Fix incorrect project name in the Projects/ folder makefile.

Copy out the EndpointStream and PipeStream functions to each architecture, so that each architecture can optimise the stream functions according to the available hardware (such as DMA, where available).

Lower the size of the HID report endpoint from 16 bytes to 8 bytes, to allow for Low Speed USB compatibility in the KeyboardMouseMultiReport demo (thanks to Kevin Chernoff).

Update MediaController project to remove the Record usage (not often needed) and explicitly declare each usage for maximum clarity.

Add new MediaController project, a simple Media Playback control manipulator (Play/Pause/Next Track/Volume/etc.).

Correct errors in the DHCP Server application in the Webserver project, that was causing random server restarts and/or incorrectly allocated IP addresses.

Add RNDIS Device support to the Webserver project, so that the files stored on the device can be viewed from a web-browser while the board is plugged into a host machine.

Add new RNDIS_Host_IsPacketReceived(), RNDIS_Device_ReadPacket() and RNDIS_Device_WritePacket() functions to the Device RNDIS Class Driver.

Modify RNDIS demos to suit the simplified Ethernet_Frame_Info_t structure.

Add new "version" makefile target to the main LUFA library makefile, to give an easy way to identify a LUFA release from the command-line.

Fixed possible invalid program execution when in host mode if corrupt descriptor lengths are supplied by the attached device.

Minor code cleanups to add const and reformat where missing, as well as abstract out the internal device signature start address into a macro, so that it can be altered to suit particular devices within a single architecture if needed.

Add missing documentation to the USB_Device_States_t enum.

Add missing LEDS_ToggleLEDs() function to the LED driver for the EVK1101 board.

Add __VA_ARGS__ support to the LUFA supplied ISR macro. Add proper result typecasting to the SWAPENDIAN_* macros.

Switch to using -1 on the UC3 target to obtain a register mask with all bits set (for clearing interrupts and status flags).

Fix incorrect USB controller mode on the UC3 when a fixed mode is specified as a compile time option due to AVR32_USBB.USBCON.uide being set by default.

Make USB_Descriptor_String_t use a uint16_t for Unicode strings on all targets except the AVR8 (retained for backwards compatibility).

Fix HID class bootloader not working on the larger USB AVRs due to the altered page addressing scheme.

Make DFU and CDC class bootloaders also toggle the LEDs on command activity in addition to the periodic flashing.

Replace cast-as-char* byte access of multibyte variables with proper shifts and masks to preserve endianness across different architectures.

Slight cleanups to the architecture specific makefile templates to make the common configuration sections as identical as possible.

Update architecture specific makefile templates so that they are as close to one-another as possible.

Perform endianness correction in the HID report parser for big-endian platforms.

Remove redundant type information for bitfield elements, other than the signed/unsignedness of the element.

Change type of USB_SelectedPipe and USB_SelectedEndpoint for the AVR32 UC3 architecture to uint32_t to reduce the compiled code size.

Replace all calls and references to _delay_ms() in the code with the architecture-agnostic Delay_MS() function.

Improve code generation for the Delay_MS() function on the AVR8 architecture when called with a constant input.

Remove use of avr-libc specific ATOMIC_BLOCK, replace with a new per-architecture set of inline functions to retrieve and manipulate the global interrupt enable bit for each architecture.

Add in documentation for the USB controller common interrupt routine which must be linked to the interrupt controller in the user application on the AVR32 UC3 architecture.

Renamed all low level Endpoint_Read_*, Endpoint_Write_* and Endpoint_Discard_* functions to use the number of bits instead of a symbolic size (Byte, Word, DWord) so that the function names are applicable and correct across all architectures.

Renamed all low level Pipe_Read_*, Pipe_Write_* and Pipe_Discard_* functions to use the number of bits instead of a symbolic size (Byte, Word, DWord) so that the function names are applicable and correct across all architectures.

Conditionally add available address spaces to the address space enum, for multiple address space architectures.

Update Doxygen configuration file for the latest Doxygen release.

Minor documentation updates.