Actions

Functions, the core of OpenServerless

Actions

An action can generally be considered as a function, a snippet of code, or generally a method.

The ops action command is designed for managing actions, featuring frequently utilized CRUD operations such as list, create, update, and delete. We will illustrate these operations through examples using a basic hello action. Let’s assume we have the following file in the current directory:

The hello.js script with the following content:

function main(args) {
    return { body: "Hello" }
}

Simple Action Deployment

If we want to deploy this simple action in the package demo, let’s execute:

$ ops package update demo
ok: updated package demo
$ ops action update demo/hello hello.js
ok: update action demo/hello

Note that we ensured the package exists before creating the action.

We can actually omit the package name. In this case, the package name is default, which always exists in a namespace. However, we advise always placing actions in some named package.

💡 NOTE

We used update, but we could have used create if the action does not exist because update also creates the action if it does not exist and updates it if it is already there. Update here is similar to the patch concept in REST API. However, create generates an error if an action does not exist, while update does not, so it is practical to always use update instead of create (unless we really want an error for an existing action for some reason).

How to Invoke Actions

Let’s try to run the action:

$ ops invoke demo/hello
{
    "body": "Hello"
}

Actually, the invoke command does not exist, or better, it’s just a handy shortcut for ops action invoke -r.

If you try to run ops action invoke demo/hello, you get:

$ ops action invoke demo/hello
ok: invoked /_/demo/hello with id fec047bc81ff40bc8047bc81ff10bc85

You may wonder where the result is. In reality, in OpenServerless, all actions are by default asynchronous, so what you usually get is the activation id to retrieve the result once the action is completed.

To block the execution until the action is completed and get the result, you can either use the flag -r or --result, or use ops invoke.

Note, however, that we are using ops to invoke an action, which means all the requests are authenticated. You cannot invoke actions directly without logging into the system first.

However, you can mark an action to be public by creating it with --web true (see below).

Public Actions

If you want an action to be public, you can do:

$ ops action update demo/hello hello.js --web true
ok: updated action demo/hello
$ ops url demo/hello
https://nuvolaris.dev/api/v1/web/mirella/demo/hello

and you can invoke it with:

$ curl -sL https://nuvolaris.dev/api/v1/web/dashboard/demo/hello
Hello

Note that the output is only showing the value of the body field. This is because the web actions must follow a pattern to produce an output suitable for web output, so the output should be under the key body, and so on. Check the section on Web Actions for more information.

💡 NOTE

Actually, ops url is a shortcut for ops action get --url. You can use ops action get to retrieve a more detailed description of an action in JSON format.

After action create, action update, and action get (and the shortcuts invoke and url), we should mention action list and action delete.

The action list command obviously lists actions and allows us to delete them:

$ ops action list
/mirella/demo/hello                                                  private nodejs:18
$ ops action delete demo/hello
ok: deleted action demo/hello

Conclusion

Actions are a core part of our entities. A ops action is a self-contained and executable unit of code deployed on the ops serverless computing platform.


Last modified August 31, 2024: Merge all work done until now (#23) (3327e3a)