Re: [Ietf-message-headers] Permanent message header field registration for traceparent and tracestate

[Julian Reschke <[e-mail]>]

On 24.11.2019 21:46, Philippe Le Hégaret wrote:
>
>
> On 11/24/2019 3:39 PM, Julian Reschke wrote:
>> The spec link's anchor doesn't work for me.
>> <https://www.w3.org/TR/trace-context/#traceparent-header> does, though.
>
> Didn't check and realize the links had changed. Here is an updated
> template:
>
> [[
> Header field name: traceparent
> Applicable protocol: http
> Status: permanent
> Change Controller:
>    Distributed Tracing Working Group
>    https://www.w3.org/2018/distributed-tracing/
>    [e-mail]
> Specification:
>    https://www.w3.org/TR/trace-context/#traceparent-header
>
> Header field name: tracestate
> Applicable protocol: http
> Status: permanent
> Change Controller:
>    Distributed Tracing Working Group
>    https://www.w3.org/2018/distributed-tracing/
>    [e-mail]
> Specification:
>    https://www.w3.org/TR/trace-context/#tracestate-header
> ]]
>
>> Quoting:
>>
>>> 3.2 Traceparent Header
>>>
>>> The traceparent HTTP header field identifies the incoming request in
>>> a tracing system. It has four fields:
>>>
>>>     version
>>>     trace-id
>>>     parent-id
>>>     trace-flags
>>
>> So what is the actual syntax of the header field?
>
> It's mentioned in the section
> [[
> [...]
>    The tracestate field value is a list of [list-members] separated by
> [commas].
>
> [...]
>
> list  = list-member 0*31( OWS "," OWS list-member )
> list-member = key "=" value
> list-member = OWS
>
> [...]
>
> key = lcalpha 0*255( lcalpha / DIGIT / "_" / "-"/ "*" / "/" )
> key = ( lcalpha / DIGIT ) 0*240( lcalpha / DIGIT / "_" / "-"/ "*" / "/"
> ) "@" lcalpha 0*13( lcalpha / DIGIT / "_" / "-"/ "*" / "/" )
> lcalpha    = %x61-7A ; a-z
>
> [...]
>
> value    = 0*255(chr) nblk-chr
> nblk-chr = %x21-2B / %x2D-3C / %x3E-7E
> chr      = %x20 / nblk-chr
> ]]
> https://www.w3.org/TR/trace-context/#tracestate-header-field-values

I see. I'd propose to improve the paragraph ordering :-).

The ABNF currently says:

> list  = list-member 0*31( OWS "," OWS list-member )
> list-member = key "=" value
> list-member = OWS

That's invalid, you can't define "list-member" twice; make it:

list  = list-member 0*31( OWS "," OWS list-member )
list-member = (key "=" value) / OWS

or simpler:

list  = #31 list-member
list-member = key "=" value

There's also a problem that the ABNF does not allow whitespace around
"=", but the prose does.

Best regards, Julian