Sounds good. A user who submitted an issue to us has the same opinion,
so I'm happy to comply.
Perhaps adding a sentence to 3.7.3 providing some extra guidance in this
regard might be useful
thanks, Sergey
On 26/09/11 12:07, Marek Potociar wrote:
>
>
> On 09/25/2011 11:07 PM, Sergey Beryozkin wrote:
>> Hi
>>
>> I'm wondering, what should the specification say about cases where
>> HTTP ContentType or Accept contain malformed values.
>
> You mean malformed in the sense of not following the definition in HTTP 1.1 spec?
> http://www.w3.org/Protocols/rfc2616/rfc2616-sec3.html#sec3.7
>
>>
>> For example, should
>>
>> Accept: text
>>
>> result in 406 or 400 ?
>
> Depends on the side IMO.
>
> It would probably make sense to follow the "be conservative in what you produce" guideline and fail early with an
> exception on the client side.
>
> On the server side, all existing implementations should follow the spec and send back 415 (in case of
> malformed/unsupported Content-type) or 406 (in case of malformed/unsupported Accept) - as per section 3.7.3, stage 3.a.
>
>>
>> I'd like to avoid defaulting 'text' to text/* - that would probably be illegal anyway
>
> +1.
>
> Marek
>
>>
>> Cheers, Sergey
>>
>>
--
Sergey Beryozkin
http://sberyozkin.blogspot.com
Talend - http://www.talend.com