Re: [Json-canon] JSON Subtype handling. Re: Review of draft-rundgren-json-canonicalization-scheme-03

[Anders Rundgren <[e-mail]>]

On 2019-02-09 06:10, Jim Schaad wrote:
> Anders,
> 
> Two items.  First note that I changed the address of the IETF mailing list as the "-request" should be omitted.  This sends mail to the administrator not to the list.

Sorry, it was just a carbon copy of the original list of recipients..

> Second.  I like the text that you have placed in Appendix E as it seems to do a good job of both describing the problem and providing a solution.

Thanx!

> The problem is that it provides two different solutions and the reference from section 3.1 does not state which of the two methods to use.  I think you probably want to state - "Use Appendix E.2 for creation of new items, but use Appendix E.1 when parsing and checking an existing JSON serialization."

I believe this rather highlights a somewhat less clear part of the 04 draft: The distinction between "Hashable" and "Comparable" JSON.

Only "Comparable" JSON actually requires Appendix E.2 measures to function while "Hashable" JSON works with both.

These are effectively forming two levels of canonicalization:
- Hashable: JSON native level canonicalization only
- Comparable: Full canonicalization including custom data types

Anders

> 
> Jim
> 
> 
>> -----Original Message-----
>> From: Anders Rundgren <[e-mail]>
>> Sent: Friday, February 8, 2019 8:54 PM
>> To: Jim Schaad <[e-mail]>
>> Cc: [e-mail]; json-canon-
>> [e-mail]; Mark Nottingham <[e-mail]>; Samuel Erdtman
>> <[e-mail]>; Bret Jordan <[e-mail]>
>> Subject: JSON Subtype handling. Re: Review of draft-rundgren-json-
>> canonicalization-scheme-03
>>
>> On 2019-01-26 03:10, Jim Schaad wrote:
>>> Here are a set of comments on the current version of the draft.
>>
>> Thank you very much for these insightful comments!
>>
>> Since the comments were many I take the liberty splitting the responses in
>> separate threads.
>>
>> Handling of subtypes in JSON strings was clearly missing in the 03 draft.
>> The 04 draft adds the following to section 3.1 "Creation of Input Data":
>>
>>       Note: parsed String data MUST NOT be altered during subsequent
>>       serializations.  For more information see Appendix E
>>
>> The added Appendix E (https://tools.ietf.org/html/draft-rundgren-json-
>> canonicalization-scheme-04#appendix-E)
>> contains a pretty extensive description of how this must be dealt with.
>>
>> A side-effect of this is that the updated draft now makes a distinction
>> between "Hashable" JSON and "true" canonicalization ("Comparable" JSON).
>> The latter is a more demanding deployment option.
>>
>> Cheers,
>> Anders
>