Re: [apps-discuss] Concise Binary Object Representation (CBOR)

[Nico Williams <[e-mail]>]

On Thu, May 23, 2013 at 11:38 AM, Carsten Bormann <[e-mail]> wrote:
> On May 23, 2013, at 18:13, Nico Williams <[e-mail]> wrote:
>
>> Also, I believe this is the fifth binary encoding of JSON proposed
>> thus far.
>
> 1) CBOR is not a "binary encoding of JSON".
> (It can be used as one, though.)

The superset of JSON in CBOR is not in itself appealing enough, I
think.  We already have a *lot* of binary encodings.

It's only the connection to JSON that makes this appealing.  You might
disagree, but you're probably biased :)  You're asking that we
standardize this though, so you should do something to justify one
more encoding in a sea of encodings.

> 2) There are many more proposals in this space, and it is hard to draw a line.
> E.g. Google protocol buffers is a prominent example.

But it requires a schema.  (And what the heck was so wrong with PER?
But whatever.)

>> An analysis of these might be nice.
>
> Yes.  Find a grad student :-)

I think it falls on you to provide some justification for this.  It
wouldn't be hard, but you might find that an existing, *deployed*
encoding is good enough (or maybe you wouldn't but we would).  I think
I'd rather standardize a good-enough encoding that's been deployed
than create a new one, but maybe the improvements CBOR brings are
compelling enough.  Either way you should definitely provide some
analysis.

> Of course, I did my own analysis, but I didn't see a need to write it up.

That's... odd.  You need consensus.  Why wouldn't your analysis help you get it?

> (More specifically, I didn't see a funding agency interested in this part of reality.
> But I'm willing to be surprised :-)

Do you intend to use CBOR for any protocols you want standardized at
the IETF?  Or are you just hoping others will?

Nico
--