[Standards] References and XHTML-IM interoperability

Kevin Smith kevin.smith at isode.com
Fri Dec 2 20:51:33 UTC 2016

> On 2 Dec 2016, at 18:38, Peter Saint-Andre <stpeter at stpeter.im> wrote:
>> On 12/2/16 11:32 AM, Kevin Smith wrote:
>>>> On 2 Dec 2016, at 18:22, Peter Saint-Andre <stpeter at stpeter.im> wrote:
>>>> On 12/2/16 8:48 AM, Kevin Smith wrote:
>>>> On 2 Dec 2016, at 15:44, Tobias M <tmarkmann at googlemail.com
>>>> <mailto:tmarkmann at googlemail.com>> wrote:
>>>>>> On 2 Dec 2016, at 16:39, Kevin Smith <kevin.smith at isode.com
>>>>>> <mailto:kevin.smith at isode.com>> wrote:
>>>>>> They’re long and (as far as I can see at the moment) unnecessary when
>>>>>> a simple per-stanza counter would suffice - only uniqueness within a
>>>>>> stanza is needed, not globally, isn’t it?
>>>>> As far as I know, technically an XML ID value needs to be unique in
>>>>> the whole document, and for the XMPP case the document is the XMPP
>>>>> stream/session. Using UUIDs would avoid having to go over and reassign
>>>>> IDs when attaching one message to another, etc.
>>>> Oh, you meant a literal ‘id’. That’s easily avoidable just by using a
>>>> different attribute :)
>>> IMHO this is exactly why 'id' was invented. :-)
>> Indeed. I'm slightly concerned with stream uniqueness guarantees that that implies, though. What if you forward a message to someone else, or happen to request a stanza from your archive twice or whatever?
>> I think we're probably already breaking that with various other cases and stanza ids though, so if we're happy doing it there do we need to worry about it here? (And if no, does mandating uuid make sense?)
> Oh, I thought we were talking about identifying fragments within XHTML-IM formatting, which is why I thought 'id' was right. Perhaps I am missing context.

We are, but that shouldn't stop me confusing things!

Tobi mentioned that we should use UUIDs, I thought it wasn't necessary, Tobi pointed out that the ID needs to be unique within an XML document (the XMPP stream), and UUIDs would achieve that for forwards of multiple stanzas and things, I mentioned that we couldn't guarantee that even with UUIDs and that we're already breaking that rule so it probably doesn't matter. 

And I very much want it blue ;)


More information about the Standards mailing list