LanX:

You're not missing anything that I know of. What I was basing my "not quite" phrasing on is the idea of using a single character to encode each group (@c) into a character, so it would use 9 characters (72) bits. Had I thought of just packing the required 51 bit records together, it would be more than sufficient to get 50% compression, as the file would take 635 bytes to encode 100 records (sans newlines).

(The 51 bits came from: 50 different possibilities for each group of 10 in the inner loop (log2(50) == 5.64.. bits/group) * (9 groups) == 50.79 bits.)

...roboticus

When your only tool is a hammer, all problems look like your thumb.