Rationale

Researchers have been working for many years on formal specification of protocols. Many languages or systems have been developed. Today, for various reasons (some of them being good reasons), these language or systems are not used in RFCs. And there is no reason to believe that the IETF can do better than what these researchers did.

For the description of state machines ("description" and not "specification", which would be more ambitious), there was was two choices:

• use a very complete language, able to describe most or all of the protocol,
• use a very simple language, able to do only one thing (describe the structure of the state machine) but trying to do it well. And let most of the protocol being described in natural language, not in Cosmogol.

The second solution was choosen, which looked like more in the spirit of IETF.

So, many things are missing for Cosmogol such as timers (see the SIP state machine in RFC 3261, which makes an heavy use of timers) or protocol behavior like TCP slow start. It is not a bug, it is a feature.

Cosmogol is a language to describe finite state machines. It is currently under study at the IETF. Web site maintained by Stéphane Bortzmeyer. © S. Bortzmeyer. Hosted by AFNIC.