jdev - 2025-03-16

https://xmpp.org/extensions/xep-0452.html

did someone implement this already? Was there some discussion around it, are there concerns?

i mean rather https://xmpp.org/extensions/xep-0372.html the mentioning part of it

this seems to be again forever in experimental

Renga and converse.js implement 0372 mentions

There's a hroblem with the spec that says you should mention a user bare jid, in a muc that doesn't make sense

(At least in a muc where users are not allowed to see each other' jid)

In GC3, wouldn't mentions best be by occupant-id? (I haven't followed GC3 discussions closely enough to remember if that's covered (yet))

thanks great, so basically this does not support yet occupant-id

which is kind of a deal breaker for me

but since everything is somekind of uri

lets just agree on some sensible uri for this or?

Yes, there was some discussion about this a few years ago in one of the sprints but I think nothing finalized came out of it

btw, whats GC3?

group chat 3?

Yes

There's an in-progress doc somewhere

i propose xmpp:groupchat@gc.server.com?;occupant-id=xmca8asd9ads

seems redundant to mention the groupchat again, but nothing else comes to mind

ahh we could simply use full jid for good measure

xmpp:groupchat@gc.server.com/nickname?;occupant-id=xmca8asd9ads

0452 is for mentions in MUCs where you're not currently present, so including the MUC name is relevant

yeah, there is just some security considerations

actually its bit tricky. I well can mention someone who is currently not joined

so the nick is unknown

If the nick is unknown, how are you (as a user) mentioning them?

although i dont need to add it, as the user will know his prefered nick anyway himself, when he gets the message forwarded

theTedd, i know only the last nick

i mean i can send any nick i want, no client should trust it

so why send it at all?

> lets just agree on some sensible uri for this or? Under GC3 the full jid will contain the occupant id ↺

"You were mentioned in "jdev" as 'lovetox'."

you mean for the notification?

but this is irelevant info, if the MUC informs me when im not joined, its enough it tells me im mentioned, and by who

not what the user thought my nick would be

It could be, but most times probably not

i mean i get the whole message forwarded anyway

so in the text my name will be in it

and reference tells me where in the message

Ok

singpolyma, not sure how that helps me implement mentions

Well it means the uri is just xmpp:fulljid I guess

ok but there is no gc3 implemented, and im not keen on delaying that feature

so im back at xmpp:groupchat@gc.server.com?;occupant-id=xmca8asd9ads because i think nickname should be irrelevant

Isn't there also a full rewrite of the references xep underway? That's why I aborted implementing this last time

its in the message anyway

would it do something conceptually different?

i guess we still have the nickname in the message, and reference it with start/end

not sure how this could work any different

Sure, at a fundamental level if it's not inline I agree it will be some sort of count like that

yeah then this would not really stop me, because adding a bit of parsing different xml is trivial at a later point

So I guess it's just to register a Uri query key for occupant id. Or agree to use gc3 style URIs now independent of any server implementation

what does a gc3 look like? just occupant id as resource?

hmm yeah why not

it would be a bit weird if GC3 never happens though :D

> what does a gc3 look like? just occupant id as resource? Yes ↺

i tend to lean to the query key for now, supporting other jids would be trivial at a later point

I have a slight preference the other way, but if you're going to do this probably I'll do whatever you do 🙂

You'll need both until GC3 is the default anyway

Not necessarily 🙂

Question from another room for Tigase devs: which of Stork, Beagle, and Siskin are still in active development?

lovetox, are you aware of https://github.com/xsf/xeps/pull/1281 and https://bouah.net/specs/mentions.html on which we had worked on in the previous Lyon sprint?