Re: Working Group Last Call: Structured Fields Revision (RFC8941bis)

On 30.04.2023 15:52, Poul-Henning Kamp wrote:
> --------
> Mark Nottingham writes:
>> Personally --
>> *If* we think we're going to ship Display Strings and can agree to do
>> so, I'd prefer to do it in this revision, rather than having a third
>> revision of Structured Fields.
> I almost agree, only I think it is better to put a stop to this horrible
> idea once and for all, rather than kick the can down the road.
> As for consensus:  I am firmly against.
> With reference to:
> 1. sf-binary already does everything this proposal claims to do, only better & safer.

That is incorrect. It labels the item as string, and defines the
encoding. And it's more compact on the wire if the field is mostly ASCII.

If we would follow your reasoning, we should remove sf-date (which does
not add anything over integers) and consequently not update the spec at all.

> 2. I have still not seen a non-speculative use-case where this is beneficial.

Many have been raised.

> 3. "HTTP is the new TCP" and just we never added I18N to TCP headers, we should never
>     do it to HTTP fields either, because it is the wrong protocol layer for it.

I'd say that's an opinion.

> 4. The escape rules are inconsistent:
>  %  -> %25
>  " -> \"
>  \ -> \\

If that's still the case, we need to fix it.

> 5. There is no negotiation mechanism for the necessary cultural (p)reference.


> and so on...
> No, just NO!
> Poul-Henning

Best regards, Julian

Received on Sunday, 30 April 2023 15:00:00 UTC