XSF Discussion - 2020-12-07

jcbrand, bonus points if you had put a <note/> with a link to dijkstra's note containing the argument why numbering should start at zero :)

flow: I was wondering how to do that

I.e. how to add footnotes

jcbrand: example at https://github.com/xsf/xeps/blob/master/xep-0373.xml#L247-L253

Thanks, done

What's the state-of-the-art for XMPP-to-other-chat transports? Neither Spectrum2 nor Matterbridge are built for XMPP specifically, which means there's guaranteed impedance mismatches as they try to cater to the lowest common denominator.

Spectrum2 isn't build for xmpp?

Spectrum2 is just horrible

deuill: biboumi is a great irc bridge, and Bifröst is slowly getting better at matrix bridging

Not in the sense that XMPP is the singular source/target abstraction used, and less so by the fact that you're mostly relegated to using libpurple-based connectors.

And libpurple hasn't followed up with xmpp for the last decade

Biboumi is indeed wonderful, and Spectrum2, though terrible, is probably the best thing around...

I realize that chasing after moving targets and working around potential ToS violations and the like make any significant effort here a fool's errand, but Matrix trumpets this idea a lot

I was going to say that matrix also has paid developers working on their bridges, but my experience with Bifröst has been very mixed in that regard

I always thought bridging was a bad idea in general, but I can't tell if it's actually part of the reason for Matrix's popularity among the Hacker News crowd or if it's just a thing they like to talk about a lot

I think there are some dedicated bridges (not multi transport) that work ok

That is to say, I'm always unsure if it's something we should focus on or not

One for IRC and one for whatsapp

Oh and JMP's SMS bridging seems to work fine

According to people...

SamWhited: having a commercial quality bridge for some commonly used silo networks would be great, but doesn't have a business model

Also you'll run into getting blacklisted really fast

Ah right, forgot about the WhatsApp one, that's another good example. My understanding is that Matrix-the-project has been sold under this idea of being the one protocol to mesh/replace others from the very start.

Actually, that's a fair point, JMP is pretty great. I was using them as my primary phone provider for a while, only switched because I was needing group SMS more and their implementation of that was terrible. Otherwise it was fantastic.

deuill: as was xmpp, twenty years ago

deuill: there are some quotes and links in https://yaxim.org/blog/2019/04/01/yaxim-enters-the-matrix/

SamWhited: jmp has somewhat recently started a group texting beta using that pre muc xep I can't recall right now... Supposedly works well

Right, and that's probably a good way to build up hype until you've reached some critical mass -- I suspect bridging will become less prominent as a goal if the project becomes popular enough, especially given how much work it is to keep everything working.

moparisthebest, muc-light (or lite I don’t remember) ?

No it's a xep-00xx

33?

https://xmpp.org/extensions/xep-0033.html yep

Honestly I've never used group SMS but people in the jmp muc seem happy about it

Context: https://wiki.soprani.ca/SGX/GroupMMS + https://github.com/iNPUTmice/Conversations/issues/3844

seeing the phone number instead of the name is pretty jarring but it works well enough

I wonder if bridges might work well for driving client features/XEPs that might otherwise need to wait for genuine support.

For example, message reactions, threading, etc. Though I'm guessing client developers aren't exactly experiencing a dearth of feature reports without this sort of additional pressure.

I guess the XEP-0033 issue above is an example of that.

moparisthebest: that was the thing that made me finally leave them actually; it's worse than the implementation they had before because it's not supported by anything. Before at least I could see who sent a message, with that groupchat thing I just see that a MUC with a dozen numbers in the title sent a message and have no idea who anybody ise

Ah, well it's still optional, and someone threatened to write a muc bridge for it

yah, I'd waited several years so when they finally got it but it was worse than the original I just decided I couldn't wait several more years for them to use a spec that anything actually supported