[rfc-i] Re: Normative ABNF [was Re: Re: Normative information in RFC imagery]

[Marc Petit-Huguenin <[e-mail]>]

On 5/20/25 7:50 AM, Paul Kyzivat wrote:
> On 5/19/25 11:55 PM, Julian Reschke wrote:
>> On 17.05.2025 16:32, Paul Kyzivat wrote:
> 
>>> I've encountered this style before while doing genart reviews.
>>> It does seem to precisely define they syntax. But it takes an order of magnitude more text than equivalent ABNF would.
>>>
>>> I fail to grasp any benefits that this provides. Are there any?
>>
>> It makes implementing easier, in case you're willing to be *told* how to do things.
> 
> I think it is the business of IETF to tell implementers *what* to implement, not *how*. We should be specifying the minimum constraints necessary to ensure interoperability. There is a place for implementation notes - a non-normative appendix.
> 
> In my many years as an implementer, I objected to being told *how*. That was my job, and I wanted the freedom to come up with the best approach within prevailing constraints.
> 

Yes, that is what I meant when I say that specifications should be descriptive, as opposed to algorithms being prescriptive.  On the one hand, I like that RFC 9651 is no longer using normative ABNF, but on the other hand, I dislike even more normative algorithms in RFCs.

-- 
Marc Petit-Huguenin
Email: [e-mail]
Blog: https://medium.com/@petithug
Profile: https://www.linkedin.com/in/petithug