Mwahahahaaa! This is going to be absolutely wonderful. Soon, the internal interface of all of my DSP plugins used by the port-specific code is going to be able to read from and write to 32 & 64-bit float data as well as 8, 16, 24 & 32-bit integer (instead of merely the former) directly, instead of pre and post-conversion and use of extra temporary buffers being neccessary when the input is in integer format. Each processing loop will automatically become able to handle all formats, as well as reading and writing mono signals, stsreo data stored in separate buffers, and interleaved buffers. The input and output formats will be independently set and handled, so every processing function could in addition to performing processing work as a format converter. If the input and output data is set to the same location, it will even be possible to do in-place, as long as the output format is not larger in storage size than that of the input.
And the presence in itself of these new capabilities will cost exactly 0 CPU cycles during processing. Long live the evils of self-modifying code!