@eclipse-emfcloud/coffee-custom-commands-example

Custom Commands Example

This package provides an example of a command provider plug-in for the Model Server. It provides an implementation of the custom 'increment-duration' command for a workflow Task in a Coffee model.

Setup

See the parent readme for details of how to set up and build the project.

How to Run

For detailed instructions how to run the example server configuration that includes this plug-in example, see the Example Server readme.

Send Requests to the Model Server

There are a variety of tools available for ad hoc REST interactions with services such as the Model Server. A recommended option is the Postman application.

Start by fetching the current state of the model. Send a GET request to

    http://localhost:8082/api/v2/models?modeluri=SuperBrewer3000.coffee

The result should look something like this:

{
  "type": "success",
  "data": {
    "$type": "http://www.eclipsesource.com/modelserver/example/coffeemodel#//Machine",
    "$id": "/",
    "children": [
      // ... structural components here ...
    ],
    "name": "Super Brewer 3000",
    "workflows": [
      {
        "$id": "//@workflows.0",
        "name": "Simple Workflow",
        "nodes": [
          {
            "$type": "http://www.eclipsesource.com/modelserver/example/coffeemodel#//AutomaticTask",
            "$id": "//@workflows.0/@nodes.0",
            "name": "PreHeat",
            "component": {
              "$type": "http://www.eclipsesource.com/modelserver/example/coffeemodel#//BrewingUnit",
              "$ref": "//@children.0"
            }
          }
        ]
      }
    ]
  }
}

Make note of the EMF-style URI fragment $id of the PreHeat task, which must be inferred from the model structure. The URI fragment is //@workflows.0/@nodes.0: the first node of the first workflow in the root Machine object. This will be used to build requests for custom commands.

Next, send a request to execute the custom "increment duration" command. This command is a composite of two commands in a four-step process:

1. Fetch the current value of the duration property of the affected Task. The default is 0 when the property is not present, as is initially the case in this example model.
2. Add 1 to the current duration and execute a set command to update the duration.
3. Extract the previous value of the duration from the CommandExecutionResult returned by the server. Add the new duration value calculated at step 2 to this previous value of the duration.
4. Execute a second command that sets the second new duration computation into the Task.

The example command provider uses a transaction on the target model to execute these two commands, receiving intermediate results that are inputs to later commands in the sequence, in a single undoable operation on the undo stack.

To execute this custom command, send a PATCH request to

    http://localhost:8082/api/v2/models?modeluri=SuperBrewer3000.coffee

with the following body:

{
  "data": {
    "type": "modelserver.emfcommand",
    "data": {
      "type": "increment-duration",
      "$type": "http://www.eclipse.org/emfcloud/modelserver/command#//Command",
      "owner": {
        "$ref": "SuperBrewer3000.coffee#//@workflows.0/@nodes.0",
        "$type": "http://www.eclipse.org/emfcloud/coffee/model#//AutomaticTask"
      }
    }
  }
}

Note that this uses the EMF URI fragment determined earlier in the owner.$ref property. The command is a custom increment-duration type and has no other parameters.

Fetch the model again with a GET request to see the updated duration of the PreHeat task.

Now repeat the same command a few times, each time getting the model again to see how the duration changes.

Do you recognize a pattern?

Undo the Commands

A critical component of this custom command provider framework is the new Model Server protocol for opening and closing transactions on models. This transaction context allows a command provider plug-in in the Model Server to execute a chain of commands with intermediate results, all captured as a single undoable unit on the stack. During the transaction, the Model Server receives private notification of the model changes and broadcast to subscribers at large is deferred to the close of the transaction. If the transaction is rolled back, of course such broadcast will not occur as all changes performed are reverted.

To send an undo command to the Model Server, simply send a GET request to

    http://localhost:8082/api/v2/undo?modeluri=SuperBrewer3000.coffee

Follow that up with a GET on the /api/v2/models endpoint to see how the duration of the PreHeat Task is reverted. The intermediate +1 increments are not evident because both original commands are undone at each step.