>> In total: >> >> 3 bits: opcode of first frame >> 1 bit: continuation state >> 4 bit: UTF-8 DFA state >> ============ >> >> 1 octet state > > forgot of course: > > for servers: > > client mask (4 octets) > last frame end % 4 : 2 Bits => to know where within mask to start for > unmasking the next masked frame received
Server shouldn't need to remember this, since each fragmented frame comes with its own masking key. At least that's how I read the spec; there's nothing in Section 5.4 that seems to indicate that masking keys get carried over from one fragment to another. > Plus any extension state as Björn already mentioned. > > Whatever. There is state anyway. WS is not stateless. Yep. I was just talking about state related to UTF-8 processing. --Richard _______________________________________________ Gen-art mailing list [email protected] https://www.ietf.org/mailman/listinfo/gen-art
