On 04.10.2017 09:01, Georg Lukas wrote:
> * Florian Schmaus <[email protected]> [2017-09-23 11:22]:
>> I think I don't understand the benefits from Georg's suggestion, so
>> currently one reason against is that it adds more complexity without any
>> gain from my PoV.
> 
> I would argue that it reduces complexity by decreasing the number of
> "unique" message IDs attached to a given message from three to two.
> 
> As a client developer, I want to have as few places as possible to look
> for a message ID, so stanza-id >> origin-id >> stanza id (*).
> 
> (*) stanza-id = XEP-0359 <stanza-id/> element
>     stanza id = 'id' attribute of the message stanza

I don't see an alternative "search for <origin-id/> first and fallback
to the 'id' attribute of the stanza" to get an ID for de-duplication.

>> Plus what you said: The scheme will break as soon as the MUC service, or
>> any other service doing some sort of reflection, rewrites the 'id'
>> attribute of the stanza (which MUC services are allowed to, and are the
>> reason for xep359 existence in the first place).
> 
> There are also MUC implementations (read: transports) that strip away
> non-body elements from reflected messages.

Of course there are, but with the current wording you have at least no
<origin-id/> (or <stanza-id/>) at all. Also it is discoverable if a MUC
service will not strip the xep359 extension elements by checking for the
xep359 disco#info feature.

But if we go with what you suggest, we could end up with a <origin-id>
with a different stanza-id than the 'id' attribute of the message
stanza, which would violate your rule. I think that causes more
confusion and complexity, not less.

> Other XEPs, like LMC and References, use the stanza id as a reference. I
> assume that MUCs will not rewrite the stanza id in the payload, thus
> breaking those XEPs. By enforcing origin-id = stanza id we can at least
> fix this particular problem on clients that support both LMC/Reactions
> and 0359.

I don't follow. How do which XEPs break? Which problem are you talking
about?

- Florian

Attachment: signature.asc
Description: OpenPGP digital signature

_______________________________________________
Standards mailing list
Info: https://mail.jabber.org/mailman/listinfo/standards
Unsubscribe: [email protected]
_______________________________________________

Reply via email to