XSF Discussion - 2024-08-11

I've now made this post publicly-visible, so I guess I should also share it here. > Wishlist for XMPP https://contrapunctus.codeberg.page/blog/xmpp-wishlist.html

Suggestion: The first time using am abbreviation, write it out. I've no idea what "cf" is. If I don't know, maybe others neither.

Menel: https://en.m.wiktionary.org/wiki/cf.#English

Basically, "cf. $app" = "in the manner of $app", or "look at $app for inspiration". To be fair, it's not a very popular abbreviation these days...

Replaced it with "like".

I don't recall having come across that one before

though, it feels weirdly familiar like maybe I did when I was a child

> I've now made this post publicly-visible, so I guess I should also share it here. >> Wishlist for XMPP > https://contrapunctus.codeberg.page/blog/xmpp-wishlist.html I agree with a lot of those points

> I've now made this post publicly-visible, so I guess I should also share it here. >> Wishlist for XMPP > https://contrapunctus.codeberg.page/blog/xmpp-wishlist.html Thanks fugata. Some of the points will be better in the monocles release next week :) ↺

The next step will be a threads page I think like you described.

fugata: I would object to > Graphics can be recognized much more quickly than text. They're also useful for illiterate/low-literate users It really depends

Graphics are trickier to convert to voice, for visually impaired viewers.

Daniel, Arne-Brün: Thanks for reading, glad to hear you agree! And thanks for your work on Conversations, Quicksy, and Monocles Chat 🙂

mathieui: I literally added that point a few hours ago, because a low-literate casual user told me that the WhatsApp UI is better because it uses icons instead of text. (They wanted to know how to share a file from Cheogram.) At the other end of the spectrum, even I have to concede that it's faster to identify an icon than to read a textual menu and select the item I need. (Despite being a fast reader, and using that menu on a daily basis for years.) I hope this 3 minute excerpt from this talk gives a good idea of the problem. Watch from 19:42 to 22:47. > "Stop Writing Dead Programs" by Jack Rusher (Strange Loop 2022) https://www.youtube.com/watch?v=8Ab3ArE8W3s&t=19m42s As for accessibility, surely a graphical icon can (behind the scenes) still be associated with a piece of text for screen readers? ✏

> The first OMEMO message to/from a new contact often can't be decrypted. I also experienced this. Is it a known issue?

I thought it was just a gajim thing. I'm curious what the cause is.

I don't think I have this issue. ✏

At least no one ever told me they couldn't decrypt my first message. ✏

I've never that issue, using conversations both sides. I think it's an issue from some clients

(well never since around 2016 or something)

fugata: glad to see so many of these are things already done by me (and now need to spread) or on my list. One that jumped out at me: > When Android users share something, the system recommends them contacts to share with Conversations and forks has this as well. I'm not sure in when contexts the OS displays it vs not but I've definitely seen it.

> They wanted to know how to share a file from Cheogram This is a curious one because the flow for that is all icons in cheogram android?

My bad, share a file * they received.

> fugata: glad to see so many of these are things already done by me (and now need to spread) or on my list. One that jumped out at me: > >> When Android users share something, the system recommends them contacts to share with > > Conversations and forks has this as well. I'm not sure in when contexts the OS displays it vs not but I've definitely seen it. Android has changed the mechanism (api) of how this worked several times. We support the first two versions of that but not the latest one

So depending on what android version you run or what api the sharing app supports this either works or doesn't

On newer apps and OSes it increasingly doesn't

Ah, an is version thing ok. I defer of course to Daniel who knows

Essentially this is one of the many things where the problem isn't "why haven't the Conversations developers thought of this" but rather why is Android coming out with a new Api every 6 month

Yeah, the icons in context menus issue is similar. Android context menus cannot have icons. I have done some reflection hacks to enable on one of mine but then in some ROMs people complain that the padding is broken. So the only way around this is to build a custom popup menu replacement. Which I guess must be possible but ugh

singpolyma: When a message is selected, WhatsApp seems to show a row of icons in the top bar. Would that be easier?

So issues in the OS is attributed to clients, issues in clients attributed to servers and issues in servers get attributed to clients. 🤔️

Daniel: thanks, I'll try and mention that in the post.

fugata: to be clear I think the article is correct in that this is currently a missing feature

I was just providing context on why some people remember seeing that feature

I still see it.

How awkward would it be to negotiate a way to say that it's okay to send error replies on the same s2s connection? Problem: Unidirectional s2s connections can succeed without having connectivity in the other direction, e.g. because of asymetric routing, firewall or DNS issues. In that case, e.g. iq-get/set can be sent in one direction but the iq-reply or any kind of error bounce can't be sent back. So, it would make some sense if this problem could be signaled on the functioning s2s connection somehow. But how?

(This didn't use to be a problem in the times of Dialback, as that would fail completely without bidirectional connectivity)

Surely just do XEP-0288?

Well mandating 288 would solve that, sure :)

No, I mean, if you want to negotiate that you can send some stuff back, then why not negotiate that you can send everything back?

Unless you want to make your partial-288 mandatory.

A smaller change that may be easier to get widely deployed?

Is it? You need to know exactly what's being sent back in order to know if it's legit. It's only really of use if you happen to have a handle of the inbound stream when you generate an error.

I appreciate that in Prosody, that's normally the case as I recall, but I'm not sure about other servers.

bidi + dialback for maximum working 😉

But yes, I should add bidi to the pile of stats I'm collecting. :-)

Oh, Prosody ought to collect some stats on bidi too

and DANE!

The real solution of course is quic, where once you have a quic session open+authenticated (regardless of which side initiated it), both sides can open new independent streams inside of it as they see fit. Doesn't even need any XMPP protocol.

fugata: > An idea pitched in the Prav community is to suggest contacts from private group chats. What does this mean? From private groups they are already in? Doesn't seem that useful because to be in such a group means you already have their contact info...

Everyone wants good decentralized contact discovery, no one has any idea how that might even be possible... 💀

moparisthebest, probably adding people from such groups into your roster, if you don't have them there already

You don't always have them in your roster already. And it's not required if OMEMO stuff is made world-readable, as is the convention in most clients now.

that could be a good thing to do, but it's certainly not "decentralized contact discovery"

I don't even know how you get in a private group with people you don't know already...

Low-hanging fruit at least

Given how much noise about the problems caused by not having public OMEMO stuffs, clearly people are in such groups.

What clients don't set OMEMO keys to public nowadays?

FWIW, for my little bubble, that sort of contact discovery would definitely be a plus, as most of the contacts I onboard are part of a large MUC I invite them to.

nicoco, a private, non-anonymous MUC ?

> moparisthebest, probably adding people from such groups into your roster, if you don't have them there already You can already do that manually via a long press on their avatar and 'start chat'. I guess doing that automatically can help people who are really really really not tech savvy but that's hardly a magic bullet for 'decentralized contact discovery'

> nicoco, a private, non-anonymous MUC ? Yup. I know that they can find each others in the participants list, the thing is, most don't. My friends are probably slow. 😉 ↺

> Everyone wants good decentralized contact discovery, no one has any idea how that might even be possible... 💀 It's called "talking to your friends and asking for their address" ↺

or sending them an account and/or account creation invite

Decentralized anonymous list? Somehow my neural net associates that with the term "blockchain" 😉

> Everyone wants good decentralized contact discovery, no one has any idea how that might even be possible... 💀 web of trust ↺

kurisu: like PGP ? I think we've got 20 years experience of that not working in practice

Not working in practice as in not popular? It's not like xmpp is particularly popular either, doesn't mean it doesn't work.

I've never known anyone who's met anyone IRL and signed their pgp key, and I know a ton of nerds

But anyway ignoring that how would you imagine it to work with XMPP ?

> Not working in practice as in not popular? It's not like xmpp is particularly popular either, doesn't mean it doesn't work. I remember PGP signing parties, today's analog would be QR code scanning parties. ✏

> I've never known anyone who's met anyone IRL and signed their pgp key, and I know a ton of nerds You could go to a Debconf, or watch Debconf24 videos, to meet some or learn, respectively. ↺

who's the target audience, nerds or normal people ?