Re: [arch-d] draft-iab-protocol-maintenance / JSON feedback

Also, a little bit of IETF self-examination is called for.

Let me quote from the datatracker history of what became RFC4627:

"I had to read quite a bit of the document before I realized that this document basically is just requesting a MIME-type registration and has good documentation to go with it first." (from Bert Wijnen)

The IETF didn't really review the original JSON spec as such. We reviewed a MIME type registration and concluded that the associated specification was sufficient. That doesn't say much about the robustness principle. It wasn't reviewed as a protocol, just as a data format, and it wasn't a standards track document.

(For the record: I was IESG Chair at the time and I balloted No Objection.)

Personally, if I ever have to write to code to receive and interpret JSON, I will strive to make it as tolerant as possible, since it seems likely that different sources will produce different variants. That doesn't mean that the existence of such variants is desirable, but any implication that people *designed* those variants in the hope that a receiver would apply the robustness principle seems wrong to me. I think Carsten's history makes that clear.

Regards
    Brian Carpenter

On 19-Jul-22 02:05, Paul Hoffman wrote:
> Carsten's detailed history of JSON (which I fully agree with, having
> been there for it in different roles) again points out the difficulty of
> draft-iab-protocol-maintenance: many of the definitive statements in it
> about what worked/works and what didn't/doesn't are easy to get wrong
> due to limited viewpoints. Having bad examples weakens the overall goal
> of the document, and many readers might be turned off to the document
> once they hit the second example they disagree with.
> 
> --Paul Hoffman
> 
> _______________________________________________
> Architecture-discuss mailing list
> Architecture-discuss@ietf.org
> https://www.ietf.org/mailman/listinfo/architecture-discuss
> .