jdev - 2026-04-15

> The vast majority of stanzas are `<presence/>` so the average will tend towards the average size of those (less than 1kb) /me starts sending 128kb presences to every muc ↺

> do you want to do TLS on it too or are you ok offloading that? TLS termination is offloaded to the ESP32 in the cartridge (same one that handles internet) ↺

taba, XMPP runs fine over military radio links as-is. Only changes needed are that you want to be doing S2S, and you want a 0-RTT connection setup. We basically have that in the mainline with SASL2, now, so possibly even C2S would work reasonably well.

> I wrote https://github.com/moparisthebest/kiss-xmpp to use from my 1983 atari ... one day I'll play with writing a native one :) So this is just a telnet terminal emulator? ↺

What is the standard procedure for proposing a new XEP?

Goot the ticklegoblin!, https://xmpp.org/extensions/xep-0001.html#submission

t-thanks

https://xmpp.org/about/standards-process/

dwd's link looks nicer

Which will then guide you to https://xmpp.org/extensions/xep-0143.html#submit for more detail, but XEP-0001 has the formalities.

jonas', It does look pretty (and is more user-friendly, I think) but it's not very obvious to find.

indeed

Goot, What are you thinking of specifying?

Is there presently a standard way to convey information about federation restrictions on accounts?

Goot the ticklegoblin!, it sounds like something MattJ may know something about and/or have some interest in

For e.g. SASL ANONYMOUS accounts

Goot the ticklegoblin!, to tell the client that they aren't permitted to federate, you mean?

Yes, or give a list of domains to which they are permitted to contact

If they are so restricted

Thinking a stream feature sent alongside/after binding (not familiar with how bind 2.0 works)

Preferably before they try and send a policy-violating stanza ✏

I'm not aware of anything like that, no

Some policies may be hard to encode in such a protocol, but for simple "allowed"/"disallowed" lists it might be useful

Working on a video game system and supporting ANONYMOUS login; while it may be a safe assumption that you'd only be able to contact other servers which are under the same authoritative control as your server, there is no way to guarantee that for e.g. anon.some-domain.example and groupchat.some-domain.example will play nicely together (and the fewer assumptions that have to be made, the better)

Being able to catch that a challenge to someperson@someremotedomain.example/someresource would fail before we send it would be helpful

> Being able to catch that a challenge to someperson@someremotedomain.example/someresource would fail before we send it would be helpful you can already do that though right? just send something ↺

send an IQ ping, you'll get an error if you can't send there?

moparisthebest, Sure, though that requires a probe prior to a send. If nothing else, it'd be useful to document how to do this. I'll bet it's not entirely consistent.

right, regardless you have to send something to know if you can send, why create new thing when existing works

why make wood wheel when stone work good

moparisthebest, Because you can make a better user experience if you know a priori whether you can communicate?

I agree, I'm saying there's an already working way to do that now

> why make wood wheel when stone work good well making something better would be a reason, but haven't seen a proposal ↺

This conversation did start by someone asking how to propose a new XEP...

and then asked if there was an existing way to do it...

Option 1: here is a list of domains you have permission to contact - have fun. Option 2: we're not telling you whether you have permission, so you're going to have to probe each individual domain beforehand, and if that's 20 different domains, err.... well, have fun!

a domain whitelist is something else entirely sure

How is that different from what was asked about?

> Being able to catch that a challenge to someperson@someremotedomain.example/someresource would fail before we send it would be helpful this is what I responded to ↺

> Yes, or give a list of domains to which they are permitted to contact This is a further clarification ↺

the one I responded to came later, it was the clarification

But still, how is "am I allowed to contact these domains?" different from a whitelist?

it isn't, but I think this might be a case of https://en.wikipedia.org/wiki/XY_problem

If we know in advance, we can tell the player something to the effect of "note: if your opponent's JID is not in one of these domains, you won't be able to play with them" ✏

Or "if you try to connect to a server whose JID isn't from one of these domains," or "if you try to host a server, only players from these domains can join", etc

> If we know in advance, we can tell the player something to the effect of "note: if your opponent's JID is not in one of these domains, you won't be able to contact them" and these users know what a JID and domain part is and further know their contact's JIDs ? ↺

Some gamers are literate

but why make anyone check what a machine can do much easier ? Why not have them paste JID and now show the message?

Great idea! And there could be a whitelist that does the checking to avoid repeated requests and roundtrips!

> If we know in advance, we can tell the player something to the effect of "note: if your opponent's JID is not in one of these domains, you won't be able to play with them" This is your game, right? ↺

Why not just make a nonstandard extension

well they can but they were seeing if something for this already exists, because why use nonstandard protocol when standard protocol already exists lol

Riot Games does the same thing, their game (LoL) is full of nonstandard extensions ✏

> Why not just make a nonstandard extension This has broader potential to be useful than video games ↺

just realize your server allowing you to send to bob.com is not the same as being able to send to or recieve from bob.com, so since you still have to handle those errors you might not want both

(note also "send to" and "recieve from" are different channels that can independently fail, aren't distributed systems fun?) ✏