[Cbor] Re: I-D Action: draft-mcnally-deterministic-cbor-10.html

[Anders Rundgren <[e-mail]>]

The only thing that remains clear, is that anybody wanting to understand the rationale for numeric reduction would have to wade through literally hundreds of pretty voluminous messages, typically mixing things actually are unrelated, like "you can build dBCOR on top of CDE".

Anders

On 2024-07-31 08:13, Carsten Bormann wrote:
> On 31. Jul 2024, at 07:48, Anders Rundgren <[e-mail]> wrote:
>>
>> - No scientific proof of the superiority of numeric reduction from a determinism point of view.
>>
>> - No down-to-earth example showing how numeric reduction would aid the targeted applications.
> 
> 
> - No scientific proof of the superiority of numeric reduction from a determinism point of view THAT I UNDERSTAND.
> 
> - No down-to-earth example showing how numeric reduction would aid the targeted applications THAT I UNDERSTAND.
> 
> FTFY.
> 
> The argument has been made here, you just chose not to follow it.
> (Yes, the argument is based on an application that you may not have; this doesn’t make it less convincing.)
> 
> (I’m not sure what “scientific proof” means; I’d substitute “technically sound argument”.)
> 
> I think the more important observation is that you don’t have to fight against these applications; they just aren’t yours.
> And, that it’s not CDE vs. dCBOR — dCBOR contains CDE as a layer, so it is actually “no additional layer” vs. “add layer that helps mapping certain numeric application data types to numeric CBOR data types in a deterministic fashion, so CDE can do its thing on the latter”.
> 
> (After last week’s bizarre discussion about layers in a different place, maybe I should use this term less.  But it is really useful...)
> 
> Grüße, Carsten
>