[Cbor] Re: Consensus call on EDN literals single ABNF

[Pete Resnick <[e-mail]>]

Rohan asked me about this, so I thought I'd pipe up on the list. 
Overall, I don't see this as a "two-layer" or "two-pass" issue; the ABNF 
as described is simply not complete. To respond directly to Christian:

On 21 Jul 2024, at 10:46, Christian Amsüss wrote:

> There is a large risk of the special cases to diverge from the general
> cases. Whenever an extension point is specified (be it in ABNF or 
> CBOR)
> as
>
> item = key1 value1 /
>        key2 value2 /
>        extkey extvalue
>
> then implementors get no guarantee that the processing of the 
> originally
> defined keys and values is consistent with the processing of 
> extensions.

I agree that the above is not a great idea, but that wouldn't be the way 
I suggest to make the change. In the current top-level syntax in 4.1, 
you have:

````
    app-prefix      = lcalpha *lcalnum ; including h and b64
                    / ucalpha *ucalnum ; tagged variant, if defined
    app-string      = app-prefix sqstr
    sqstr           = "'" *single-quoted "'"
    bstr            = app-string / sqstr / embedded
                      ; app-string could be any type
````

With this syntax, bstr can never be parsed to anything that results in 
"app-string-h" or "app-string-dt", etc. That's (I believe) not what you 
want. Try this instead:

````
    app-string      = app-string-h
                    / app-string-b64
                    / app-string-dt
                    / app-string-ip
                    / app-string-gen

    app-string-gen  = app-prefix sqstr ; generic app-string
    app-prefix      = lcalpha *lcalnum ; including h and b64
                    / ucalpha *ucalnum ; tagged variant, if defined
    sqstr           = "'" *single-quoted "'"
    bstr            = app-string / sqstr / embedded
````

Then leave everything in 4.2 exactly as it is now. That will allow all 
of the current possibilities, and if you add a new app-string type in a 
later document, you can simply say:

````
    app-string      =/ app-string-new
````

Correct ABNF requires something like the above AFAICT.

pr
-- 
Pete Resnick https://www.episteme.net/
All connections to the world are tenuous at best