Re: [Json] [Technical Errata Reported] RFC8259 (7650)

[Lucas TESSON <[e-mail]>]

Hi,

thanks for your response.
Indeed it is not clear whether the "core rules" are always present, but I think that is the idea behind the name "core".

For instance, multiple RFCs makes use of those core rules without ever redefining them or clearly define that they "import" it from RFC 5234:
- RFC 5147: DIGIT, HEXDIG
- RFC 5321: CRLF, ALPHA, DIGIT, SP, DQUOTE, HEXDIG
- RFC 5954: DIGIT, HEXDIG
- RFC 6338: ALPHA, DIGIT, HEXDIG
- RFC 5404: HEXDIG, DIGIT
- ... and so on, even in NIST-IR and other specifications
It has been widely accepted that ABNF Core rules always lie in the grammar, leading to this errata.

Note that the BAP tool used by the IETF to validate ABNF grammar don't raise an error on this, which may be an actual issue in its implementation.

I still think the "char" rule name should be updated to avoid interpretations or forcing behavior.

Best regards,
   Lucas.

------- Original Message -------
Le mercredi 20 septembre 2023 à 5:10 PM, Carsten Bormann <[e-mail]> a écrit :


> Interesting.
> 

> I don’t read RFC 5234 to imply that Core Rules (B.1) are always present; these are just useful rules that can be imported when needed. (This is supported by B saying "Note that these rules are only valid for ABNF encoded in 7-bit ASCII or in characters sets that are a superset of 7-bit ASCII.”, so these rules appear to have been designed with limited applicability in mind.)
> 

> RFC 8259 does not define DIGIT or HEXDIG; the assumption here seems to be that those come from the Core Rules. This does not appear to be said in the document, which maybe is the real errata behind this report.
> 

> Grüße, Carsten