[Standards] LAST CALL: XEP-0359 (Unique and Stable Stanza IDs)

Daniel Gultsch daniel at gultsch.de
Sun Nov 25 23:01:03 UTC 2018


Am So., 25. Nov. 2018 um 23:47 Uhr schrieb Florian Schmaus <flo at geekplace.eu>:
>
> On 22.11.18 18:07, Daniel Gultsch wrote:
> > Am Mi., 14. Nov. 2018 um 19:52 Uhr schrieb Georg Lukas <georg at op-co.de>:
> >> * Jonas Schäfer <jonas at wielicki.name> [2018-10-20 13:55]:
> >>
> >> A point that I brought up back then, and that I think needs to be added
> >> in §2.2 is this:
> >>
> >> | The message sender MUST set the stanza's @id to the same value as the
> >> | origin-id.
> >>
> >> The example should be changed accordingly.
> >>
> >> There is really no drawback in specifying that (as a MUST, or at least a
> >> SHOULD), and there is a huge amount of pain and madness later down the
> >> road if we don't mandate it. In my eyes, this is the only reasonable way
> >> to move forward.
> >>
> >> Therefore I change my vote to -1 unless the above statement is added
> >> with either a SHOULD or (preferrably) a MUST.
> >
> > I raised the some concerns multiple times on I’m also -1 on that
> > before this addressed.
> > Especially since I don’t see a reason for *not* doing this even if
> > some people thing it is not needed.
>
>
> Thanks Daniel and Georg for you feedback.
>
> The requests to require both values to be equal have always been very
> vague: No actual arguments were given why that would be beneficial.
> Maybe I missed them, and I'm sorry if that is the case. I looked into my
> notes, which I keep for every XEP I care about or I am personally
> involved in, but could not find any records regarding the potential
> upsides of doing so. Also, a quick search for the previous discussion of
> this topic yielded no results (*Summoning Zash* because I could bet
> there was such a discussion).
>
> On the other hand, there are reasons against:
>   - origin-id is entirely unrelated to the stanza id attribute
>   - it adds another rule to the XEP, hence increasing its complexity
>   - even if we would mandate it, you are not guaranteed that you will
> receive stanzas where origin-id is the same value as the stanza id
> attribute because - of MUC id rewrite (yes I know of the latest changes
> to XEP-0045)
>     - because there may be XEP-0359 implementations which do not do it
> (unless you want a namespace bump)
>
> I am happy to be convinced that your suggestion improves the XEP and the
> XMPP ecosystem as a whole. But I also hope that it is understandable
> that it is hard for me to become convinced of a change without providing
> any arguments in favour of the particular change. Arguments that could
> put weight in against the counterarguments.

If origin id and message id are the same and I react to it (message
correction, read marker, delivery receipt) then I can use the origin
id in my reaction and don’t care that a muc serice or who ever has
rewritten the message id and you’ll still be able to know what message
I’m refering to.
By setting origin-id==message-id you do not introduce ambiguity into
which of those ids should be used for #whatever.
Also if I’m storing the origin id/message id for de duplication
purposes I don’t have to store two ids

In Conversations code I only track on ID for reactions, dedubbing
purposes. (origin-id trumps message-id)

I believe that what ever purpose you thought out for the origin-id is
better served if message-id==origin-id unless it doesn’t serve a
purpose at all - Or I’m not understanding the purpose of the
origin-id.

That might be the crux here by the way. What is the purpose of the origin-id?

cheers
Daniel


More information about the Standards mailing list