I've come up with a better way of encrypting the owner-id information, so I'd like to hold up any release till I put it in. I'll try to do it tomorrow.
The current setup XOR's the owner-id with the beginning of each zlib-compressed segment. This makes it (cryptographically) fairly easy to get the owner-id out of the document. A better scheme would be to construct a sequence of hash bits from the owner-id and use them to XOR the segments. This could have a more predictable length, as well. Bill
