Mon Aug 26 23:16:58 EDT 2019

Different sync encoding

Pick one special byte to act as packet boundary.  This leaves 255 values.

A n in m encoding would satisfy 

255^m >= 256^n

basically, encode in base 255 instead of base 256.

The midi 7 in 8 encoding is a special case of that, leaving 128
control characters (e.g. with high bit set).

Is there a natural chunk size to do that?

Say that we only use 64 codes.  This is 0.5 bits per byte that are
missing.  Can this be 15 in 16?

Or can it be 255 in 256?

Too tired and wacky brained to do this intuitive shit..

At the bit level, bit stuffing can be used.  This can also be done in
uart bytes.  If the last bit is 0, this was a 0xFF byte, and the
entire stream should be shifted by one.

That would be more efficient, but requires a more complex decoder.