grainsv2 9

INSTALLATION

pip install grainsv2

INSTALLATION FOR DEVELOPMENT

Clone the grains repo and install with pip:

git clone https://gitlab.com/saltstack/pop/grains.git
pip install -e grains

EXECUTION

After installation the grains command should now be available.

The command will have no output until it is vertically extended (details below).

TESTING

install requirements-test.txt with pip and run pytest:

pip install -r grains/requirements-test.txt
pytest grains/tests

VERTICAL APP-MERGING

Instructions for extending grains

Install pop:

pip install --upgrade pop

Create a new directory for the project:

mkdir grains_{project_name}
cd grains_{project_name}

Use pop-seed to generate the structure of a project that extends grains:

pop-seed -t v grains_{project_name} -d grains

• “-t v” specifies that this is a vertically app-merged project

• “-d grains” says that we want to implement the dynamic name of “grains”

Notice that some structure has been created for you. We especially care about the new directory in grains_{project_name}/grains

Add “grains” to the requirements.txt:

echo "grains" >> requirements.txt

CREATING GRAINS

• Create a new file in “grains_{project_name}/grains”

  • The directory should already have been created by pop-seed.

  • The file name is arbitrary.

  • Every file in this directory, then it’s subdirectories, will be parsed “simultaneously” (as far as that makes sense for asyncio).

  • Grains that are dependant on each other should be assigned in the same function.

  • Do NOT rely on the collector’s recursion strategy for grains to depend on each other.

  • Use hub.grains.init.wait_for(“grains”) to wait for grains to be generated by another sub

• Add your new project to the python path

  • alternatively, run “pip install -e .” from your project’s root directory

    • you only need to do this once

• Create an async function in this file with a descriptive yet arbitrary name

  • By convention it should start with “load”

  • Make it async unless you have a really really really good reason.

  • Grains that depend on each other already belong in the same function, don’t depend on synchronous programming for determinism

  • Use hub.grains.init.wait_for(“specific_grain”) to explicitly wait for an external grain to become available before moving on

  • Grains can be accessed and assigned like a dictionary, but our convention is to use the namespace

  • Use hub.grains.GRAINS._dict() or copy.copy(hub.grains.GRAINS) to retrieve a JSON serializable version of grains.

Example:

async def load_my_grain(hub):
    hub.grains.GRAINS.new_grain = "Hello World!"

And that’s it! Now verify that your grain is collected from the command line:

grains new_grain

Output:

new_grain:
    Hello World!