ember-rewind-stack

ember-rewind-stack

Many applications have code that orchestrates a LOT of API calls. Particularly related to creation of complex solutions, it is quite reasonable to expect that one user action will result in 5 or more API calls, some of which will result in the creation of ancillary records/entities.

The question is - what happens when some critical path call fails in the middle of one of these giant chains?

Enter the rewind-stack-service, which allows you to manage multiple, possibly parallel, operation stacks. In the event of a failure, your .catch() block can ask the service for all the completed operations, and then take action to roll-back the changes.

Installation

ember install ember-rewind-stack

When should I use this?

Only use this when you are orchestrating a lot of calls. If you are just issuing a single save operation, this is massive overkill. Basically, if you can't easily figure out how to clean things up in a .catch(), then consider using this.

API

method	returns	description
addOperation (stackName, operationObj)	n/a	add an operation to a named stack. Creates the stack if it does not exist. Sets the .current to the passed in operation. If there is a .current, it is pushed into the stack's .completed array. A .startedAt property is added to the operation when pushed in. A .completedAt property is set when it's pushed into the .completed array.
completeOperation (stackName, options)	n/a	Moves the current operation onto the .completed array, adding the .completedAt property, and clearing the stack's .current property. The options are merged into the current operation, allowing useful stuff like itemId's to be added after an operation completes.
getCompleted (stackName)	array of completed operations, reversed	Used by exception handlers to rewind the completed actions
removeStack (stackName)	n/a	clean up a stack
getCurrent (stackName)	Current Operation Obj	Returns the current operation
getStack (stackName)	stack object	returns the entire stack object

Operation Object

The operation object can be whatever you need it to be. The intent here is that you provide enough info in that object to allow you to un-do the operation.

{
  "name": "create site item", // some sort of name that makes sense to you
  "type": "create-item",      // something that would make sense in a log
  "cleanup": "remove-item",   // a key for your clean up code to use...
  "inputs": {                 // whatever your cleanup code needs...
    "id": "3ef...",           // in this case, to delete an item we need the id and owner
    "owner": "dcadmin"
  }
}