Both kits let you create so-called MIDI endpoints, but the endpoints from midi1 cannot be shared between different applications. The midi2 kit solves that problem, but unlike midi1 it does not include a General MIDI softsynth, nor does it have a facility for reading and playing Standard MIDI Files. Don't worry: both kits are compatible and you can mix-and-match them in your applications.
The main differences between the two kits:
Here is a pretty picture:
The first and most important point to note is that BMidiConsumer and BMidiProducer in the Midi Kit are NOT directly analogous to BBufferConsumer and BBufferProducer in the Media Kit! In the Media Kit, consumers and producers are the data consuming and producing properties of a media node. A filter in the Media Kit, therefore, inherits from both BBufferConsumer and BBufferProducer, and implements their virtual member functions to do its work.
In the Midi Kit, consumers and producers act as endpoints of MIDI data connections, much as media_source and media_destination do in the Media Kit. Thus, a MIDI filter does not derive from BMidiConsumer and BMidiProducer; instead, it contains BMidiConsumer and BMidiProducer objects for each of its distinct endpoints that connect to other MIDI objects. The Midi Kit does not allow the use of multiple virtual inheritance, so you can't create an object that's both a BMidiConsumer and a BMidiProducer.
This also contrasts with the old Midi Kit's conception of a BMidi object, which stood for an object that both received and sent MIDI data. In the new Midi Kit, the endpoints of MIDI connections are all that matters. What lies between the endpoints, i.e., how a MIDI filter is actually structured, is entirely at your discretion.
Also, rather than use token structs like media_node to make connections via the MediaRoster, the new kit makes the connections directly via the BMidiProducer object.
BMidiRoster only gives you access to BMidiEndpoints, BMidiConsumers, and BMidiProducers. When you want to talk to remote MIDI objects, you do so through the proxy objects that BMidiRoster provides. Unlike BMidiLocalConsumer and BMidiLocalProducer, these classes do not provide a lot of functions. That is intentional. In order to hide the details of communication with MIDI endpoints in other applications, the Midi Kit must hide the details of how a particular endpoint is implemented.
So, what can you do with remote objects? Only what BMidiConsumer, BMidiProducer, and BMidiEndpoint will let you do. You can connect objects, get the properties of these objects -- and that's about it.
new operator, on the stack, or as globals. Nor can you delete them. Instead, these objects are obtained through BMidiRoster. The only two exceptions to this rule are BMidiLocalConsumer and BMidiLocalProducer. These two objects may be directly created and subclassed by developers.
This means that, to delete an endpoint, you don't call the delete operator directly; instead, you call Release() . To balance this call, there's also an Acquire() , in case you have two disparate parts of your application working with the endpoint, and you don't want to have to keep track of who needs to Release() the endpoint.
When you're done with any endpoint object, you must Release() it. This is true for both local and remote objects. Repeat after me: Release() when you're done.
The Midi Kit already provides a set of commonly used spray functions, such as SprayNoteOn() , SprayControlChange() , and so on. These correspond one-to-one with the message types from the MIDI spec. You don't need to be a MIDI expert to use the kit, but of course some knowledge of the protocol helps. If you are really hardcore, you can also use the SprayData() to send raw MIDI events to the consumers.
At the consumer side, a dedicated thread invokes a hook function for every incoming MIDI event. For every spray function, there is a corresponding hook function, e.g. NoteOn() and ControlChange() . The hardcore MIDI fanatics among you will be pleased to know that you can also tap into the Data() hook and get your hands dirty with the raw MIDI data.
If you override a hook function in one of your consumer objects, it should look at the time argument, wait until the designated time, and then perform its action. The preferred method is to use the Kernel Kit's snooze_until() function, which sends the consumer thread to sleep until the requested time has come. (Or, if the time has already passed, returns immediately.)
Like this:
void MyConsumer::NoteOn( uchar channel, uchar note, uchar velocity, bigtime_t time) { snooze_until(time, B_SYSTEM_TIMEBASE); ...do your thing... }
If you want your producers to run in real time, i.e. they produce MIDI data that needs to be performed immediately, you should pass time 0 to the spray functions (which also happens to be the default value). Since time 0 has already passed, snooze_until() returns immediately, and the consumer will process the events as soon as they are received.
To schedule MIDI events for a performance time that lies somewhere in the future, the producer must take into account the consumer's latency. Producers should attempt to get notes to the consumer by or before (scheduled_performance_time - latency). The time argument is still the scheduled performance time, so if your consumer has latency, it should snooze like this before it starts to perform the events:
snooze_until(time - Latency(), B_SYSTEM_TIMEBASE);
Note that a typical producer sends out its events as soon as it can; unlike a consumer, it does not have to snooze.
snooze_until(), it will not read its port. Consequently, any producer that tries to write a new event to this port will block until the consumer thread is ready to receive a new message. This is intentional, because it prevents producers from generating and queueing up thousands of events.This mechanism, while simple, puts on the producer the responsibility for sorting the events in time. Suppose your producer sends three Note On events, the first on t + 0, the second on t + 4, and the third on t + 2. This last event won't be received until after t + 4, so it will be two ticks too late. If this sort of thing can happen with your producer, you should somehow sort the events before you spray them. Of course, if you have two or more producers connected to the same consumer, it is nearly impossible to sort this all out (pardon the pun). So it is not wise to send the same kinds of events from more than one producer to one consumer at the same time.
The article Introduction to MIDI, Part 2 in OpenBeOS Newsletter 36 describes this problem in more detail, and provides a solution. Go read it now!
Filters should ignore the time argument from the spray and hook functions, and simply pass it on unchanged. Objects that only filter data should process the event as quickly as possible and be done with it. Do not snooze_until() in the consumer endpoint of a filter!
Information about MIDI in general: