ to the currently set prefix if:
— def set_package_prefix(self, prefix):
Set the CIPD package name prefix (str).
All CIPDSpecs for built packages (not sources) will have this string prepended to them.
DEPS: depot_tools/bot_update, depot_tools/gclient, depot_tools/git, depot_tools/infra_paths, recipe_engine/buildbucket, recipe_engine/context, recipe_engine/file, recipe_engine/json, recipe_engine/path, recipe_engine/properties, recipe_engine/raw_io, recipe_engine/runtime, recipe_engine/step
— def __call__(self, source, source_repo_checkout_name, dest, source_ref=‘refs/heads/master’, dest_ref=‘refs/heads/master’, extra_submodules=None, deps_path_prefix=None, disable_path_prefix=False):
Args: source: URL of the git repository to mirror. source_repo_checkout_name: Name of the directory that the source repo should be checked out into. dest: URL of the git repository to push to. source_ref: git ref in the source repository to checkout. dest_ref: git ref in the destination repository to push to. extra_submodules: a list of “path=URL” strings. These are added as extra submodules. deps_path_prefix: path prefix used to filter out DEPS. DEPS with the prefix are included. disable_path_prefix: disable filtering out DEPS by path prefix.
DEPS: depot_tools/cipd, depot_tools/git, depot_tools/gitiles, windows_sdk, recipe_engine/context, recipe_engine/file, recipe_engine/json, recipe_engine/path, recipe_engine/platform, recipe_engine/properties, recipe_engine/python, recipe_engine/raw_io, recipe_engine/step, recipe_engine/url
— def build_package(self, name, workdir, root, install_mode):
@property
— def cmake(self):
@property
— def cross_platform(self):
@property
— def dep(self):
— def does_package_exist(self, name, version):
@dry_run.setter
— def dry_run(self, v):
— def ensure_package(self, workdir, repo_url, package_name_prefix, install_fn, tag, version, cipd_install_mode, test_fn=None):
Ensures that the specified CIPD package exists.
@property
— def firebase(self):
@property
— def gcloud(self):
— def get_latest_release_tag(self, repo_url, prefix=‘v’):
— def get_package_name(self, package_name_prefix):
@property
— def git(self):
@property
— def go(self):
@property
— def gsutil(self):
— def init_cross_platform(self, cross_platform):
@property
— def ninja(self):
@property
— def python(self):
— def register_package(self, package_file, name, version):
— def support_prefix(self, base):
@property
— def swig(self):
DEPS: recipe_engine/cipd, recipe_engine/context, recipe_engine/json, recipe_engine/path, recipe_engine/platform, recipe_engine/step
API for using Windows SDK distributed via CIPD.
@contextmanager
— def __call__(self, path=None, version=None, enabled=True):
Setups the SDK environment when enabled.
Args: path (path): Path to a directory where to install the SDK (default is ‘[start_dir]/cipd/windows_sdk’) version (str): CIPD instance ID, tag or ref of the SDK (default is set via $infra/windows_sdk.version property) enabled (bool): Whether the SDK should be used or not.
Raises: StepFailure or InfraFailure.
DEPS: depot_tools/git, support_3pp, recipe_engine/file, recipe_engine/path, recipe_engine/properties, recipe_engine/python, recipe_engine/runtime, recipe_engine/step
This recipe builds and packages third party software, such as Git.
— def RunSteps(api, package_locations, to_build, platform, force_build, package_prefix):
DEPS: depot_tools/cipd, conda, recipe_engine/buildbucket, recipe_engine/file, recipe_engine/path, recipe_engine/platform, recipe_engine/properties
Recipe to build CIPD package with sealed Conda environment.
Supposed to be used from manually triggered Buildbot builders. We aren't expecting rebuilding this environment often, so setting up and periodic schedule is a waste of resources.
To build a new package for all platforms:
— def RunSteps(api):
DEPS: depot_tools/depot_tools, depot_tools/gsutil, recipe_engine/context, recipe_engine/file, recipe_engine/path, recipe_engine/platform, recipe_engine/properties, recipe_engine/python, recipe_engine/step
— def RunSteps(api):
DEPS: build/zip, depot_tools/cipd, depot_tools/gsutil, recipe_engine/file, recipe_engine/path, recipe_engine/platform, recipe_engine/properties, recipe_engine/python, recipe_engine/raw_io, recipe_engine/step
— def RunSteps(api):
DEPS: build/chromite, build/repo, depot_tools/gsutil, recipe_engine/context, recipe_engine/path, recipe_engine/platform, recipe_engine/properties, recipe_engine/python, recipe_engine/raw_io, recipe_engine/step, recipe_engine/tempfile
This recipe is used to flash a CrOS DUT on a Chromium bot.
This essentially calls out to the cros-sdk's flash tool (located at https://codesearch.chromium.org/chromium/src/third_party/chromite/cli/flash.py).
That tool however has a dependency on the cros SDK‘s chroot (see https://chromium.googlesource.com/chromiumos/docs/+/master/developer_guide.md#Create-a-chroot for more info). Though it’s often used in CrOS development, the chroot isn't found in Chromium development at all, and has very limited support on Chromium bots. Consequently, this recipe will take care of setting it up prior to flashing the DUT. The basic steps of this recipe are:
— def RunSteps(api):
DEPS: build/swarming_client, depot_tools/gitiles, depot_tools/gsutil, recipe_engine/buildbucket, recipe_engine/context, recipe_engine/json, recipe_engine/path, recipe_engine/platform, recipe_engine/properties, recipe_engine/python, recipe_engine/raw_io, recipe_engine/step, recipe_engine/tempfile
This recipe is used to keep Chrome's pools of CrOS DUTs up to date.
It does so by launching tasks into the DUT pools which flash the devices. When ran, this recipe aims to get a portion of a given pool on the latest CHROMEOS_LKGM version. It will never flash more than a third of the pool at a single time. This is to ensure the remainder of the pool is online for tests. Consequently, this recipe will need to be run multiple times to upgrade the entire pool.
This recipe is intended to run several times during MTV's off-peak hours. Its builder should be backed by a single thin Ubuntu VM, while the tasks it launches run the cros_flash recipe and run on DUT swarming bots.
— def RunSteps(api, swarming_server, swarming_pool, device_type, bb_host, random_seed):
— def get_bots_in_pool(api, swarming_server, pool, device_type):
Returns the list of bots that belong to the given pool.
This uses swarming.py's bot/list query, and returns the resulting bots.
— def get_closest_available_version(api, board, lkgm_base):
Returns the GS path of the latest image for the given board and lkgm.
This finds the first LATEST-$lkgm file in GS closest to the current lkgm. It'll decrement the lkgm until it finds one, up to 100 attempts. This logic is taken from: https://codesearch.chromium.org/chromium/src/third_party/chromite/cli/cros/cros_chrome_sdk.py?rcl=63924982b3fdaf3c313e0052fe0c07dae5e4628a&l=350
Once it finds a valid LATEST-$lkgm file, it returns its contents appended to the board's directory in the GS image bucket, which contains the images built for that board at that version. (eg: gs://chromeos-image-archive/kevin-full/R72-11244.0.0-rc2/)
Returns tuple of: The 5-digit manifest for the latest image. GS path for the latest image.
— def trigger_flash(api, bot, gs_image_path):
DEPS: build/zip, depot_tools/cipd, depot_tools/git, depot_tools/gsutil, recipe_engine/file, recipe_engine/json, recipe_engine/path, recipe_engine/platform, recipe_engine/properties, recipe_engine/raw_io, recipe_engine/step, recipe_engine/tempfile
Recipe to build windows depot_tools bootstrap zipfile.
— def RunSteps(api, revision):
DEPS: build/docker, depot_tools/bot_update, depot_tools/gclient, recipe_engine/file, recipe_engine/path, recipe_engine/properties, recipe_engine/raw_io, recipe_engine/runtime, recipe_engine/service_account, recipe_engine/step, recipe_engine/time
— def RunSteps(api):
DEPS: build/gae_sdk, build/zip, depot_tools/cipd, depot_tools/gsutil, recipe_engine/path, recipe_engine/python, recipe_engine/raw_io, recipe_engine/runtime, recipe_engine/step
— def RunSteps(api):
DEPS: depot_tools/bot_update, depot_tools/gclient, recipe_engine/buildbucket, recipe_engine/properties, recipe_engine/step
— def RunSteps(api):
DEPS: recipe_engine/context, recipe_engine/file, recipe_engine/path, recipe_engine/platform, recipe_engine/properties, recipe_engine/step, recipe_engine/time
Pushes a trivial CL to Gerrit to verify git authentication works on LUCI.
— def RunSteps(api):
DEPS: build/goma, recipe_engine/context, recipe_engine/file, recipe_engine/path, recipe_engine/platform, recipe_engine/properties, recipe_engine/step, recipe_engine/time
Compiles trivial C++ program using Goma.
Intended to be used as a very simple litmus test of Goma health on LUCI staging environment. Linux and OSX only.
— def RunSteps(api):
DEPS: depot_tools/gclient, recipe_engine/context, recipe_engine/path, recipe_engine/properties, recipe_engine/python
Runs git submodule daemon (gsubmodd) against a given source repo.
Intended to be called periodically (see CYCLE_TIME_SEC). Runs several iterations of the daemon and then quits so that recipe has a chance to resync the source code.
— def RunSteps(api, source_repo, target_repo, limit=‘‘, epoch=’’):
DEPS: depot_tools/gclient, recipe_engine/context, recipe_engine/file, recipe_engine/json, recipe_engine/path, recipe_engine/properties, recipe_engine/python, recipe_engine/step
Runs git subtree daemon (gsubtreed) against Chromium src repo.
Intended to be called periodically (see CYCLE_TIME_SEC). Runs several iteration of the daemon and then quits so that recipe has a chance to resync the source code.
— def RunSteps(api, target_repo, cycle_time_sec, max_error_count):
DEPS: depot_tools/depot_tools, depot_tools/gsutil, recipe_engine/file, recipe_engine/path, recipe_engine/platform, recipe_engine/properties, recipe_engine/python, recipe_engine/step, recipe_engine/time
Pushes a trivial CL to Gerrit to verify git authentication works on LUCI.
— def RunSteps(api):
DEPS: depot_tools/tryserver, infra_checkout, recipe_engine/buildbucket, recipe_engine/context, recipe_engine/json, recipe_engine/platform, recipe_engine/python, recipe_engine/runtime
— def RunSteps(api):
DEPS: infra_cipd, recipe_engine/buildbucket, recipe_engine/context, recipe_engine/json, recipe_engine/path, recipe_engine/properties, recipe_engine/python, recipe_engine/runtime, recipe_engine/step
— def RunSteps(api):
DEPS: depot_tools/bot_update, depot_tools/cipd, depot_tools/depot_tools, depot_tools/gclient, depot_tools/infra_paths, depot_tools/osx_sdk, infra_checkout, infra_cipd, infra_system, recipe_engine/buildbucket, recipe_engine/context, recipe_engine/file, recipe_engine/json, recipe_engine/path, recipe_engine/platform, recipe_engine/properties, recipe_engine/python, recipe_engine/runtime, recipe_engine/step
— def RunSteps(api):
— def build_main(api, checkout, buildername, project_name, repo_url, rev):
DEPS: depot_tools/bot_update, depot_tools/gclient, recipe_engine/context, recipe_engine/path, recipe_engine/properties, recipe_engine/python, recipe_engine/step
Builds and tests infra.git code.
Very dumb for now, with no side effects. Runs continuously on LUCI in both staging and prod environments.
— def RunSteps(api):
DEPS: depot_tools/bot_update, depot_tools/gclient, depot_tools/git, depot_tools/osx_sdk, infra_checkout, infra_system, recipe_engine/buildbucket, recipe_engine/context, recipe_engine/path, recipe_engine/platform, recipe_engine/properties, recipe_engine/python, recipe_engine/raw_io, recipe_engine/step
— def RunSteps(api):
DEPS: infra_system, recipe_engine/context, recipe_engine/platform, recipe_engine/step
— def RunSteps(api):
DEPS: depot_tools/bot_update, depot_tools/gclient, recipe_engine/cipd, recipe_engine/context, recipe_engine/path, recipe_engine/platform, recipe_engine/step
— def RunSteps(api):
DEPS: build/chromium_checkout, depot_tools/bot_update, depot_tools/gclient, depot_tools/git, depot_tools/gitiles, depot_tools/gsutil, recipe_engine/buildbucket, recipe_engine/context, recipe_engine/file, recipe_engine/path, recipe_engine/properties, recipe_engine/python, recipe_engine/raw_io, recipe_engine/runtime, recipe_engine/step
— def RunSteps(api, project, repo, ref, config, lkgr_status_gs_path, allowed_lag):
DEPS: depot_tools/bot_update, depot_tools/gclient, recipe_engine/context, recipe_engine/json, recipe_engine/path, recipe_engine/properties, recipe_engine/python
— def RunSteps(api):
DEPS: depot_tools/tryserver, infra_checkout, recipe_engine/buildbucket, recipe_engine/context, recipe_engine/json, recipe_engine/properties, recipe_engine/runtime
— def RunSteps(api):
DEPS: depot_tools/osx_sdk, depot_tools/tryserver, infra_checkout, recipe_engine/buildbucket, recipe_engine/context, recipe_engine/json, recipe_engine/platform, recipe_engine/properties, recipe_engine/runtime
— def RunSteps(api, GOARCH):
DEPS: depot_tools/bot_update, depot_tools/gclient, recipe_engine/buildbucket, recipe_engine/properties
— def RunSteps(api):
DEPS: depot_tools/bot_update, depot_tools/gclient, depot_tools/infra_paths, infra_checkout, recipe_engine/buildbucket, recipe_engine/json, recipe_engine/path, recipe_engine/platform, recipe_engine/properties, recipe_engine/python, recipe_engine/step
— def RunSteps(api):
DEPS: build/chromium, build/trigger, depot_tools/bot_update, depot_tools/depot_tools, depot_tools/gclient, depot_tools/git, depot_tools/gsutil, omahaproxy, recipe_engine/context, recipe_engine/file, recipe_engine/path, recipe_engine/platform, recipe_engine/properties, recipe_engine/python, recipe_engine/raw_io, recipe_engine/runtime, recipe_engine/scheduler, recipe_engine/step
— def RunSteps(api):
@contextlib.contextmanager
— def copytree_checkout(api):
@recipe_api.composite_step
— def export_lite_tarball(api, version):
@recipe_api.composite_step
— def export_nacl_tarball(api, version):
@recipe_api.composite_step
— def export_tarball(api, args, source, destination):
— def gsutil_upload(api, source, bucket, dest, args):
— def published_all_tarballs(version, ls_result):
— def published_full_tarball(version, ls_result):
— def published_lite_tarball(version, ls_result):
— def published_nacl_tarball(version, ls_result):
— def published_test_tarball(version, ls_result):
DEPS: build/luci_config, build/puppet_service_account, recipe_autoroller, recipe_engine/json, recipe_engine/properties, recipe_engine/raw_io, recipe_engine/step, recipe_engine/time
Rolls recipes.cfg dependencies for public projects.
— def RunSteps(api, projects, auth_token, service_account):
DEPS: depot_tools/cipd, recipe_engine/path, recipe_engine/properties, recipe_engine/step
— def RunSteps(api, recipe_bundler_pkg, recipe_bundler_vers, repo_specs, package_name_prefix, package_name_internal_prefix):
DEPS: build/luci_config, depot_tools/bot_update, depot_tools/gclient, depot_tools/git, depot_tools/tryserver, recipe_engine/context, recipe_engine/file, recipe_engine/json, recipe_engine/path, recipe_engine/properties, recipe_engine/python, recipe_engine/raw_io, recipe_engine/runtime, recipe_engine/service_account, recipe_engine/step
— def RunSteps(api, upstream_project, downstream_project):
DEPS: build/luci_config, build/puppet_service_account, depot_tools/bot_update, depot_tools/gclient, recipe_engine/buildbucket, recipe_engine/context, recipe_engine/file, recipe_engine/json, recipe_engine/path, recipe_engine/properties, recipe_engine/python, recipe_engine/step
A continuous builder which runs recipe tests.
— def RunSteps(api, project_under_test, auth_with_account):
DEPS: depot_tools/bot_update, depot_tools/cipd, depot_tools/gclient, recipe_engine/buildbucket, recipe_engine/path, recipe_engine/properties
— def RunSteps(api):
DEPS: build/puppet_service_account, depot_tools/bot_update, depot_tools/gclient, recipe_engine/context, recipe_engine/path, recipe_engine/properties, recipe_engine/python, recipe_engine/step
This recipe is used to execute Dataflow workflows.
If you want a workflow to run at regular intervals, you can configure a builder to run this recipe. Dataflow workflows run on an internal builder, so this step must be completed by a Google employee. See this change for an example: https://chrome-internal-review.googlesource.com/c/chrome/tools/build/+/412934
Builders configured with the name matching “dataflow-workflow-.*” will be automatically monitored for failures.
This recipe uses the dataflow-launcher service account. That account must have the permission to schedule a Dataflow job for your project.
— def RunSteps(api, workflow, job_name, gcp_project_id, num_workers, timeout):
DEPS: recipe_engine/buildbucket, recipe_engine/cipd, recipe_engine/json, recipe_engine/path, recipe_engine/platform, recipe_engine/properties, recipe_engine/step
— def RunSteps(api):
— def get_value(pairs, key):
Returns a the value for the given key in the given pairs.
Args: pairs: A list of {“key”: key, “value”: value} dicts. key: A key whose value to get. If the key appears more than once, only the first value is returned.
Returns: The value for the given key.
Raises: ValueError: If the key doesn't exist.
DEPS: recipe_engine/buildbucket, recipe_engine/cipd, recipe_engine/path, recipe_engine/platform, recipe_engine/properties, recipe_engine/step, recipe_engine/time, recipe_engine/url, recipe_engine/uuid
— def RunSteps(api):
— def normalize(s):
Normalizes a string for use in a resource label.
DEPS: support_3pp, recipe_engine/buildbucket, recipe_engine/cipd, recipe_engine/file, recipe_engine/path, recipe_engine/platform, recipe_engine/properties, recipe_engine/raw_io, recipe_engine/runtime, recipe_engine/step
— def RunSteps(api, GOOS, GOARCH, load_dupe):
DEPS: sync_submodules, recipe_engine/properties, recipe_engine/runtime
— def RunSteps(api, disable_path_prefix):
DEPS: depot_tools/cipd, depot_tools/gitiles, third_party_packages, recipe_engine/platform, recipe_engine/properties, recipe_engine/runtime, recipe_engine/step
This recipe builds and packages third party software, such as Git.
— def RunSteps(api, cross_platform):
DEPS: depot_tools/cipd, depot_tools/gitiles, third_party_packages, recipe_engine/platform, recipe_engine/properties, recipe_engine/url
Recipe for ‘cmake’ building.
During testing, it may be useful to focus on building CMake. This can be done by running this recipe module directly.
— def RunSteps(api, dry_run):
DEPS: depot_tools/cipd, depot_tools/gitiles, third_party_packages, recipe_engine/platform, recipe_engine/properties, recipe_engine/url
Recipe for Go dependency management tool building.
During testing, it may be useful to focus on building dep. This can be done by running this recipe module directly.
— def RunSteps(api, dry_run):
DEPS: depot_tools/cipd, third_party_packages, recipe_engine/platform, recipe_engine/properties, recipe_engine/url
Recipe for firebase-tools building.
During testing, it may be useful to focus on building the firebase-tools. This can be done by running this recipe directly.
— def RunSteps(api, dry_run):
DEPS: depot_tools/cipd, third_party_packages, recipe_engine/platform, recipe_engine/properties, recipe_engine/url
Recipe for Google Cloud SDK building.
During testing, it may be useful to focus on building the SDK. This can be done by running this recipe module directly.
— def RunSteps(api, platform_name, platform_bits, dry_run):
DEPS: depot_tools/cipd, depot_tools/gitiles, third_party_packages, recipe_engine/platform, recipe_engine/properties, recipe_engine/url
Recipe for ‘git’ building.
During testing, it may be useful to focus on building Git. This can be done by running this recipe module directly.
— def RunSteps(api, dry_run):
DEPS: depot_tools/cipd, third_party_packages, recipe_engine/platform, recipe_engine/properties, recipe_engine/url
Recipe for Go toolchain building.
During testing, it may be useful to focus on building Go. This can be done by running this recipe module directly.
— def RunSteps(api, platform_name, platform_bits, dry_run):
DEPS: depot_tools/cipd, third_party_packages, recipe_engine/platform, recipe_engine/properties, recipe_engine/url
Recipe for gsutil building.
During testing, it may be useful to focus on building the gsutil. This can be done by running this recipe module directly.
— def RunSteps(api, dry_run):
DEPS: depot_tools/cipd, depot_tools/gitiles, third_party_packages, recipe_engine/platform, recipe_engine/properties, recipe_engine/url
Recipe for ‘ninja’ building.
During testing, it may be useful to focus on building Ninja. This can be done by running this recipe module directly.
— def RunSteps(api, dry_run):
DEPS: depot_tools/cipd, depot_tools/gitiles, third_party_packages, recipe_engine/platform, recipe_engine/properties
Recipe for ‘python’ building.
During testing, it may be useful to focus on building Python. This can be done by running this recipe module directly.
— def RunSteps(api, dry_run, cross_platform):
DEPS: depot_tools/cipd, depot_tools/gitiles, third_party_packages, recipe_engine/platform, recipe_engine/properties, recipe_engine/url
Recipe for ‘swig’ building.
During testing, it may be useful to focus on building Swig. This can be done by running this recipe module directly.
— def RunSteps(api, dry_run):
DEPS: recipe_engine/properties, recipe_engine/tricium
Recipe for Tricium that outputs no results.
This is intended to be used as the default recipe for the tricium builder, which shouldn't actually be used in runs triggered by Tricium.
— def RunSteps(api):
DEPS: depot_tools/gclient, depot_tools/git, depot_tools/gitiles, recipe_engine/buildbucket, recipe_engine/context, recipe_engine/file, recipe_engine/json, recipe_engine/path, recipe_engine/properties, recipe_engine/raw_io, recipe_engine/step
— def RunSteps(api, source_repo, target_repo):
DEPS: windows_sdk, recipe_engine/platform, recipe_engine/properties, recipe_engine/step
— def RunSteps(api):
DEPS: build/chromium, depot_tools/bot_update, depot_tools/gclient, depot_tools/git, recipe_engine/json, recipe_engine/path, recipe_engine/properties, recipe_engine/python, recipe_engine/runtime
Exports commits in Chromium to the web-platform-tests repo.
This recipe runs the wpt-export script; it is expected to be run as a recurring job at a short interval. It creates pull requests on GitHub for Chromium commits that contain exportable changes, merges these pull requests.
See: //docs/testing/web_platform_tests.md (https://goo.gl/rSRGmZ)
— def RunSteps(api):
DEPS: build/chromium, depot_tools/bot_update, depot_tools/gclient, depot_tools/git, depot_tools/git_cl, recipe_engine/context, recipe_engine/file, recipe_engine/json, recipe_engine/path, recipe_engine/properties, recipe_engine/python, recipe_engine/runtime, recipe_engine/step
Imports changes from web-platform-tests into Chromium.
This recipe runs the wpt-import script. The import process involves first fetching the latest changes from web-platform-tests, then running the tests via try jobs and and updating any baselines and expectations, before committing to Chromium.
See: //docs/testing/web_platform_tests.md (https://goo.gl/rSRGmZ)
— def RunSteps(api):
— def git_cl_issue_link(api):
Runs a step which adds a link to the current CL if there is one.