XSF Discussion - 2024-06-15

s2s smacks gets you this even for iq, but probably only for quite short windows

Email you have a solid 4-5 days, usually a week

Right. I think that's much too long for chat, but probably depends on use case

Too long for no feedback, feels like it'd be fine if you told the sender it still hadn't been delivered

The context was a "digital nomad" self-hosting, email works, XMPP not so much, but this would be helpful for server reboots and such too

I hope my s2s smacks timeout is long enough for server reboots.

Hello, XSF, I was wondering: How does the "fiscal hosting" work? What does that actually mean?

I have read the page now twice and I don't really get it. 😅️

It means the XSF can take donations on behalf of a project, they are charity receiptable in the USA, and they hold the money for you but you can mostly direct the spending as needed (via process and under restrictions necessary to remain charity status)

egnun: excellent question

singpolyma: impeccable answer

Thank you both 👍

> I hope my s2s smacks timeout is long enough for server reboots. Menel: does prosody smacks even persist across prosody restarts? I don't actually know ↺

My own restart? , I doubt it,

Bu this shouldn't be a problem, I restart often and juet rejoin without a hickup all the rooms

It has to persist on both sides or can't be used by either

If I send you a 1:1 message while you are restarting though it's lost

Hm. Isn't it in your smacks Quere still? And will be delivered when I'm connected again?

Want to try it, I'm prepared to restart, you could message test@snikket.de

No you have to have the specific stream id to resume and the counter, if both sides don't persist, can't resume https://xmpp.org/extensions/xep-0198.html#example-9

I don't think I have smacks s2s enabled...

no, it's not on by default yet

https://prosody.im/doc/modules/mod_smacks Should be on by default according to

heh. I love that that page says both experimental and default true

It's default lol yes

default true as of what version. not as of 0.12 AFAIK

It was experimentsl and off, I think at some point it was changed and forgot to delete the experimental

Hm thr doc should be for the current release version 0.12.4

I think zash said at one point Yolo. But am not 100% sure it's not only default for trunk. But generally the docs are for release not trunk ✏

Indeed it doesn't work with me restarting (with deliberate pause of 15s) with an prosody 0.12.4 on the other side. But ai don't know the settings of that one

> It means the XSF can take donations on behalf of a project, they are charity receiptable in the USA, and they hold the money for you but you can mostly direct the spending as needed (via process and under restrictions necessary to remain charity status) Ah, then I understood it correctly. Thank you, singpolyma.

I don't know, if you already know this, but apparently the NGI 0 funding is even open for people and projects outside of Europe. https://nlnet.nl/core/faq/ > Can anyone in the whole world submit? > […] The grant from the European Commission that allows us to run NGI0 is funded by European tax payers. It is a knock-out criterion for each project to have a "European dimension". […]

And anyone can apply. Individuals, organizations, etc.

egnun: yes. For example I have a grant right now. They did think I'd get past a review board easier if it was slightly joint with a European, so I did do that, but it was never clear if that was a hard requirement or more like "it would be nice" and either way my part of the money goes to me

> Indeed it doesn't work with me restarting (with deliberate pause of 15s) with an prosody 0.12.4 on the other side. But ai don't know the settings of that one Menel: Matt confirmed memory only in prosody@ mystery solved I will add investigating s2s-message-buffering and writing a plugin and/or XEP if needed to my infinite length todo list 😬 ↺

Need smacks suspend to disk feature

The own clients would also profit from it. Seamless server upgrades and clients not even noticing anything

Clients already buffer outgoing messages when their server is down

But it would be nice though probably not easy to persist smacks, you'd have to handle a ton more edge cases

I do persistent smacks in my new clients, it's very nice for lots of things

Especially web and mobile where the app might just die or reload whenever

As you know I'm clueless about the internals, in my mind I would stop listening and sending apruptly to / from clients and save all smacks state. On reboot restore smacks first and then open connections

Like same as if the network failed. Smacks should handle the rest?

Menel: oh so only on clean shutdown? Maybe, easy first step

Also likely depends how many users, jabber.org may not be able to write all that to disk before systems kills it, but it'd work for my server

Anyway s2s buffering isn't mutually exclusive with smacks, why not both? Clients already do both

Yeah, nothing will help on hard kill, won't it?

> Also likely depends how many users, jabber.org may not be able to write all that to disk before systems kills it, but it'd work for my server Need a normal shutdown way, with not saving state and a soft reboot version, (I'll be back shortly) that saves everything and restores everything. Best startet from within the server not systemd. ↺

Persisting state while running will but has it's trade-offs

> Persisting state while running will but has it's trade-offs Like buffer everything through disk? ↺

Write changes when they happen