Imported from previous forum
I think we should modify the definition of the newly introduced excess-1 encoding of subtraction lengths to be asymmetric in the same sense that the excess encoding of nullable integers is asymmetric.
In the current writing, the absolute value of both positive and negative subtraction lengths is incremented by one as per Yuriys original proposal. This leaves us with the encoded value 0 undefined (or at least nonsensical: “remove -1 characters from the back”). So I suggest that we only increment the absolute value of negative subtraction lengths (or put shorter: decrement negative subtraction lengths by one). This way the encoded 0 would mean “remove zero characters from the back” and -1 would mean “remove zero characters from the front”.
Best regards,
David