As described in the H4v3n testing page, the ability to send batches of method calls in a single request method (along with the tolerant reader and built-in assertions within each method result) makes the implementation of remote end-to-end testing fairly simple. One other facet of the request message batching of method calls is used to simplify the retrieval of system notifications and action messages. It can best be described as a sort of polling mechanism, without any actual polling.
It works for each client application with a simple timestamp stored as a property of the object that handles client communication with the server. Each time a request message is posted to the web service, the timestamp is compared with the current time. If the interval between the two is greater than the defined minimum value (configurable), then a method call to retrieve system messages is automatically appended to the outbound request message. This method call to retrieve system messages can retrieve many different types of messages from the system.
Any results of the query for system messages is handled automatically by the logic of the AppLib classes which process response messages received by the client. System messages are structured somewhat like a request message, except they are a request from the server to execute logic within the client application. They are processed by a mini-pipeline on the client just like on the server, complete with a controller to route the request to the proper internal method if properly authenticated. Client authorization for the system messages is automatic, since any client-side data is partitioned per system.
This mechanism can be used for any type of automated processing that makes sense to be distributed down to each client application.