Re: p1: handling obs-fold

[Amos Jeffries <[e-mail]>]

On 20/04/2013 7:43 p.m., Mark Nottingham wrote:
> On 20/04/2013, at 5:00 PM, Willy Tarreau <[e-mail]> wrote:
>
>> On Sat, Apr 20, 2013 at 02:07:39PM +1000, Mark Nottingham wrote:
>>> p1 3.2.4 defines requirements for handling obs-fold:
>>>
>>>> When an obs-fold is received in a message, recipients MUST do one of:
>>>>
>>>> 	? accept the message and replace any embedded obs-fold whitespace with either a single SP or a matching number of SP octets (to avoid buffer copying) prior to interpreting the field value or forwarding the message downstream;
>>>> 	? if it is a request, reject the message by sending a 400 (Bad Request) response with a representation explaining that obsolete line folding is unacceptable; or,
>>>> 	? if it is a response, discard the message and generate a 502 (Bad Gateway) response with a representation explaining that unacceptable line folding was received.
>>>>
>>>> Recipients that choose not to implement obs-fold processing (as described above) MUST NOT accept messages containing header fields with leading whitespace, as this can expose them to attacks that exploit this difference in processing.
>>> This seems to repeat itself; what is the difference between choosing to reject the request in the manner described in the last two bullet points, and not accepting the message?
>>>
>>> I think that the last sentence can be removed.
>> I think it was here before the addition above. In fact it targets a different
>> audience which is not aware of OBS at all. The simple fact that we talk about
>> prepending spaces before a header field means that the reader doesn't
>> understand that this field is not one but the continuation of previous one.
>>
>> Maybe this confusing sentence should be removed and replaced with something
>> like this before the block you quoted :
>>
>>   Presence of a space or tab character at the beginning of a line must not
>>   be taken as a new header field but as the continuation of previous header
>>   field (obs-fold). As such it cannot happen on the first header field.
>>
>> That way readers looking for what to do with these spaces will find their
>> response here and will be able to decide what to do with the options that
>> are offered to them.
>
> Seems reasonable; I think this one is largely editorial; the only way I'd be really concerned would be if nothing changed.
>
> Recorded as:
>    http://trac.tools.ietf.org/wg/httpbis/trac/ticket/444
> with suggestions.

I would just go with a MUST NOT on whitespace before or after the 
field-name. Not just between the field-name and its colon.

Amos