Re: [Sip] MIME: Nested bodies with option-tag?

[Paul Kyzivat <[e-mail]>]

Christer Holmberg (JO/LMF) wrote:
> Hi,
> 
> My main concern is not what wording (SHALL, MUST, etc) we use.
> 
> The issue is that I think we need text saying what implementations that
> DO NOT support alternative and/or nested shall do. If we say that they
> shall reject the message I believe that many of the existing
> implementations will be compliant with the spec.
> 
> And, if we give the possibility to simply reject the message, I believe
> it will be easier to get people to change their implementations, if they
> currently simply discard alternative and/or nested today.

I agree that simply ignoring a body can cause huge problems.

I believe we have addressed that by clarifying the use of 
Content-Disposition and its handling parameter. In the absence of 
handling=optional, the request must be rejected. This part must be of 
MUST strength. And the default of handling=required in the absence of a 
C-D header must also be of MUST strength.

But there is little we can do about existing implementations that don't 
do that right. I guess I can see writing up somewhere that even if you 
aren't going to make your implementation compliant to this full draft 
you should at least do make sure you support Content-Disposition. Would 
it be helpful in that regard to make that part a separate draft???

	Paul

> Regards,
> 
> Christer
>  
> 
>> -----Original Message-----
>> From: Gonzalo Camarillo [mailto:[e-mail]] 
>> Sent: 29. toukokuuta 2007 16:06
>> To: Paul Kyzivat
>> Cc: Christer Holmberg (JO/LMF); [e-mail]
>> Subject: Re: [Sip] MIME: Nested bodies with option-tag?
>>
>> Hi,
>>
>> what I am hearing is that there are implementations out there 
>> that support multipart but not nested. Therefore, we need to 
>> decide two things:
>>
>> 1) do we want to have a MUST for multipart and a SHOULD for 
>> nested? I would say that we should have the same level (e.g., 
>> MUST, if we decided that MUST is appropriate) for both.
>>
>> 2) do we need a way for implementations that support 
>> multipart but not nested to be quickly updated to, at least, 
>> report that fact with an error response? This may make sense.
>>
>> Cheers,
>>
>> Gonzalo
>>
>> Paul Kyzivat wrote:
>>> More or less repeating what I said before:
>>>
>>> I expect we do have to account in some way for implementations that 
>>> have already been deployed, in absence of a clarifying document. 
>>> Exactly how we deal with that is still TBD.
>>>
>>> But as we define what is required to support this in the future, I 
>>> think there is *no* benefit to defining two levels of 
>> support - full 
>>> and partial. Anybody that sets out to provide support for this 
>>> document should be expected to do it all - its not that much harder.
>>>
>>>     Paul
>>>
>>> Christer Holmberg (JO/LMF) wrote:
>>>> Hi,
>>>>>   I wonder whether we should define an option-tag for the 
>> support of
>>>>>   nested bodies.
>>>>>
>>>>> I don't think there's a lot to be gained from defining such an 
>>>>> option-tag.  The sender should already be aware that 
>> there is a risk 
>>>>> the recipient can't understand nested bodies, and have 
>> arranged for 
>>>>> suitable fallbacks.  Conversely, the recipient should (at 
>> least) be 
>>>>> able to skip the nested multipart body part in the proper 
>> "I don't 
>>>>> understand this body part" way.  All an option-tag would 
>> do is allow 
>>>>> the sender to not add a fallback.
>>>> In that case we need some specific text saying that if the 
>> receiver 
>>>> does not support nested multiparts it MUST do-this-and-do-that.
>>>>
>>>> Because, as I said earlier, I don't think we will achieve what we 
>>>> want by saying that one MUST be able to parse nested 
>> multiparts. It 
>>>> can be rather tricky to implement (depending on how the parser is 
>>>> implemented, though), and since there aren't really any 
>> use-cases out 
>>>> there yet I am pretty sure some people will choose not to 
>> implement 
>>>> it (and saying that people are not compliant in that case will not 
>>>> really help from an interop perspective). So, because of 
>> that I think 
>>>> it would be good not to mandate the support of nested 
>> multiparts, but 
>>>> to mandate appropriate behavior if not supported - just 
>> like in any 
>>>> other case when a MIME body contains an unsupported but 
>> required content type.
>>>> Regards,
>>>>
>>>> Christer
>>>>
>>>>
>>>>
>>>>> Sip mailing list  https://www1.ietf.org/mailman/listinfo/sip
>>>>> This list is for NEW development of the core SIP Protocol Use 
>>>>> [e-mail] for questions on current sip Use 
>>>>> [e-mail] for new developments on the application of sip
>>>>>
>>>>
>>>> _______________________________________________
>>>> Sip mailing list  https://www1.ietf.org/mailman/listinfo/sip
>>>> This list is for NEW development of the core SIP Protocol Use 
>>>> [e-mail] for questions on current sip Use 
>>>> [e-mail] for new developments on the application of sip
>>>>
>>>
>>> _______________________________________________
>>> Sip mailing list  https://www1.ietf.org/mailman/listinfo/sip
>>> This list is for NEW development of the core SIP Protocol Use 
>>> [e-mail] for questions on current sip Use 
>>> [e-mail] for new developments on the application of sip
>>>
>>
> 


_______________________________________________
Sip mailing list  https://www1.ietf.org/mailman/listinfo/sip
This list is for NEW development of the core SIP Protocol
Use [e-mail] for questions on current sip
Use [e-mail] for new developments on the application of sip