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

[Wolf McNally <[e-mail]>]

Anders,

> On Jul 30, 2024, at 1:46 PM, Anders Rundgren <[e-mail]> wrote:
> 
> Hi Christopher,
> 
>> On 2024-07-30 22:22, Christopher Allen wrote:
>> On Mon, Jul 29, 2024 at 11:21 PM Anders Rundgren <[e-mail] <mailto:[e-mail]>> wrote:
>>    On 2024-07-29 15:49, Joe Hildebrand wrote:
>>     > - I'm still not sold on bidirectional diagnostic notation, and have not planned on writing a parser for it, but I'm still thinking about that.  However, I don't see what it has to do with this discussion.
>>    Well, it has.  This feature permits writing and reading fully compatible CDE from external text based databases [1].  Gordian Envelopes rely on a similar feature but using dCBOR.
>> To be clear, we do have something called Envelope Notation, but it is not bidirectional. It is designed to make the envelope structures clear for debugging purposes only.
>> ```
>> "Alice" [
>>     "knows": "Bob"
>> ]
>> ```
>> or with --tree
>> ```
>> e54d6fd3 NODE
>>     27840350 subj "Alice"
>>     55560bdf ASSERTION
>>         7092d620 pred "knows"
>>         9a771715 obj "Bob"
>> ```
>> We do have a `dcbor-cli` app
>> https://github.com/BlockchainCommons/bc-dcbor-cli <https://github.com/BlockchainCommons/bc-dcbor-cli> that supports basic CBOR diagnostic output, but I would not say that it is a full implementation of latest notation specs — again, our purpose is not round trip but to enable clarity of structures for debugging. I'm not even sure we test round-tripping.
> 
> Thanx!
> 
>> An important point here is that round-tripping of binary to text and back can actually get in the way of useful debugging support for software engineers.
> 
> FWIW, this ability is why I prefer CDE.  If you write 2.0 in the left window of https://cbor.me and press the -> button it will come out as CBOR hex in the right window.  If you then push the <- button it will still render as 2.0.  Using the dCBOR option breaks this scheme.

It actually depends on what you mean by “this scheme”, and whether you think the concept of “two” should have many different encodings. For the purpose of supporting determinism for most common purposes we have identified, we don’t think that conceding the semantics of numbers to specifics of processor architectures (any more than is absolutely necessary) is as desirable as you clearly do.

In the case of the CBOR playground used with the dCBOR option on, I would argue that the semantics round-trip perfectly for every purpose we care about, and *that* is what matters. There is absolutely nothing broken there.

It is on you to argue why deterministic encoding of identical semantics for such a common data type as “number” should be a secondary consideration, and you have so far failed to do this.

~ Wolf