4. Subscriptions

Messages published to a topic on an AMPS server are available to other clients via a subscription. Before messages can be received, a client must subscribe to one or more topics on the AMPS server so that the server will begin sending messages to the client. The server will continue sending messages to the client until the client unsubscribes, or the client disconnects. With content filtering, the AMPS server will limit the messages sent only to those messages that match a client-supplied filter. In this chapter, you will learn how to subscribe, unsubscribe, and supply filters for messages using the AMPS C/C++ client.

Subscribing

Subscribing to an AMPS topic takes place by calling Client.subscribe(). Here is a short example showing the simplest way to subscribe to a topic (error handling and connection details are omitted for brevity):

Client client(...);
client.connect(...);

/* Here we have created or received a Client that is properly connected to an AMPS server. */
client.logon();


/* Here we subscribe to the topic messages. We do not provide a filter, so AMPS
 * does not content-filter the subscription. Although we don't use the object
 * explicitly here, the subscribe function returns a MessageStream object that we
 * iterate over. If, at any time, we no longer need to subscribe, we can break out
 * of the loop. When we break out of the loop, the MessageStream goes out of scope,
 * the MessageStream destructor runs, and the AMPS client sends an unsubscribe
 * command to AMPS.
 */
for (auto message : client.subscribe("messages"))
{
    /* Within the body of the loop, we can process the message as we need to. In this
     * case, we simply print the contents of the message.
     */
    std :: cout << "Received message: " << message.getData () << std :: endl ;
}

Example 4.1: Subscribing to a topic

AMPS creates a background thread that receives messages and copies them into a MessageStream that you iterate over. This means that the client application as a whole can continue to receive messages while you are doing processing work.

The simple method described above is provided for convenience. The AMPS C++ client provides convenience methods for the most common form of the AMPS commands. The client also provides an interface that allows you to have precise control over the command. Using that interface, the example above becomes:

Client client(...);
client.connect(...);

/* Here we have created or received a Client that is properly connected to an AMPS server. */
client.logon();


/* Here we subscribe to the topic messages. We do not provide a filter, so AMPS
 * does not content-filter the subscription. Although we don't use the object
 * explicitly here, the execute function returns a MessageStream object that we
 * iterate over. If, at any time, we no longer need to subscribe, we can break out
 * of the loop. When we break out of the loop, the MessageStream goes out of scope,
 * the MessageStream destructor runs, and the AMPS client sends an unsubscribe
 * command to AMPS.
 *
 * Here we create a command object for the subscribe command, specifying the topic
 * messages.
 */
for (auto message : ampsClient.execute(Command("subscribe").setTopic("messages")))
{
    std :: cout << "Received message: "<< message.getData () << std :: endl;
}

Example 4.2: Subscribing to a topic

The Command interface allows you to precisely customize the commands you send to AMPS. For flexibility and ease of maintenance, 60East recommends using the Command interface (rather than a named method) for any command that will receive messages from AMPS. For publishing messages, there can be a slight performance advantage to using the named commands where possible.

Asynchronous Message Processing Interface

The AMPS C++ client also supports an interface that allows you to process messages asynchronously. In this case, you add a message handler to the function call. The client returns the command ID of the subscribe command once the server has acknowledged that the command has been processed. As messages arrive, the client calls your message handler directly on the background thread. This can be an advantage for some applications. For example, if your application is highly multithreaded and copies message data to a work queue processed by multiple threads, there is usually a performance benefit to enqueuing work directly from the background thread. See Understanding Threading for a discussion of threading considerations, including considerations for message handlers.

Here is a short example (error handling and connection details are omitted for brevity):

Client client(...);
client.connect(...);
/* Here we have created or received a Client that is properly connected to an AMPS server. */


client.logon();

/* Here we create a subscription with the following parameters:
 * command        : This is the AMPS Command object that contains the subscribe command.
 * MessageHandler : This is an AMPS MessageHandler object that refers to our message
 *                  handling function myHandlerFunction. This function is
 *                  called on a background thread each time a message arrives.
 *                  The second parameter, NULL, is passed as-is from the
 *                  client.subscribe() call to the message handler with every message,
 *                  allowing you to pass context about the subscription through to the
 *                  message handler.
 *
 * We create a command object for the subscribe command, specifying the topic
 * messages.
 */
string subscriptionId = client.executeAsync(Command("subscribe").setTopic("messages"),
                                            MessageHandler(myHandlerFunction, NULL));

...

/* The myHandlerFunction is a global function that is invoked by AMPS whenever a
 * matching message is received. The first parameter, message, is a reference to an
 * AMPS Message object that contains the data and headers of the received message.
 * The second parameter, userData, is set to whatever value was provided in the
 * MessageHandler constructor -- NULL in this example.
 */
void myHandlerFunction(const Message& message, void* userData)
{
    std::cout << message.getData() << std::endl;
}

Example 4.3: Subscribing to a process with asynchronous processing

Caution

The AMPS client resets and reuses the message provided to this function between calls. This improves performance in the client, but means that if your handler function needs to preserve information contained within the message, you must copy the information rather than just saving the message object. Otherwise, the AMPS client cannot guarantee the state of the object or the contents of the object when your program goes to use it.

With newer compilers, you can use additional constructs to specify a callback function. Recent improvements in C++ have added lambda functions – unnamed functions declared in-line that can refer to names in the lexical scope of their creator. If available on your system, both Standard C++ Library function objects and lambda functions may be used as callbacks. Check functional.cpp in the samples directory for numerous examples.

Using an Instance Method as a Message Handler

One of the more common ways of providing a message handler is as an instance method on an object that maintains message state. It’s simple to provide a handler with this capability, as shown below.

class StatefulHandler
{
private:
    std::string _handlerName;
public:
    /* Construct the handler and save state. */
    StatefulHandler(const std::string& name) : _handlerName(name) {}

    /* Message handler method. */
    void operator()(const AMPS::Message & message)
    {
      std::cout << _handlerName << " got " << message.getData() << std::endl;
    }
};

You can then provide an instance of the handler directly wherever a message handler is required, as shown below:

client.subscribe(StatefulHandler("An instance"), "topic");

Understanding Threading and Message Handlers

The first time a command causes an instance of the Client or HAClient to connect to AMPS (typically, the logon() command), the client creates a thread that runs in the background. This background thread is responsible for processing incoming messages from AMPS, which includes both messages that contain data and acknowledgments from the server.

When you call a command on the AMPS client, the command typically waits for an acknowledgment from the server and then returns. (The exception to this is publish. For performance, the publish command does not wait for an acknowledgment from the server before returning.)

In the simple case, using synchronous message processing, the client provides an internal handler function that populates the MessageStream. The client receive thread calls the internal handler function, which makes a deep copy of the incoming message and adds it to the MessageStream. The MessageStream is used on the calling thread, so operations on the MessageStream do not block the client receive thread.

When using asynchronous message processing, AMPS calls the handler function from the client receive thread. Message handlers provided for asynchronous message processing must be aware of the following considerations.

The client creates one client receive thread at a time, and the lifetime of the thread lasts for the lifetime of the connection to the AMPS server. A message handler that is only provided to a single client will only be called from a single thread at a time. If your message handler will be used by multiple clients, then multiple threads will call your message handler. In this case, you should take care to protect any state that will be shared between threads. Notice that if the client connection fails (or is closed), and the client reconnects, the client will create a different thread for the new connection.

For maximum performance, do as little work in the message handler as possible. For example, if you use the contents of the message to update an external database, a message handler that adds the relevant data to an update queue that is processed by a different thread will typically perform better than a message handler that does this update during the message handler.

While your message handler is running, the thread that calls your message handler is no longer receiving messages. This makes it easier to write a message handler, because you know that no other messages are arriving from the same subscription. However, this also means that you cannot use the same client that called the message handler to send commands to AMPS. Acknowledgments from AMPS cannot be processed, and your application will deadlock waiting for the acknowledgment. Instead, enqueue the command in a work queue to be processed by a separate thread, or use a different client object to submit the commands.

The AMPS client resets and reuses the Message provided to this function between calls. This improves performance in the client, but means that if your handler function needs to preserve information contained within the message, you must copy the information (either by making a copy of the entire message or copying the required fields) rather than just saving the message object. Otherwise, the AMPS client cannot guarantee the state of the object or the contents of the object when your program goes to use it. Likewise, a message handler should not modify the Message – this will result in modifying the message provided to other handlers (including handlers internal to the AMPS client.).

Unsubscribing

The AMPS server continues a subscription until the client explicitly ends the subscription (that is, unsubscribes) or until the connection to the client is closed.

With the synchronous interface, AMPS automatically unsubscribes to the topic when the destructor for the MessageStream runs. You can also explicitly call the close() method on the MessageStream object to remove the subscription.

In the asynchronous interface, when a subscription is successfully made, messages will begin flowing to the message handler, and the subscribe() or executeAsync() call will return a string for the subscription id that serves as the identifier for this subscription. A Client can have any number of active subscriptions, and this subscription id is how AMPS designates messages intended for this particular subscription. To unsubscribe, we simply call unsubscribe with the subscription identifier:

Client client = ...;

// Register asynchronous subscription
std::string subId = client.executeAsync(
                         Command("subscribe").setTopic("messages"),
                         MessageHandler(myHandlerFunction, NULL));

... other work here ...

client.unsubscribe(subId);

Example 4.4: Unsubscribing form a topic

In this example, as in the previous section, we use the client.executeAsync() method to create a subscription to the messages topic. When our application is done listening to this topic, it unsubscribes by passing in the subId returned by subscribe(). After the subscription is removed, no more messages will flow into our myHandlerFunction().

When an application calls unsubscribe(), the client sends an explicit unsubscribe command to AMPS. The AMPS server removes that subscription from the set of subscriptions for the client, and stops sending messages for that subscription. On the client side, the client unregisters the subscription so that the MessageStream or MessageHandler for that subscription will no longer receive messages for that subscription.

Notice that calling unsubscribe does not destroy messages that the server has already sent to the client. If there are messages on the way to the client for this subscription, the AMPS client must consume those messages. If a LastChanceMessageHandler is registered, the handler will receive the messages. Otherwise, they will be discarded since no message handler matches the subscription ID on the message.

Understanding messages

So far, we have seen that subscribing to a topic involves working with objects of AMPS::Message type. A Message represents a single message to or from an AMPS server. Messages are received or sent for every client/server operation in AMPS.

Header properties

There are two parts of each message in AMPS: a set of headers that provide metadata for the message, and the data that the message contains. Every AMPS message has one or more header fields defined. The precise headers present depend on the type and context of the message. There are many possible fields in any given message, but only a few are used for any given message. For each header field, the Message class contains a distinct property that allows for retrieval and setting of that field. For example, the Message.get_command_id() function corresponds to the commandId header field, the Message.get_batch_size() function corresponds to the BatchSize header field, and so on. For more information on these header fields, consult the AMPS User Guide and AMPS Command Reference.

To work with header fields, a Message contains getXxx()/setXxx() methods corresponding to the header fields. 60East does not recommend attempting to parse header fields from the raw data of the message.

In AMPS, fields sometimes need to be set to a unique identifier value. For example, when creating a new subscription, or sending a manually constructed message, you’ll need to assign a new unique identifier to multiple fields such as CommandId and SubscriptionId. For this purpose, Message provides newXxx() methods for each field that generates a new unique identifier and sets the field to that new value.

getData() method

Access to the data section of a message is provided via the getData() method. The data contains the unparsed data in the message, returned as a series of bytes (a string or const char *). Your application code parses and works with the data.

The AMPS C++ client contains a collection of helper classes for working with message types that are specific to AMPS (for example, FIX, NVFIX, and AMPS composite message types). For message types that are widely used, such as JSON or XML, you can use whichever library you typically use in your environment.

Advanced Messaging Support

The client.subscribe() function provides options for subscribing to topics even when you do not know their exact names, and for providing a filter that works on the server to limit the messages your application must process.

Regex topics

Regular Expression (Regex) Topics allow a regular expression to be supplied in the place of a topic name. When you supply a regular expression, it is as if a subscription is made to every topic that matches your expression, including topics that do not yet exist at the time of creating the subscription.

To use a regular expression, simply supply the regular expression in place of the topic name in the subscribe() call. For example:

for (auto message : client.subscribe("client.*"))
{
    /* receive messages for any topic that begins with 'client' */
    std::cout << "Received a message on topic '" << message.getTopic() << "' "
              << "with the data: " << message.getData() << std::endl;
}

Example 4.5: Regex topic subscription

In this example, messages on topics client and client1 would match the regular expression, and those messages will be returned by the MessageStream . As in the example, you can use the getTopic() method to determine the actual topic of the message sent to the lambda function.

Content filtering

One of the most powerful features of AMPS is content filtering. With content filtering, filters based on message content are applied at the server, so that your application and the network are not utilized by messages that are uninteresting for your application. For example, if your application is only displaying messages from a particular user, you can send a content filter to the server so that only messages from that particular user are sent to the client. The AMPS User Guide provides full details on content filtering.

To apply a content filter to a subscription, simply pass it into the client.subscribe() call:

for (auto message : ampsClient.subscribe("messages", 0, "/sender = 'mom'"))
{
    // process messages from mom
}

Example 4.6: Using content filters

In this example, we have passed in a content filter /sender = 'mom'. This will cause the server to only send us messages from the messages topic that additionally have a sender field equal to mom.

For example, the AMPS server will send the following message, where /sender is mom:

{
    "sender" : "mom",
    "text" : "Happy Birthday!",
    "reminder" : "Call me Thursday!"
}

The AMPS server will not send a message with a different /sender value:

{
    "sender" : "henry dave",
    "text" : "Things do not change; we change."
}

Updating the Filter on a Subscribe

AMPS allows you to update the filter on a subscription. When you replace a filter on the subscription, AMPS immediately begins sending only messages that match the updated filter. Notice that if the subscription was entered with a command that includes a SOW query, using the replace option can re-issue the SOW query (as described in the AMPS User Guide).

To update a the filter on a subscription, you create a subscribe command. You set the SubscriptionId provided on the Command to the identifier of the existing subscription, and include the replace option on the Command. When you send the Command, AMPS atomically replaces the filter and sends messages that match the updated filter from that point forward.

Next steps

At this point, you are able to build AMPS programs in C/C++ that publish and subscribe to AMPS topics. For an AMPS application to be truly robust, it needs to be able to handle the errors and disconnections that occur in any distributed system. In the next chapter, we will take a closer look at error handling and recovery, and how you can use it to make your application ready for the real world.