RtMidi is a set of C++ classes (RtMidiIn, RtMidiOut and API-specific classes) that provides a common API (Application Programming Interface) for realtime MIDI input/output across Linux (ALSA & JACK), Macintosh OS X (CoreMIDI & JACK), and Windows (Multimedia Library) operating systems. RtMidi significantly simplifies the process of interacting with computer MIDI hardware and software. It was designed with the following goals:
Where applicable, multiple API support can be compiled and a particular API specified when creating an RtAudio instance.
MIDI input and output functionality are separated into two classes, RtMidiIn and RtMidiOut. Each class instance supports only a single MIDI connection. RtMidi does not provide timing functionality (i.e., output messages are sent immediately). Input messages are timestamped with delta times in seconds (via a
double floating point type). MIDI data is passed to the user as raw bytes using an std::vector<unsigned char>.
There were no API changes made in the current release. The primary changes involved updates to the build system and some small bug fixes. With respect to the previous release (2.1.0), a minor API change was made. The RtError class was renamed RtMidiError and embedded directly in RtMidi.h. Thus, all references to RtError should be renamed to RtMidiError and the RtError.h file should be deleted. The Windows Kernel Streaming code was moved to a separate branch because it was uncompilable and incomplete.
Latest Release (11 February 2016): Version 2.1.1
The first thing that must be done when using RtMidi is to create an instance of the RtMidiIn or RtMidiOut subclasses. RtMidi is an abstract base class, which itself cannot be instantiated. Each default constructor attempts to establish any necessary "connections" with the underlying MIDI system. RtMidi uses C++ exceptions to report errors, necessitating try/catch blocks around many member functions. An RtMidiError can be thrown during instantiation in some circumstances. A warning message may also be reported if no MIDI devices are found during instantiation. The RtMidi classes have been designed to work with "hot pluggable" or virtual (software) MIDI devices, making it possible to connect to MIDI devices that may not have been present when the classes were instantiated. The following code example demonstrates default object construction and destruction:
Obviously, this example doesn't demonstrate any of the real functionality of RtMidi. However, all uses of RtMidi must begin with construction and must end with class destruction. Further, it is necessary that all class methods that can throw a C++ exception be called within a try/catch block.
RtMidi uses a C++ exception handler called RtMidiError, which is declared and defined in RtMidi.h. The RtMidiError class is quite simple but it does allow errors to be "caught" by RtMidiError::Type. Many RtMidi methods can "throw" an RtMidiError, most typically if a driver error occurs or an invalid function argument is specified. There are a number of cases within RtMidi where warning messages may be displayed but an exception is not thrown. A client error callback function can be specified (via the RtMidi::setErrorCallback function) that is invoked when an error occurs. By default, error messages are not automatically displayed in RtMidi unless the preprocessor definition RTMIDI_DEBUG is defined during compilation. Messages associated with caught exceptions can be displayed with, for example, the RtMidiError::printMessage() function.
A client generally must query the available MIDI ports before deciding which to use. The following example outlines how this can be done.
The RtMidiOut class provides simple functionality to immediately send messages over a MIDI connection. No timing functionality is provided.
In the following example, we omit necessary error checking and details regarding OS-dependent sleep functions. For a complete example, see the
midiout.cpp program in the
The RtMidiIn class uses an internal callback function or thread to receive incoming MIDI messages from a port or device. These messages are then either queued and read by the user via calls to the RtMidiIn::getMessage() function or immediately passed to a user-specified callback function (which must be "registered" using the RtMidiIn::setCallback() function). We'll provide examples of both usages.
The RtMidiIn::getMessage() function does not block. If a MIDI message is available in the queue, it is copied to the user-provided
std::vector<unsigned char> container. When no MIDI message is available, the function returns an empty container. The default maximum MIDI queue size is 1024 messages. This value may be modified with the RtMidiIn::setQueueSizeLimit() function. If the maximum queue size limit is reached, subsequent incoming MIDI messages are discarded until the queue size is reduced.
In the following example, we omit some necessary error checking and details regarding OS-dependent sleep functions. For a more complete example, see the
qmidiin.cpp program in the
When set, a user-provided callback function will be invoked after the input of a complete MIDI message. It is possible to provide a pointer to user data that can be accessed in the callback function (not shown here). It is necessary to set the callback function immediately after opening the port to avoid having incoming messages written to the queue (which is not emptied when a callback function is set). If you are worried about this happening, you can check the queue using the RtMidi::getMessage() function to verify it is empty (after the callback function is set).
In the following example, we omit some necessary error checking. For a more complete example, see the
cmidiin.cpp program in the
The Linux ALSA, Macintosh CoreMIDI and JACK APIs allow for the establishment of virtual input and output MIDI ports to which other software clients can connect. RtMidi incorporates this functionality with the RtMidiIn::openVirtualPort() and RtMidiOut::openVirtualPort() functions. Any messages sent with the RtMidiOut::sendMessage() function will also be transmitted through an open virtual output port. If a virtual input port is open and a user callback function is set, the callback function will be invoked when messages arrive via that port. If a callback function is not set, the user must poll the input queue to check whether messages have arrived. No notification is provided for the establishment of a client connection via a virtual port.
In order to compile RtMidi for a specific OS and API, it is necessary to supply the appropriate preprocessor definition and library within the compiler statement:
|OS:||MIDI API:||Preprocessor Definition:||Library or Framework:||Example Compiler Statement:|
|Linux or Mac||JACK MIDI||__UNIX_JACK__|
|Macintosh OS X||CoreMIDI||__MACOSX_CORE__|
|Windows||Multimedia Library||__WINDOWS_MM__||compiler specific|
The example compiler statements above could be used to compile the
midiprobe.cpp example file, assuming that
RtMidi.cpp all exist in the same directory.
If you are having problems getting RtMidi to run on your system, try passing the preprocessor definition
__RTMIDI_DEBUG__ to the compiler (or define it in RtMidi.h). A variety of warning messages will be displayed that may help in determining the problem. Also try using the programs included in the
tests directory. The program
midiprobe displays the queried capabilities of all MIDI ports found.
Support for each MIDI API is encapsulated in specific MidiInApi or MidiOutApi subclasses, making it possible to compile and instantiate multiple API-specific subclasses on a given operating system. For example, one can compile both CoreMIDI and JACK support on the OS-X operating system by providing the appropriate preprocessor definitions for each. In a run-time situation, one might first attempt to determine whether any JACK ports are available. This can be done by specifying the api argument RtMidi::UNIX_JACK when attempting to create an instance of RtMidiIn or RtMidiOut. If no available ports are found, then an instance of RtMidi with the api argument RtMidi::MACOSX_CORE can be created. Alternately, if no api argument is specified, RtMidi will first look for JACK ports and if none are found, then CoreMIDI ports (in linux, the search order is JACK and then ALSA. In theory, it should also be possible to have separate instances of RtMidi open at the same time with different underlying API support, though this has not been tested.
RtMidi is designed to provide a common API across the various supported operating systems and audio libraries. Despite that, some issues should be mentioned with regard to each.
RtMidi for Linux was developed using the Fedora distribution. Two different MIDI APIs are supported on Linux platforms: ALSA and JACK. A decision was made to not include support for the OSS API because the OSS API provides very limited functionality and because ALSA support is now incorporated in the Linux kernel. The ALSA sequencer and JACK APIs allows for virtual software input and output ports.
The Apple CoreMIDI API allows for the establishment of virtual input and output ports to which other software applications can connect.
The RtMidi JACK support can be compiled on Macintosh OS-X systems, as well as in Linux.
configure script provides support for the MinGW compiler.
The Windows Multimedia library MIDI calls used in RtMidi do not make use of streaming functionality. Incoming system exclusive messages read by RtMidiIn are limited to a length as defined by the preprocessor definition RT_SYSEX_BUFFER_SIZE (set in RtMidi.cpp). The default value is 1024. There is no such limit for outgoing sysex messages via RtMidiOut.
RtMidi was originally developed with Visual C++ version 6.0 but has been tested with Virtual Studio 2010.
In years past, the following people provided bug fixes and improvements:
RtMidi: realtime MIDI i/o C++ classes<BR> Copyright (c) 2003-2016 Gary P. Scavone Permission is hereby granted, free of charge, to any person obtaining a copy of this software and associated documentation files (the "Software"), to deal in the Software without restriction, including without limitation the rights to use, copy, modify, merge, publish, distribute, sublicense, and/or sell copies of the Software, and to permit persons to whom the Software is furnished to do so, subject to the following conditions: The above copyright notice and this permission notice shall be included in all copies or substantial portions of the Software. Any person wishing to distribute modifications to the Software is asked to send the modifications to the original developer so that they can be incorporated into the canonical version. This is, however, not a binding provision of this license. THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
|©2003-2016 Gary P. Scavone, McGill University. All Rights Reserved.|
Maintained by Gary P. Scavone, gary at music.mcgill.ca