Ensure that the USB_DeviceState variable is properly set upon wakeup events, to the Addressed or Configured state as needed.

Add explicit support for the AT90USB646 to the Teensy bootloader now that PJRC have released the Teensy++ board.

Ensure unfinished SideShow demo uses an unallocated PID value within the LUFA range.

Changed USB_Init() and USB_Shutdown() so that they are no longer dependant on oneanother for possible code savings when the interface is never explicitly shut down.

Added new Endpoint_ClearStatusStage() convenience function to assist with the status stages of control transfers.

Removed vague USB_IsConnected global - test USB_DeviceState or USB_HostState explicitly to gain previous functionality.

Removed USB_IsSuspended global - test USB_DeviceState against DEVICE_STATE_Suspended instead.

Fixed possible enumeration errors from spinloops which may fail to exit if the USB connection is severed before the exit condition becomes true.

Add extra tests to the MassStorage device demo and class driver for validating command blocks from the host.

Add explicit guards to all device mode tasks to ensure the device is connected and configured before running the task, to prevent any user tasks from locking up the main USB task if the device has not been properly configured.

Change Doxygen configuration files to not produce 4096 directories for the documentation files -- the LUFA components do not generate enough documentation files to justify the large performance hit of creating so many directories.

Fix missing Doxygen documentation in the Class Driver Mass Storage Device demo.

Fixed PrinterHost demo Printer_GetDeviceID() routine not removing the Device ID string length from the start of the returned array (thanks to John Andrews).

Fixed error in new pipe stream function template system not setting the right device token for R/W operations (also thanks to John Andrews).

Remove ClearPipeStall routine from the StillImageHost demo, as there is now a version of the function built into the library itself.

Cleaner solution to the PrinterHost data send routine problem of embedded NULLs - use a special structure to hold the data string plus the length in bytes of the data.

Fix unfinished printer host demo - some printer languages use embedded NULLs and so strlen() won't work in all cases. Ensure NULL terminator for the test page is not transmitted to the device.

Don't send terminal RESET codes via serial - just set the foreground colour and assume the user has the background set to something sane, so that the demos don't wipe out custom user terminal settings.

Use sane ANSI terminal escape codes for the Host mode demos, so that they do not cause eye-searing, unreadable text on terminals which parse the escape codes properly.

Change over Printer_GetDeviceID() to require a pointer to the destination buffer plus the buffer size, rather than using a pointer to a special structure.

Make new Printer_SendData() function to hide the implementation of sending data to an attached printer, cleaning up the main demo source file body.

Fix PrinterHost demo so that it will only enumerate printers with Bidirectional protocol encapsulation. Change enumeration code to automatically select the correct alternate setting for the printer interface to select the bidirectional protocol.

Fixed incorrect PIPE_EPNUM_MASK mask causing pipe failures on devices with endpoint addresses of 8 and above (thanks to John Andrews).

Magstripe Project: Ensure that empty tracks still print out a newline seperator so that the host always knows what track data is being sent.

Updates to PrinterHost demo to include some PCL test data plus fixes to the GetDeviceID routine.

Fixed error in GenericHID descriptors preventing it from passing the USB-IF HID tests (thanks to Søren Greiner).

Added new TOTAL_NUM_CONFIGURATIONS option, removed USE_SINGLE_DEVICE_CONFIGURATION compile time option (but silently convert it to USE_SINGLE_DEVICE_CONFIGURATION internally for compatibility).

Added new USE_FLASH_DESCRIPTORS compile time option. By default, descriptors can now lie in mixed memory spaces (specified by a new parameter to the CALLBACK_USB_GetDescriptor() function) unless one of the USE_*_DESCRIPTORS compile time option is specified.

Fix HID class device driver -- if a SetIDle request is issued with the LSB of wValue set to zero, the idle period must be set for all HID interfaces.

Fix Keyboard and Mouse demos, Idle period is now multiplied by 4 as the period is read into and sent out of the device to ensure it is always stored as a multiple of 1ms. Fixes Keyboard demo using an initial Idle period of 2s rather than 500ms (thanks to Brian Dickman).

Move out the internal device serial descriptor reading routine into a seperate static function, rather than being part of USB_Device_GetDescriptor.

Added new EEPROM and FLASH buffer versions of the Endpoint and Pipe stream functions. Changed Endpoint.c and Pipe.c to use a templated system to build the seperate functions, rather than duplicating each function's code many times.

Documented FAST_STREAM_TRANSFERS compile time option.

Reduced the TCP window size for the RNDIS demos (Class and LowLevel) to make them compatible with the AT90USB64x.

Added new HOST_STATE_WaitForDeviceRemoval host state machine state for non-blocking disabling of device communications until the device has been removed (for use when an error occurs or communications with the device have completed). Changed over all host mode demos to use the new state.

Added verbose documentation for each of the USB Host state machine states.

Fixed Host mode to Device mode UID change not causing a USB Disconnect event when a device was connected.

Seperated out Lib components of the incomplete BluetoothHost demo application out into a seperate Lib subfolder.

Changed F_CLOCK entries in project makefiles to alias to F_CPU by default, as this is the most common case.