Re: [jose] Performance of Canonical JSON

On 2018-12-07 19:25, Bret Jordan wrote:
> Thanks for the test data. Are you going to request a side meeting for Prague?

That's a good question. Since few have read the existing (somewhat constrained), serialization-only proposal [1], I'm considering other alternatives [4,5]

As you know REST [2] is currently held as the only "real" way architecting Web applications.  In spite of that, there is no standard for signing REST requests but signed REST requests are still used in the wild including by Amazon [3].

A rebooted JSON WG would likely settle on a full-fledged counterpart to XML's "CN14" which I have no interest in because it presumes that the canonicalization process is schema driven for both parsing and serialization, making deployment much more complex.

Anyway, IF there actually is GENUINE interest in a BoF session in Prague, could you guys on the list indicate your interest?

Thanx,
Anders

1] https://tools.ietf.org/html/draft-rundgren-json-canonicalization-scheme-01

2] https://en.wikipedia.org/wiki/Representational_state_transfer

3] https://docs.aws.amazon.com/general/latest/gr/sigv4-create-canonical-request.html

4] https://www.rfc-editor.org/about/independent/

5] Ignoring the standards process and rather let associated applications like https://cyberphone.github.io/doc/two-visions-4-mobile-payments.pdf set a de-facto standard.

> 
> Bret
> 
> Sent from my Commodore 128D
> 
> PGP Fingerprint: 63B4 FC53 680A 6B7D 1447  F2C0 74F8 ACAE 7415 0050
> 
> On Dec 7, 2018, at 9:23 AM, Anders Rundgren <anders.rundgren.net@gmail.com <mailto:anders.rundgren.net@gmail.com>> wrote:
> 
>> Since XML Canonicalization has a reputation of not only being brittle but also terribly slow, I tested JCS [1] with the following JSON file:
>> {
>>   "1": {"f": {"f": "hi","F": 5} ,"\n": 56.0},
>>   "10": { },
>>   "": "empty",
>>   "a": { },
>>   "111": [ {"e": "yes","E": "no" } ],
>>   "A": { }
>> }
>>
>> Expected output: {"":"empty","1":{"\n":56,"f":{"F":5,"f":"hi"}},"10":{},"111":[{"E":"no","e":"yes"}],"A":{},"a":{}}
>>
>> Since JCS only is a serialization concept (parsing is unaffected), I compared the execution speed of standard serialization versus canonicalized serialization.
>>
>> Using https://www.npmjs.com/package/canonicalize the performance penalty was about 2.4 compared to JSON.stringify().
>> Using my homegrown JSON tools written in Java having an integrated "canonicalize" serializer option the performance penalty was about 1.4
>>
>> Anders
>>
>> 1] https://tools.ietf.org/html/draft-rundgren-json-canonicalization-scheme-01
>>
>> _______________________________________________
>> jose mailing list
>> jose@ietf.org <mailto:jose@ietf.org>
>> https://www.ietf.org/mailman/listinfo/jose