[video-codec] Lessons Learned from Audio Codec Group - Normative Text not Code

["Cullen Jennings (fluffy)" <[e-mail]>]

The audio codec WG tried many things that were sort of new for the IETF and some of theses worked and some did not. Having co-chaired that WG, there are some things I would like to comment on if we were to do something like the codec WG again. Note that on almost all of theses, there is someone who can tell me "I told you so", and to all theses people, I owe you a beverage :-) 


The Normative spec should be Text not Code

Code works many places and has many advantages but the IETF process for reviewing does not work well for reviewing large blocks of code.  We should use the text in the draft, not code as the normative spec. It's great if some site happens to have some running code but that is not the spec. It will help find bugs in the spec and in the long run, running, interoperable code is the important thing. But the normative conceptual design should be in the text of the draft with enough detail that someone that was incredibly smart, had lots of time, and with access to all the normative references in the draft, could implement and interoperable version of the codec. 

Send comments - tell me I am right - tell me I am wrong - whatever - the grass is definitely greener on the other side. But lets sort this one out before the BOF.