Documentation/dev-bazel.txt

= Gerrit Code Review - Building with Bazel

[[installation]]
== Prerequisites

To build Gerrit from source, you need:

* A Linux or macOS system (Windows is not supported at this time)
* A JDK for Java 8|9|10|11|...
* Python 2 or 3
* Node.js
* link:https://docs.bazel.build/versions/master/install.html[Bazel] directly
or through link:https://github.com/bazelbuild/bazelisk[Bazelisk]
* Maven
* zip, unzip
* gcc

[[java]]
=== Java

[[java-10]]
==== Java 10 support

Java 10 (and newer) is supported through vanilla java toolchain
link:https://docs.bazel.build/versions/master/toolchains.html[Bazel option].
To build Gerrit with Java 10 and newer, specify vanilla java toolchain and
provide the path to JDK home:

```
  $ bazel build \
    --define=ABSOLUTE_JAVABASE=<path-to-java-10> \
    --host_javabase=@bazel_tools//tools/jdk:absolute_javabase \
    --host_java_toolchain=@bazel_tools//tools/jdk:toolchain_vanilla \
    --java_toolchain=@bazel_tools//tools/jdk:toolchain_vanilla \
    :release
```

To run the tests, `--javabase` option must be passed as well, because
bazel test runs the test using the target javabase:

```
  $ bazel test \
    --define=ABSOLUTE_JAVABASE=<path-to-java-10> \
    --javabase=@bazel_tools//tools/jdk:absolute_javabase \
    --host_javabase=@bazel_tools//tools/jdk:absolute_javabase \
    --host_java_toolchain=@bazel_tools//tools/jdk:toolchain_vanilla \
    --java_toolchain=@bazel_tools//tools/jdk:toolchain_vanilla \
    //...
```

To avoid passing all those options on every Bazel build invocation,
they could be added to ~/.bazelrc resource file:

```
$ cat << EOF > ~/.bazelrc
> build --define=ABSOLUTE_JAVABASE=<path-to-java-10>
> build --javabase=@bazel_tools//tools/jdk:absolute_javabase
> build --host_javabase=@bazel_tools//tools/jdk:absolute_javabase
> build --host_java_toolchain=@bazel_tools//tools/jdk:toolchain_vanilla
> build --java_toolchain=@bazel_tools//tools/jdk:toolchain_vanilla
> EOF
```

Now, invoking Bazel with just `bazel build :release` would include
all those options.

Note that the follow option must be added to `container.javaOptions`
in `$gerrit_site/etc/gerrit.config` to run Gerrit with Java 10|11|...:

```
[container]
  javaOptions = --add-opens=jdk.management/com.sun.management.internal=ALL-UNNAMED
```

[[java-9]]
==== Java 9 support

Java 9 is supported through alternative java toolchain
link:https://docs.bazel.build/versions/master/toolchains.html[Bazel option].
The Java 9 support is backwards compatible. Java 8 is still the default.
To build Gerrit with Java 9, specify JDK 9 java toolchain:

```
  $ bazel build \
      --host_java_toolchain=@bazel_tools//tools/jdk:toolchain_java9 \
      --java_toolchain=@bazel_tools//tools/jdk:toolchain_java9 \
      :release
```

Note that the follow option must be added to `container.javaOptions`
in `$gerrit_site/etc/gerrit.config` to run Gerrit with Java 9:

```
[container]
  javaOptions = --add-opens=jdk.management/com.sun.management.internal=ALL-UNNAMED
```

[[build]]
== Building on the Command Line

=== Gerrit Development WAR File

To build the Gerrit web application:

----
  bazel build gerrit
----

[NOTE]
PolyGerrit UI may require additional tools (such as npm). Please read
the polygerrit-ui/README.md for more info.

The output executable WAR will be placed in:

----
  bazel-bin/gerrit.war
----

[[release]]
=== Gerrit Release WAR File

To build the Gerrit web application that includes the PolyGerrit UI,
core plugins and documentation:

----
  bazel build release
----

The output executable WAR will be placed in:

----
  bazel-bin/release.war
----

=== Headless Mode

To build Gerrit in headless mode, i.e. without the PolyGerrit UI:
Web UI:

----
  bazel build headless
----

The output executable WAR will be placed in:

----
  bazel-bin/headless.war
----

=== Extension and Plugin API JAR Files

To build the extension, plugin and acceptance-framework JAR files:

----
  bazel build api
----

The output archive that contains Java binaries, Java sources and
Java docs will be placed in:

----
  bazel-bin/api.zip
----

Install {extension,plugin,acceptance-framework}-api to the local
maven repository:

----
  tools/maven/api.sh install
----

Install gerrit.war to the local maven repository:

----
  tools/maven/api.sh war_install
----

=== Plugins

----
  bazel build plugins:core
----

The output JAR files for individual plugins will be placed in:

----
  bazel-bin/plugins/<name>/<name>.jar
----

The JAR files will also be packaged in:

----
  bazel-bin/plugins/core.zip
----

To build a specific plugin:

----
  bazel build plugins/<name>
----

The output JAR file will be be placed in:

----
  bazel-bin/plugins/<name>/<name>.jar
----

Note that when building an individual plugin, the `core.zip` package
is not regenerated.

To build with all Error Prone warnings activated, run:

----
  bazel build --java_toolchain //tools:error_prone_warnings_toolchain //...
----


[[IDEs]]
== Using an IDE.

=== IntelliJ

The Gerrit build works with Bazel's link:https://ij.bazel.build[IntelliJ plugin].
Please follow the instructions on <<dev-intellij#,IntelliJ Setup>>.

=== Eclipse

==== Generating the Eclipse Project

Create the Eclipse project:

----
  tools/eclipse/project.py
----

and then follow the link:dev-eclipse.html#setup[setup instructions].

==== Refreshing the Classpath

If an updated classpath is needed, the Eclipse project can be
refreshed and missing dependency JARs can be downloaded by running
`project.py` again. For IntelliJ, you need to click the `Sync Project
with BUILD Files` button of link:https://ij.bazel.build[Bazel plugin].

[[documentation]]
=== Documentation

To build only the documentation for testing or static hosting:

----
  bazel build Documentation:searchfree
----

The html files will be bundled into `searchfree.zip` in this location:

----
  bazel-bin/Documentation/searchfree.zip
----

To build the executable WAR with the documentation included:

----
  bazel build withdocs
----

The WAR file will be placed in:

----
  bazel-bin/withdocs.war
----

[[tests]]
== Running Unit Tests

----
  bazel test --build_tests_only //...
----

Debugging tests:

----
  bazel test --test_output=streamed --test_filter=com.gerrit.TestClass.testMethod  testTarget
----

Debug test example:

----
  bazel test --test_output=streamed --test_filter=com.google.gerrit.acceptance.api.change.ChangeIT.getAmbiguous //javatests/com/google/gerrit/acceptance/api/change:api_change
----

To run a specific test group, e.g. the rest-account test group:

----
  bazel test //javatests/com/google/gerrit/acceptance/rest/account:rest_account
----

To run only tests that do not use SSH:

----
  bazel test --test_env=GERRIT_USE_SSH=NO //...
----

To exclude tests that have been marked as flaky:

----
  bazel test --test_tag_filters=-flaky //...
----

To exclude tests that require a Docker host:

----
  bazel test --test_tag_filters=-docker //...
----

To ignore cached test results:

----
  bazel test --cache_test_results=NO //...
----

To run one or more specific groups of tests:

----
  bazel test --test_tag_filters=api,git //...
----

The following values are currently supported for the group name:

* annotation
* api
* docker
* edit
* elastic
* git
* notedb
* pgm
* rest
* server
* ssh

[[elasticsearch]]
=== Elasticsearch

Successfully running the Elasticsearch tests requires Docker, and
may require setting the local virtual memory on
link:https://www.elastic.co/guide/en/elasticsearch/reference/current/vm-max-map-count.html[linux] and
link:https://www.elastic.co/guide/en/elasticsearch/reference/current/docker.html#_set_vm_max_map_count_to_at_least_262144[macOS].

On macOS, if using link:https://docs.docker.com/docker-for-mac/[Docker Desktop],
the effective memory value can be set in the Preferences, under the Advanced tab.
The default value usually does not suffice and is causing premature container exits.
That default is currently 2 GB and should be set to at least 5 (GB).

If Docker is not available, the Elasticsearch tests will be skipped.
Note that Bazel currently does not show
link:https://github.com/bazelbuild/bazel/issues/3476[the skipped tests].

== Dependencies

Dependency JARs are normally downloaded as needed, but you can
download everything upfront.  This is useful to enable
subsequent builds to run without network access:

----
  bazel fetch //...
----

When downloading from behind a proxy (which is common in some corporate
environments), it might be necessary to explicitly specify the proxy that
is then used by `curl`:

----
  export http_proxy=http://<proxy_user_id>:<proxy_password>@<proxy_server>:<proxy_port>
----

Redirection to local mirrors of Maven Central and the Gerrit storage
bucket is supported by defining specific properties in
`local.properties`, a file that is not tracked by Git:

----
  echo download.GERRIT = http://nexus.my-company.com/ >>local.properties
  echo download.MAVEN_CENTRAL = http://nexus.my-company.com/ >>local.properties
----

The `local.properties` file may be placed in the root of the gerrit repository
being built, or in `~/.gerritcodereview/`.  The file in the root of the gerrit
repository has precedence.

== Building against unpublished Maven JARs

To build against unpublished Maven JARs, like gwtorm or PrologCafe, the custom
JARs must be installed in the local Maven repository (`mvn clean install`) and
`maven_jar()` must be updated to point to the `MAVEN_LOCAL` Maven repository for
that artifact:

[source,python]
----
 maven_jar(
   name = 'gwtorm',
   artifact = 'gwtorm:gwtorm:42',
   repository = MAVEN_LOCAL,
 )
----

== Building against artifacts from custom Maven repositories

To build against custom Maven repositories, two modes of operations are
supported: with rewrite in local.properties and without.

Without rewrite the URL of custom Maven repository can be directly passed
to the maven_jar() function:

[source,python]
----
  GERRIT_FORGE = 'http://gerritforge.com/snapshot'

  maven_jar(
    name = 'gitblit',
    artifact = 'com.gitblit:gitblit:1.4.0',
    sha1 = '1b130dbf5578ace37507430a4a523f6594bf34fa',
    repository = GERRIT_FORGE,
 )
----

When the custom URL has to be rewritten, then the same logic as with Gerrit
known Maven repository is used: Repo name must be defined that matches an entry
in local.properties file:

----
  download.GERRIT_FORGE = http://my.company.mirror/gerrit-forge
----

And corresponding WORKSPACE excerpt:

[source,python]
----
  GERRIT_FORGE = 'GERRIT_FORGE:'

  maven_jar(
    name = 'gitblit',
    artifact = 'com.gitblit:gitblit:1.4.0',
    sha1 = '1b130dbf5578ace37507430a4a523f6594bf34fa',
    repository = GERRIT_FORGE,
 )
----

[[consume-jgit-from-development-tree]]

To consume the JGit dependency from the development tree, edit
`lib/jgit/jgit.bzl` setting LOCAL_JGIT_REPO to a directory holding a
JGit repository.

[[bazel-local-caches]]

To accelerate builds, several caches are activated per default:

* ~/.gerritcodereview/bazel-cache/downloaded-artifacts
* ~/.gerritcodereview/bazel-cache/repository
* ~/.gerritcodereview/bazel-cache/cas

Currently none of these caches have a maximum size limit. See
link:https://github.com/bazelbuild/bazel/issues/5139[this bazel issue] for
details. Users should watch the cache sizes and clean them manually if
necessary.

[[npm-binary]]
== NPM Binaries

Parts of the PolyGerrit build require running NPM-based JavaScript programs as
"binaries". We don't attempt to resolve and download NPM dependencies at build
time, but instead use pre-built bundles of the NPM binary along with all its
dependencies. Some packages on
link:https://docs.npmjs.com/misc/registry[registry.npmjs.org] come with their
dependencies bundled, but this is the exception rather than the rule. More
commonly, to add a new binary to this list, you will need to bundle the binary
yourself.

[NOTE]
We can only use binaries that meet certain licensing requirements, and that do
not include any native code.

Start by checking that the license and file types of the bundle are acceptable:
[source,bash]
----
  gerrit_repo=/path/to/gerrit
  package=some-npm-package
  version=1.2.3

  npm install -g license-checker && \
  rm -rf /tmp/$package-$version && mkdir -p /tmp/$package-$version && \
  cd /tmp/$package-$version && \
  npm install $package@$version && \
  license-checker | grep licenses: | sort -u
----

This will output a list of the different licenses used by the package and all
its transitive dependencies. We can only legally distribute a bundle via our
storage bucket if the licenses allow us to do so. As long as all of the listed
license are allowed by
link:https://opensource.google.com/docs/thirdparty/licenses/[Google's
standards]. Any `by_exception_only`, commercial, prohibited, or unlisted
licenses are not allowed; otherwise, it is ok to distribute the source. If in
doubt, contact a maintainer who is a Googler.

Next, check the file types:
[source,bash]
----
  cd /tmp/$package-$version
  find . -type f | xargs file | grep -v 'ASCII\|UTF-8\|empty$'
----

If you see anything that looks like a native library or binary, then we can't
use the bundle.

If everything looks good, create the bundle, and note the SHA-1:
[source,bash]
----
  $gerrit_repo/tools/js/npm_pack.py $package $version && \
  sha1sum $package-$version.tgz
----

This creates a file named `$package-$version.tgz` in your working directory.

Any project maintainer can upload this file to the
link:https://console.cloud.google.com/storage/browser/gerrit-maven/npm-packages[storage
bucket].

Finally, add the new binary to the build process:
----
  # WORKSPACE
  npm_binary(
      name = "some-npm-package",
      repository = GERRIT,
  )

  # lib/js/npm.bzl
  NPM_VERSIONS = {
    ...
    "some-npm-package": "1.2.3",
  }

  NPM_SHA1S = {
    ...
    "some-npm-package": "<sha1>",
  }
----

To use the binary from the Bazel build, you need to use the `run_npm_binary.py`
wrapper script. For an example, see the use of `crisper` in `tools/bzl/js.bzl`.



[[RBE]]
== Google Remote Build Support

The Bazel build can be used with Google's Remote Build Execution.


This needs the following setup steps:

```
gcloud auth application-default login
gcloud services enable remotebuildexecution.googleapis.com  --project=${PROJECT}
```

Create a worker pool. The instances should have at least 4 CPUs each
for adequate performance.

```
gcloud alpha remote-build-execution worker-pools create default \
    --project=${PROJECT} \
    --instance=default_instance \
    --worker-count=50 \
    --machine-type=n1-highcpu-4 \
    --disk-size=200
```

To use RBE, execute

```
bazel test --config=remote \
    --remote_instance_name=projects/${PROJECT}/instances/default_instance \
    javatests/...
```




GERRIT
------
Part of link:index.html[Gerrit Code Review]

SEARCHBOX
---------