Re: [Json] ABNF nits

[Stefan Drees <[e-mail]>]

On 07.06.13 16:10, Matt Miller (mamille2) wrote:
> Would one of you be so kind as to put all of this into a proposal we can
> all get our heads wrapped around?

I will try for an estimated consensual definition of all and this ;-)

Proposal:

Replace the closing ABNF rules inside section 2.4. Numbers

OLD:
"""
       number = [ minus ] int [ frac ] [ exp ]

       decimal-point = %x2E       ; .

       digit1-9 = %x31-39         ; 1-9

       e = %x65 / %x45            ; e E

       exp = e [ minus / plus ] 1*DIGIT

       frac = decimal-point 1*DIGIT

       int = zero / ( digit1-9 *DIGIT )

       minus = %x2D               ; -

       plus = %x2B                ; +

       zero = %x30                ; 0
"""

with

NEW:
"""
       number = [ minus ] int [ frac ] [ exp ]

       decimal-point = %x2E       ; .

       digit1-9 = %x31-39         ; 1-9

       e = %x65 / %x45            ; e E

       exp = e [ minus / plus ] 1*DIGIT

       frac = decimal-point 1*DIGIT

       int = zero / ( digit1-9 *DIGIT )

       DIGIT = %x30-39            ; 0-9
             ; DIGIT equivalent to DIGIT rule in [RFC5234]

       minus = %x2D               ; -

       plus = %x2B                ; +

       zero = %x30                ; 0

"""

and in section 2.5 Strings (as there is a 4HEXDIG, and no HEXDIG 
declared) also the ABNF rules section replace:

OLD:
"""
       string = quotation-mark *char quotation-mark

       char = unescaped /
           escape (
               %x22 /          ; "    quotation mark  U+0022
               %x5C /          ; \    reverse solidus U+005C
               %x2F /          ; /    solidus         U+002F
               %x62 /          ; b    backspace       U+0008
               %x66 /          ; f    form feed       U+000C
               %x6E /          ; n    line feed       U+000A
               %x72 /          ; r    carriage return U+000D
               %x74 /          ; t    tab             U+0009
               %x75 4HEXDIG )  ; uXXXX                U+XXXX

       escape = %x5C              ; \

       quotation-mark = %x22      ; "

       unescaped = %x20-21 / %x23-5B / %x5D-10FFFF

"""

with:
NEW:
"""
       string = quotation-mark *char quotation-mark

       char = unescaped /
           escape (
               %x22 /          ; "    quotation mark  U+0022
               %x5C /          ; \    reverse solidus U+005C
               %x2F /          ; /    solidus         U+002F
               %x62 /          ; b    backspace       U+0008
               %x66 /          ; f    form feed       U+000C
               %x6E /          ; n    line feed       U+000A
               %x72 /          ; r    carriage return U+000D
               %x74 /          ; t    tab             U+0009
               %x75 4HEXDIG )  ; uXXXX                U+XXXX

       escape = %x5C              ; \

       quotation-mark = %x22      ; "

       unescaped = %x20-21 / %x23-5B / %x5D-10FFFF

       HEXDIG = DIGIT / %x41-46 / %x61-66   ; 0-9, A-F, or a-f
              ; HEXDIG equivalent to HEXDIG rule in [RFC5234]

"""

Did it catch everything? Or do we want to explain, that 4HEXDIG means 
exactly 4 hex digits? I guess no, as the comment is depicting this nice 
and it is common ABNF practice to deal with multiplicity.

All the best and second to top posting as this is a summarizing job ;-)
Stefan

>
>
> kthxbye,
>
> - m&m
>
> Matt Miller < [e-mail] >
> Cisco Systems, Inc.
>
> On Jun 7, 2013, at 12:31 AM, Stefan Drees <[e-mail]> wrote:
>
>> On 2013-06-07 05:15, James H. Manger wrote:
>>>>>> Some things I found futzing with the ABNF:
>>>>>>
>>>>>> - normalize the indentation (makes it easier to pull out)
>>>>>> - Hex literals need to be uppercase.  The false and null rules are
>>>>>> wrong.
>>>>>
>>>>> Lowercase is allowed in hex literals.
>>>>> It links back to HEXDIG in RFC5234 (or RFC 4234).
>>>>> You have to remember the sneaky ABNF rule that quoted strings (eg "A")
>>>>> are case insensitive. "A" is equivalent to %x41 / %x61.
>>>>
>>>> Crap.  I didn't see that because it's in a "Note:" in section 2.3.
>>>
>>> I didn't call it "sneaky" for nothing ;) ...
>>
>> lucky JSON, it only needs short tokens (cowardly avoiding the word single character here ;-)
>>
>> If defining a grammar with real case-sensitive "word"-ish tokens in it, ABNF is IMO barely usable, unless you extend it with say single quotes as case-sensitve string delimiters. But then dealing with two kinds of strings also has it's catches =(
>>
>>
>>>> Regardless, we should change  both of those rules for consistency with
>>>> the rest of the doc.
>>>>
>>>>>> - DIGIT and HEXDIG are neither defined nor imported, as far as I can
>>>>>> tell.
>>>>>> I suggest:
>>>>>>
>>>>>> DIGIT = %x30-39                        ; 0-9
>>>>>> HEXDIG = %x30-39 / %x41-46 / %x61-66   ; 0-9, A-F, or a-f
>>>>
>>>> I still recommend being explicit about this, like the rule for e.
>>>
>>> I agree.
>>> Add DIGIT and HEXDIG to the doc so all the ABNF is present.
>>> Use your definition of HEXDIG above, not the RFC5234 one, to avoid the sneaky ABNF rule.
>>> Optionally add a note "; DIGIT and HEXDIG are equivalent to rules with the same names in [RFC5234]".
>>
>> +1 on your proposal James (including the ABNF comment).
>>
>> Stefan.
>>
>> _______________________________________________
>> json mailing list
>> [e-mail]
>> https://www.ietf.org/mailman/listinfo/json
>