[Json] Re: [Errata Verified] RFC8259 (7383)

[Carsten Bormann <[e-mail]>]

Hi Austin,

> On Jun 8, 2026, at 06:11, Austin Wright <[e-mail]> wrote:
> 
> I believe this errata is unnecessary, 

I think you are trying to be too kind in your choice of words.
The errata report is not “unnecessary", it is “wrong" and needs to be “rejected".

The errata report says that "Hex values should be capitalized” and refers to https://www.rfc-editor.org/rfc/rfc5234#appendix-B.1 to support this statement.
I have no idea what “should” means here.
In any case, as you mention, “A” and similar constructs match in a case-insensitive way in RFC 5234, so the definition of HEXDIG in Appendix B.1 cannot be used to support “should be capitalized”, in whatever meaning this phrase is used.

In summary, this errata report needs to be REJECTED, preferably with an explanation that 

         HEXDIG = DIGIT / "A" / "B" / "C" / "D" / "E" / “F”
,
         HEXDIG = DIGIT / “a" / “b" / “c" / “d" / “e" / “f”
, and
         HEXDIG = DIGIT / "A" / "B" / "C" / "D" / "E" / “F"
                        / “a" / “b" / “c" / “d" / “e" / “f”

all three would mean EXACTLY the same thing in ABNF.

(The situation here is a bit confusing, as RFC 5234 uses ABNF to define ABNF, which is then used by RFC 8259.
The definition of HEXDIG uses the char-val construct, which is only implicitly referenced in Section 2.3 when it says "ABNF permits the specification of literal text strings directly,
enclosed in quotation marks.”
A HEXDIG defined this way is then used to define hex-val and thus num-val, which governs the definition of false in RFC 8259.)

The fact that ABNF matches char-val elements (“literal text strings”) in a case-insensitive way is a Frequently Encountered Gap In Understanding RFC 5234 and the way ABNF is defined.
(The fact that the case insensitivity of char-val is mentioned in a “NOTE” in Section 2.3 of RFC 5234, and without even mentioning the ABNF rule name, does not help.)

It would be nice if there was a common record of this FEGIU somewhere that could be referenced in responses to errata reports, but of course we can also repeat that information again and again with each rejected errata report.

Practically speaking, users of “literal text strings” (char-val) in ABNF that contain letters *always* need to include comments pointing out the FEGIU at each place where it might be incited.
this should be a check point when looking at new ABNF to be standardized, possibly by simply using RFC 7405 notation to make case insensitivity explicit, or also by adding a note such as the last bullet item in Section 2.3.1.1 in RFC 9535 [1].

Grüße, Carsten

[1]: https://www.rfc-editor.org/info/rfc9535/#section-2.3.1.1-5.2
(Yes, that was the last RFC I’ll ever contribute to that references RFC 5234 without using RFC 7405 as well.
Ashes on my head.)