I asked for help on comp.lang.python
with the implementation of the 8 bit codec that I introduced in the
previous posting.
The question triggered a very helpful
discussion that made me change the proposed encoding algorithm.
Part of the discussion covered the
question why I would want to use such an obscure algorithm.
There are many algorithms available,
each with their own merits.
The answer is that I am looking for the
algorithm that requires the minimal number of bits to encode
arbitrary integer values.
None of the currently available
algorithms does that.
Dave Angel showed me that the algorithm
I proposed didn't either.
So I went back to the drawing board
with the objective to find something better then the MIDI algorithm
he used to show me.
This time I tested the new proposal
more thoroughly myself.
It beats the MIDI encoding for every
character size and it beats the the first proposal for character
sizes larger then 2.
At character size 2 the two proposals
do compete with each other, depending on the integer value.
For now I consider the new proposal to
be my best option.
I will present it in the next posting
and I hope that it may be improved with your help.
Geen opmerkingen:
Een reactie posten