Re: Multiple header fields with the same field name - unwritten assumption about quoted commas in values?

On 2013-01-09 17:29, Zhong Yu wrote:
> On Wed, Jan 9, 2013 at 5:00 AM, Julian Reschke <julian.reschke@gmx.de> wrote:
>> On 2013-01-09 03:22, Zhong Yu wrote:
>>>
>>> On Tue, Jan 8, 2013 at 6:28 PM, Piotr Dobrogost <p@ietf.dobrogost.net>
>>> wrote:
>>>>
>>>> Hi!
>>>>
>>>> In
>>>> http://trac.tools.ietf.org/html/draft-ietf-httpbis-p1-messaging-21#section-3.2
>>>> one reads
>>>>
>>>> "Multiple header fields with the same field name MUST NOT be sent in a
>>>>    message unless the entire field value for that header field is
>>>>    defined as a comma-separated list [i.e., #(values)]."
>>>>
>>>> Is the unwritten assumption that each value out of "values" has no
>>>> unquoted comma?
>>>
>>>
>>> Good question. I think it is true for all headers defined in the spec
>>> that use the #rule; I guess it is intended to be true in general.
>>
>>
>> It's a good idea to do, but not always the case. Example: "If-Match" and
>> "If-None-Match". So to separate values in a list production, code needs to
>> understand the syntax of the individual values.
>
> The commas in entity tags are "quoted", so it's not a problem for a
> generic parser which pairs quotes.
>
>      If-Match: "12,34", W/"56,78"

But they do not use the quoted-string ABNF production (anymore). Note 
that that would imply handling of "\" which we found almost nobody does. 
This also means that just "pairing" double quotes is not sufficient.

> The Link header can contain unquoted commas, since comma is legal in URIs
>
>      Link: </page,6>; rel="Previous",  </page,8>; rel="Next"
>
> fortunately these commas will be surround by <>, no ambiguity there.

Yes, but the parser needs to understand the ABNF of the field value in 
order to decide which is which.

> The same bis section also specifically mentions the exception to
> `Set-Cookie` - each cookie must be declared in its own line, because
> unquoted comma can appear in the line
>
>      Set-Cookie: name=value; path=/ab,1; expires=Wed, 09 Jun 2021 10:18:14 GMT
>
> If we are to put multiple cookies on one line, the comma in the expiry
> date does not cause ambiguity, but the comma in the path value could
>
>      Set-Cookie: n1=v1; path=/ab,1 , n2=v2

Yes, Set-Cookie is a mess. That's why we needed that to call out 
specifically.

Best regards, Julian

Received on Wednesday, 9 January 2013 16:38:41 UTC