Re: [EAI] Proposed resolution of List-* and downgrade- Issues

["Charles Lindsey" <[e-mail]>]

On Tue, 29 Jul 2008 13:59:36 +0100, Randall Gellens <[e-mail]>  
wrote:

> At 10:41 AM +0100 7/29/08, Charles Lindsey wrote:
>
>>  On Mon, 28 Jul 2008 18:34:57 +0100, Randall Gellens  
>> <[e-mail]> wrote:

>>>  EAI to define new versions of List-* headers which can have unencoded  
>>> UTF8 values....
>>
>>  But is that not equivalent to just saying that IRIs are to be  
>> permitted (within the UTF8SMTP universe) in places where URIs are  
>> currently allowed?
>>
>>  And if it is not equivalent, then why cannot it be made so?
>
> I'm no expert in IRIs, but my understanding is that an IRI can be  
> represented either in UTF8 or in percent-encoded form.  I specifically  
> said "unencoded UTF8 values" because I think this is the key point.

I think we need Martin to come in here and explain it all to us.
>
>>
>>>  A preference is for List-ID to be considered special and remain ASCII  
>>> (same as Message-ID).
>>
>>  Agreed.
>
> Yes, but, there is the sticky point in the next paragraph....

But the my paragraph was not addressed to List-ID, but to those List-*  
headers that currently use URIs.
>
>>
>>>  ... There is an open question on internationalized domain names:  
>>> since these are used in Sieve scripts, users need to be able to type  
>>> them in.

>
> The point about internationalized domain names and Sieve scripts is  
> specifically about List-ID. Since this is the unique identifier for a  
> list and hence used in Sieve scripts, it should be permitted to be  
> something meaningful that users can type in.  So, the example might be
>
> 	List-ID: list.mañana.com

Well either we allow UTF-8 in there, or we don't. If we do, then we have  
to address the consequences.


>>>  List-* headers which contain native UTF8 should be copied into
>>>  Downgrade-List-* and the UTF8 IRI should be dropped (alternate:
>>>  encoded per IRI).
>>
>>  No! Don't drop it. Downgrade it to a URI, and let mailto-bis sort out  
>> the result.
>
> mailto is just one possible scheme that can be used in a List-* header.   
> I'm OK with encoding it, if the encoded result is always as valid and  
> usable as the unencoded form.

If it's a valid IRI (of whatever scheme), then there is already an agreed  
mechanism to turn it into a valid URI. Whether it is "valid" delepnds on  
the scheme. The Mailto case is off interest because it is likely to be  
modified to handle EAI addresses, and we are in a position to influence  
the nature of that modification. Again, we need Martin to come in here.

>>  Having invented alt-addresses for use elsewhere within EAI, let us  
>> stick with them here. The last thing we want is two mechanisms for  
>> achieving the same result.
>
> I don't follow.  What are the two versions?  The point is that an  
> EAI-aware list can be created that operates behind a UTF8 email address  
> which has no ASCII version.  That's perfectly OK, if people want to do  
> this.  Presumably, the list is only for people who have EAI-aware  
> clients.

There may be mailing lists which are only intended for EAI-aware clients.

There may be mailing lists where EAI messages are accepted, but are  
downgraded before forwarding.

There may be mailing lists where EAI messages are accepted and forwarded,  
but will get downgraded when sent to non-EAI subscribers.

There may be mailing lists which are all-ASCII, but which accept  
subscriptions from EAI addresses.

All of these cases need to work in a sensible manner.

-- 
Charles H. Lindsey ---------At Home, doing my own thing------------------------
Tel: [phone]                       
   Web: http://www.cs.man.ac.uk/~chl
Email: [e-mail]      Snail: 5 Clerewood Ave, CHEADLE, SK8 3JU, U.K.
PGP: 2C15F1A9      Fingerprint: 73 6D C2 51 93 A0 01 E7 65 E8 64 7E 14 A4 AB A5
_______________________________________________
IMA mailing list
[e-mail]
https://www.ietf.org/mailman/listinfo/ima