New for 1.3.2 (2007-11-04)

   (class MTBuffer and descendants)
   o Leopard fix for incorrect use of NSConditionLocks (now uses POSIX locks and
     condition variables -- NSCondition would have been nice but doesn't work
     correctly pre-10.5)
     
   (sample app AudioMonitor)
   o update project to Xcode
   
   (misc)
   o build universal framework and universal AudioMonitor


New for 1.3.1

   (class MTCoreAudioDevice)
   o notifications (such as MTCoreAudioHardwareDeviceListDidChangeNotification) will
     not be sent before the first use of class MTCoreAudioDevice

   (class MTBuffer and descendants)
   o new method -close
   o new method -configureForSingleThreadedOperation
   
   (sample app AudioMonitor)
   o made MTConversionBuffer class better for general buffered sample rate conversion
   o new class MTVarispeedConversionBuffer.  uses Varispeed and Converter AUs
   
   (misc)
   o new utility function MTAudioBufferListSetFrameCount()
   o new value transformers (between linear and decibels) for use with Cocoa Bindings


New for 1.3.0

   (class MTCoreAudioDevice)
   o new method +devicesWithName:havingStreamsForDirection:.
   o -description gives useful information.
   o new methods +allDevicesByRelation and -relatedDevices.
   o new delegate method -audioDeviceSomethingDidChange:.
   o new method -deviceSafetyOffsetFramesForDirection:.
   o new method -setDevicePaused:.
   
   (class MTCoreAudioStream)
   o -description gives useful information.
   
   (class MTCoreAudioStreamDescription)
   o support for localized stream description strings (see StreamDescription.strings
     and -bundleForDescriptionStrings) (only English so far, please feel free to
     submit localizations for other languages)
   o new methods -isLinearPCMFormat, -isCanonicalFormat, -isInterleaved, -setIsInterleaved:.
   o -setChannelsPerFrame: and -setIsInterleaved: will update bytesPerPacket and bytesPerFrame
     if the format is canonical (native, interleaved or de-interleaved).
     
   (class MTCircularQueue/MTByteBuffer)
   o class MTCircularQueue has been renamed to MTByteBuffer
   o initializer renamed to -initWithCapacity:
   o method -bytesInQueue renamed to -count
   o method -size renamed to -capacity
   o methods -writeBytesFrom:length: and -writeBytesWithoutBlockingFrom:length: combined
     to -writeFromBytes:count:waitForRoom:
   o methods -readBytesTo:length: and -readBytesWithoutBlockingTo:length: combined
     to -readToBytes:count:waitForData:
   o reimplemented as a subclass of MTBuffer
   o finally documented!
   
   (new class MTAudioBuffer)
   o similar in operation to MTByteBuffer, only storing N-channel Float32 frames
     instead of bytes
   o reads from and writes to AudioBufferLists
   
   (new class MTBuffer)
   o abstract class intended to be subclassed to create thread-safe, flow-controlled
     buffers, such as MTByteBuffer and MTAudioBuffer

   (sample app AudioMonitor)
   o changed MTConversionBuffer class to use an MTAudioBuffer instead of MTByteBuffer
   
   (misc)
   o fixed a bug where NULL CFStringRefs obtained from a successful
     HAL query would be used as if they were valid, causing an exception.
   o enabled prebinding for framework (with -seg1addr 0xB0000000).
   o removed internal class MTObjectProxy after i noticed that +[NSValue valueWithNonretainedObject:]
     does the same thing.  :)
   o new utility functions for AudioBufferLists


New for 1.2.0

   (class MTCoreAudioDevice)
   o -deviceStart now lets you know if the device started or not.
   o new delegate method -audioDeviceStartDidFail:forReason:
   o new method -deviceMaxVariableBufferSizeInFrames.
   o new method -supportsNominalSampleRate:.
   
   (class MTCoreAudioStreamDescription)
   o -set... methods return the receiver, so sets can be chained.
   o new methods +nativeStreamDescription, -setNativeFormat and -isNativeFormat.
   o -setChannelsPerFrame: now also updates bytesPerFrame and
     bytesPerPacket if the stream description is in native format.
   o conforms to <NSCopying>
   
   (sample app AudioMonitor)
   o totally changed input->output copying to deal with
     different sample rates, numbers of streams, etc.