yEnc v1 uses a fixed offset of 42. This is meant to move characters that tend to occur fewer times on average to the critical values. The result should be that less critical values will need to be escaped. (Each time a critical value occurs it needs to be escaped, adding an extra character in the output increasing the file size).

While having a fixed offset makes the encoding simple to understand and implement, it's obvious that this technique can only hope to perform a little better than average, at best.. and not at all at worst.

Having a variable offset would mean that no more escaping is done than is necessary for each file, ALWAYS. This may not not mean much improvement on average, but as good as eliminates the possibility of the worst case scenario.

Having a variable offset would be easy to implement: just add a "offset=xxx" parameter to the =ybegin line. Calculating the best offset is very easy too.

Here are some examples of file-scans to calculate the best offset. The bar-graph displays how often each byte value (0...255) occurs. Note: The red line is the value 0, the two purple lines are the values 10 and 13. The light green and two darker green lines are the values that are calculated to be the best best values to be mapped onto the afore mentioned critical values. Note: the scale is different on all pics!

JPEG CAB ALL WAV