Re: CACHING-6 failure message: what is an "invalid" value?

Thanks Jo,

That's a feature then, cool ;)
The warning is much more useful when alerting about unusual values, in 
any case.

Francois, who should stop reading RFCs.


Jo Rabin wrote:
> I can tell you what was in my mind when I wrote that far-from-precise 
> phrase :-)
> 
> If you send Cache-Control: croque-monsieur that should be construed as 
> being invalid, though it is syntactically correct. The warn serves to 
> remind content authors that toasted-open-face-ham-and-cheese-sandwich is 
> not widely understood as a cache control directive and is likely to have 
> undesirable interoperability consequences. It is only a warn because as 
> you point out, Francois, RFC2616 makes it clear that any value is in 
> principle acceptable, providing it conforms to the syntax for a token.
> 
> Jo
> 
> On 03/09/2009 16:59, Francois Daoust wrote:
>> Hi,
>>
>> The CACHING test defines CACHING-6 as:
>>  [[ If any cache related header contains an invalid value, warn ]]
>>  http://www.w3.org/TR/mobileOK-basic10-tests/#CACHING
>>
>> What exactly constitutes an "invalid" value? For instance, looking at 
>> the definition of the Cache-Control header field in the HTTP RFC, I 
>> see that cache-response-directive may be a cache-extension:
>>  http://www.w3.org/Protocols/rfc2616/rfc2616-sec14.html#sec14.9
>>
>> Are such cache-extension directives considered to be invalid values? 
>> I.e. are we trying to alert authors on the fact that unusual 
>> cache-control directives are unlikely to be understood, or is this 
>> warning only motivated by real invalid values such as: "Cache-Control: 
>> =" ?
>>
>> I'd go for the latter proposal, but the mobileOK Checker library 
>> currently does the former.
>> Minor bug or feature?
>>
>> Francois.
>>
>>
>>
>

Received on Friday, 4 September 2009 08:08:41 UTC