YapDatabaseCloudCorePipelineDelegate

This method is invoked when the operation is ready to be started (i.e. ready to start network IO).

The delegate should attempt to perform the corresponding network task.

If the network task completes, the delegate should:

• perform a readWriteTransaction on the database
• update any object(s) in the database (as needed)
• invoke [[transaction ext:MyCloudCore] completeOperation:operation]

This allows for updating your object(s) and deleting the operation from the queue in the same atomic commit.

In the event the network task fails, the delegate should generally give the operation back to the pipeline. There are a few ways to accomplish this.

If the network task failed due to an Internet disconnection, typically the delegate will:

• suspend the pipeline or parent YDBCloudCore instance (usually done from a manager class that monitors reachability)
• invoke [pipeline setStatusAsPendingForOperationUUID:operation.uuid]

Since the pipeline has been suspended, it won’t be able to restart the operation until it’s resumed. Once resumed (presumably due to Internet reconnection) it will automatically re-start the operation by invoking ‘startOperation:withContext:forPipeline: again.

If the network task failed due to a rate-limiting error from the server, the delegate can:

• calculate a delay, possibly using an exponential backoff algorithm
• invoke [pipeline setStatusAsPendingForOperationUUID:operation.uuid withRetryDelay:delay]

The pipeline will use an internal timer to ensure the operation isn’t started again until after the delay expires.

Tip: You can use the pipeline’s ephemeralInfo dictionary to store a failCount for the operation, in order to calculate exponential backoff delay.

If the network task failed due to some unrecoverable error, then it may be the case that the operation needs to be skipped. In this case, the delegate will:

• perform a readWriteTransaction on the database
• update any object(s) in the database (as needed)
• invoke [[transaction ext:MyCloudCore] skipOperation:operation]

NOTE: The pipeline will attempt to start as many concurrent operations as it can. The number of concurrent operations is limited by:

• pipeline.maxConcurrentOperationCount
• the operations within the pipeline, and their corresponding dependencies