XSF Discussion - 2017-11-02

moparisthebest, huh

thanks for the hint

moparisthebest, I always only saw DKIM_INVALID but I assumed that’s the mailing lists fault

moparisthebest, even though, I can find the key in the DNS without issues?

jonasw, DKIM doesn't really work with mailing lists. It only works if all you do is 1:1 mail.

dwd, I am aware of that.

it’s not that I *want* to do DKIM.

Only three days remaining for Council and Board candidates. Wow.

I might roll back DKIM. It doesn’t seem to have helped with delivery to google or anyone anyways.

and it only casues damage

I'm wondering about the MAM last call, do we have implementations of MAM for Pubsub? At least one we made for SàT Pubsub, but any other one?

goffi, Not that I know of. And our on MIX is out of date, since it needs the pointless id added.

is not needed to have 2 implementations to do a last call on MAM then?

actually the one on SàT Pubsub may be outdated to.

goffi, That's Final, not Draft, isn't it?

ah maybe

in retrospect, DKIM-Spam-Issues probably also explain some of the XHTML-IM discussion :/

iteam, can we configure the ML so that it mangles DKIM-signed messages in a way which avoids incorrect signatures?

some people don’t have a choice on whether they send DKIM signatures

jonasw, if you want to review component/protocol etc it's to a working state, pre-compiled .apk here or compile it yourself: https://github.com/moparisthebest/Conversations/releases

a compiled build is not really useful to review something, is it?

I can’t into mobile development, at all

it links to both the mobile code and the component

I think additionally the component needs to advertise in disco that it's an 'echo component' and that should be good enough

Which kind of review?

And of what?

any kind and of any of it :)

moparisthebest, frankly, I’d prefer a written specfiication over having to read java code.

but I’m at work anyways

the specification is there too in the readme, such as it is https://github.com/moparisthebest/xmpp-echo-self

ok, might take a look this evening

also the entire component is 130 lines of java code including comments

I wasted about 4 hours trying to get sleekxmpp or slixmpp to work for this and failed...

anyway being able to seamlessly SMS through my phone from gajim is sweet, or any other xmpp client supporting carbons

I still find your (ab-)use of carbons... non-satisfactory

this is a hack for something which deserves a proper solution.

Does this even make sense?

I wouldn't call it a hack, multi-client is useless without carbons anyway, not sure why the component should be made complicated and require registration just to support use without carbons...

it is a hack.

Transports being attached to the server is one of the very earliest design decisions in Jabber, so doing this in a non-awkward fashion will be Hard.

this is one of the problems with some XMPP specs, instead of solving the simple problem simply, they attempt to solve complicated problems no one actually has yet but might have in the future in a more complicated way

I think it makes sense to have per-user transports if the transport is linked to a hardware token owned by the user (e.g. SMS)

moparisthebest: IM is not a simple problem

It may look simple, but it's not.

this solves the problem of a client-side sms transport and works perfectly so it seems rather simple and it works

and requires no changes to other clients or servers, most importantly

"Works" and "Hack" are not mutually exclusive.

Hacks usually work.

why is relying on other already defined XEPs a hack?

moparisthebest, but doesn’t it strain the server by intentionally sending messages to non-existant domains?

Making something hacky that works is a good first step, as it helps understand the problem.

nothing stops anyone from making a new XEP that does complicated state management and registration to avoid carbons, it's just useless

jonasw, the domain exists

ah, the domain of the "echo" component?

it's a standard xmpp domain/component

yep

in fact my pre-compiled apk is pointed to my component running at echo.burtrum.org and works for anyone that can federate to that

this should be replaced by discovery and fallback to a hard-coded component, eventually

Eh, is it wise to publish a thing that would leak all your SMSes to you?

as a super alpha with a warning about it? sure

Zash, is it wise to install a random apk from a random persons github page? ;-)

I wouldn't, I'd compile it myself pointed to my own echo server instance :)

I also just found out about this today: https://github.com/SilenceIM/Silence/pull/390 the SMS app is planning on adding a transport-over-xmpp option hehe

yetanothermessengerapp

moparisthebest, what kind of issue did you have with slixmpp?

Did you start from the echo_component example?

yes Link Mauve it alternatively would return strings or parsed element structures but I could never get them back into a Message that I could .send etc, my 3 failed attempts are here: https://github.com/moparisthebest/xmpp-echo-self/tree/master/failed_python_attempts

Strings or parsed element structures?

so depending what I tried the <message> in the <forwarded> block would either be an Element or just a string

Some of your code reads like it should work.

moparisthebest, you need to load the xep_0297 plugin in order to get a Message, probably.

Link Mauve, yea I did that too check xmpp-echo-self.forward.py

xmpp-echo-self.py doesn't use forward and works with sleekxmpp but not slixmpp, but then I decided to go with forwards anyhow

moparisthebest, the usual way is:

forwarded_message = msg['forwarded']['stanza'] forwarded_message['to'] = your_user forwarded_message.send()

From what I understand you want to do.

Link Mauve, yea so sometimes I'd get an error on forwarded_message['to'] = your_user like 'str does not support .attr()' and sometimes I'd get an error on forwarded_message.send() like 'Element does not have .send() method'

moparisthebest, that’s if you don’t have the plugin loaded.

(Yes, slixmpp defaults to '' for unknown attributes…)

(Yes, I hate it.)

(And my branch changing that broke a ton of code in poezio alone, so I didn’t try to go forward.)

(I want to test aioxmpp, if it managed to be less non-typed than slixmpp.)

(But what I want the most, is a proper Rust library.)

(xmpp-parsers is based on the opposite idea from slixmpp’s, types are good and you should use them, always.)

(But it only parses and serialises stanzas and payloads, it isn’t an XMPP library per se.)

Link Mauve, aioxmpp is rather strongly typed

:-)

I should try to move poezio to it, someday.

sounds like a plan ;-)

yes Link Mauve completely agree I wanted to do it in rust but not quite there yet :)

anyway I spent a solid 4 hours trying different things with sleek and then slix then gave up so meh...

would have tried aioxmpp but jonasw said no component support yet

I didn’t say "yet" ;-)

sorry sorry :P

moparisthebest, both xmpp and tokio-xmpp support components, fyi.

anyway Guus whack is nice and simple, typed but also super easy to just grab / manipulate XML directly, I like it :)

I personally added that. :)

The only constraint is that you can’t use both a client and a component when you import them in your project, it’s a feature which toggles the component or client thing.

still need a new version of xmpp released though, current doesn't compile with stable rust, had to use git for my client thing

Wait what?

Someone mentioned me?

yep, whack compliment

Yey!

Didn't work on that for ages though 😀

Also mostly code from Jive

everything doesn't need to be developed for latest node.js in the last week to be good :)

Happy to hear it's helpful for you.

moparisthebest, You're still developing on last week's node.js? Luddite.

My node.js is from last month!

Okay, that was a lie. My node.js is 4.4.7 LTS, which is from mid-2016.

wow that's like 6 versions behind

8 is LTS now

I had an update to 9.0 like yesterday, you’re all so out of date…

node has to be the only software where LTS means like "ok I guess we'll support this version for 3 months"

lolwut

Sounds like node.js is the software pendant to memory manufacturers. 🙃

What should happen if you send <presence type="error"/> (with nothing else)?

should be dropped by the server?

Zash, A little window should pop up telling you you have too much time on your hands.

> Discarding unhandled error presence (<nil>, unknown condition) from c2s: <presence type='error'>

Zash, I preferred my error message.

The words "time" and "hands" should be replaced by the respective Emoji..

:time::hands:

I just tried it with Openfire. A world of nothing happened.

same for ejabberd :)

https://github.com/openspace42/aenigma "aenigma provisions a fully functional and secure out of the box XMPP server you can get running today."

never heard of that before, looks like bash script to set up ejabberd? odd

there are also nginx and sslh in the bundle ;)

I only saw it because dev emailed a question to the sslh mailing list :)

which clients do support "xmpp over ssh"? Or do you need an ssh-client setting up a tunnel?

ahh s/ssh/ssl/

xep-368 style? conversations does

I had a list of implementations in the email hang on...

moparisthebest, He didn't mean that.

oh got it

we need more board & council candidates here ;-) https://wiki.xmpp.org/web/Board_and_Council_Elections_2017#

I might volunteer

Ge0rG: go, go , go

what happens if there aren't enough candidates actually

I've already started my election page, but had to leave home.

I was kind of hoping that we’d have more candidates than seats so that the election actually gets interesting :/

moparisthebest: :-S we did not have such a case in the entire career of the XSF yet. Would have to lookup the bylaws then or apply myself ;-)

in the last years applications for board and council always came in close to the deadline

Didn't we have situations in the past where the number of candidates matched the seats, making the election rather pointless?

It's not entirely pointless.

You don't get elected automatically you, still need to get elected.

Not entirely, no

I imagine that my council application will be rather controversial

that’s a good thing, right?

not that Ge0rG guy...

haha <3

Make XMPP Great Again!

I'd buy that hat

Tell the SCAM team

well played

That motto is kind of burned. Besides, it should be "Make Jabber Great Again"...

Ha

I realized my error some time ago, but then Drumpf was elected already, doing bad things, so...

oh right, no I'd vote against you if you ran on a slogan with Jabber in it

"Make Jabber Ours Again"

moparisthebest: why so?

Ge0rG: For your own protection (against trademark infrigement)(?)

yep!

I know my trademarks

zinid, so the sslh question guy is saying ejabberd will support '368 next version, in what way(s) ?

moparisthebest: what is 368

I don't remember all those stupid xep numbers

-xep 368

Zash: XEP-0368: SRV records for XMPP over TLS (Standards Track, Draft, 2017-03-09) See: https://xmpp.org/extensions/xep-0368.html

it supports it already for outgoing server connections

"Legacy SSL on port 5223 but with SRV records"

presumably it already supported legacy ssl on port 5223 for c2s

but what else?

in theory it should support sni, but it doesn't

sni for incoming connections