When you build with bazel you sometimes want to pass a environment variable that changes something, it can be the name of the output or a config inside the app.
A good example could be passing a SHA during a build.
Bazel has a lot of CLI arguments which makes it very difficult to easily find what you are looking for.
There is a few instances where you want to do it:
You can pass environment variables with "--action_env", an example is here
bazel build //:myproject --action_env=MYENV=myvalue
You can test it easily by making a new project like this:
genrule( name = "myproject", cmd = "printenv > $@", outs = ["envs.txt"] )
You can use
--test_env and those will be available during bazel test.
bazel test //:mytest --test_env=MYENV=myvalue
genrule( name = "myproject", cmd = "echo printenv > $@", outs = ["envs.sh"], executable = True ) sh_test( name = "mytest", srcs = ["envs.sh"], data = [":myproject"], )
And you can see the logs even if the test passes with
So it looks like
bazel test --test_output=all //:mytest --test_env=MYENV=myvalue
Here you can prepend the environment variable when you run the rule.
genrule( name = "myproject", cmd = "echo printenv > $@", outs = ["envs.sh"], executable = True )
MYRUN=value bazel run //:myproject
It is not easy to find out how to pass information during build, test and run with bazel, but that is on purpose. When you allow external values it can make the build inconsistent and fragile so think about what you introduce, but don't worry so much if you are just getting started!
|Newer post:||How to get started with Bazel buildtool|