XSF Discussion - 2018-03-08

waqas, I haven’t used it in a long time, could you write those down somewhere?

XMPP could use a pretty webclient.

Pretty according to whom?

But yeah. Do it. Hire UX engineers and graphics designers and a huge marketing department and do it! :)

according to whatever the designers who make pretty but non functional UI's people love.

converse.js is pretty good

movim is pretty good but one would have to pull out all the other stuff so it's easy to just throw it up on a webserver

It's very much optimized for "communicate with the org". There's the basic MUC-like part. Room discovery+creation is much better than most other clients. Search is a big deal. Room history (reading it, and coming back to a room after being away for a while) works better than any XMPP client I've used so far. Multi-user private chats (temp MUCs?) are very easy and very actively used. File sharing (within a MUC or single user chat stream) is actively used. Threads are started to catch on, and leading to easier conversation in active rooms. Slack bots are cool (and there are many, tied to services that can be used without any ops overhead). The team overall finds things like "/giphy", emojis and custom emjois fun. We don't like the audio/video function, as it's always breaking for somebody. Nobody likes how heavy/slow the client is.

My off the cuff summary ^

hah, they had to restart their irc server 3 times this evening to add the banner warning about the death of gateways

The price/user is basically irrelevant. It's under $10/user, and for an actual business that's just noise on the balance sheet. The ops free nature is attractive.

Last time I used slack in the browser it had some serious performance issues

It's slow and bloated. I use it in the browser only, and don't use the "native" app.

it still does, waqas addressed that already

Thanks waqas.

But is it slow and bloated enough for the higher ups to care?

There was something very wrong - like I would leave the window open for a day and come back and the browser had some to a halt - I started running it in firefox

(I assume not)

*come to a halt

Zash: The lower-downs want it. Because it's as close to friction-free as I've seen anything get for various team chat scenarios.

Slack in our org was not something that management mandated, it was something that people on the grounded tested out and wanted. I assume that's how it gained massive popularity.

I think the main attraction to slack is it's easy to use and setup your own private space, and even though the client isn't amazing it's good enough and just a webpage. Easy way to onboard people.

Although I hate hate hate Slack threads.

compared to slack I think for casual stuff discord got it way more right - send you a i

*a 'invite' link - you don't need to give your email or anything, just enter your name

then we can use voice chat, chat, etc.

jonasw: editor issue, are these CFE's meant to include the text about advancing to final?

Note that our org used to be an XMPP-using one. One of the things that caused hate for XMPP: Jabber file transfer never worked. Root causes varied, everything network issues to client issues to bad client UX. It was a running joke.

My friend was just complaining to me that conversations prompts which resource to send a file to.

I don't think I have ever sent a file over xmpp but isn't it a lot better now with http upload?

And we transitioned to slack over time, the engineering team was the last holdout, but eventually everyone gave in to the "ease of everything"

Just tried now with movim sending to this account. It was pretty painless. Went to my phone and gajim without issue :)

oh weird - I tried to send a image from conversations to the movim account and it is prompting me to choose from 3 resources. I'm guessing this doesn't happen with http upload

No upload service available?

yeah that's my guess. I don't think my server has http upload (we never send files over chat)

oh weird hah I try and send it and my browser pops up with "is trying to call you"

Well, it got Jingle payloads confused?

I guess so?

does http upload play nice with omemo or is it unencrypted?

HTTP upload doesn't really need to know what the bytes you stuff in it are

Not sure how standardized the encryption methods in common user are

yeah looks like the file is encrypted then the key is exchanged over the secure session

https://conversations.im/omemo/audit.pdf

> waqas: Note that our org used to be an XMPP-using one. One of the things that caused hate for XMPP: Jabber file transfer never worked. Root causes varied, everything network issues to client issues to bad client UX. It was a running joke.

oh he left, but my work uses lync or now skype for business

and file transfers essentially never work

Zash, pretty sure, yeah. that’s what a CFE is about

>> 1. What software has XEP-0048 implemented? > Just about every client, library and server (indirectly, via either storage mechanism) I've ever seen. except Pidgin. ✏

jonasw: pidgin doesn't have bookmarks?

Really

daniel, yeah

MUCs are only in their local roster

jonasw, daniel: Pidgin does support it, but only in a plugin that's not enabled by default

MattJ, how’s that called? it doesn’t seem to be installed by default on debian

maybe in the "pidgin-plugin-pack"

I imagine pidgin-plugins-bad and pidgin-plugins-ugly, gstreamer style

hah

in case of pidgin it would rather be: pidgin-plugins-normal, pidgin-plugins-without-zerodays

(at least so I heard)

in the case of pidgin I’d also be worried what it’ll do to my already existing bookmarks when I turn the plugin on. on either side of the wire.

> In addition, the service SHOULD send an invitation to any user who has been added to the member list of a members-only room if the user is not currently affiliated with the room Nice things you discover in 0045 by accident.

Kev: I'm confused by what you said yesterday in the meeting. The muc_* features seem to actually be used as runtime configuration options, but the XEP only ever talks of them as being supported features

Give me an hour to wake up, and then I can think about this

Ge0rG, here you go https://www.youtube.com/watch?v=rTyN-vvFIkE

Tobias: that's perfect!

Great score while I wait for some integration tests to complete.

they also take exactly an hour?

Sure feels like it.

jonasw, https://developer.pidgin.im/ticket/6353 :/

Opened 10 years ago Last modified 4 years ago

sweeeet

Ge0rG: As far as I can see, the 'features' are done as http://jabber.org/protocol/muc#something, while current room configuration is done as muc_something.

Kev, Example 9

(in § 6.4 Querying for Room Information)

Kev: while I agree from reading the examples, this is not really answering my question.

What was the question, then? :)

Kev: besides, there are no *features* with http://jabber.org/protocol/muc#something in the XEP, nor in any other XEPs.

Kev: my question is, how you came to the conclusion that http://jabber.org/protocol/muc#something are features and muc_something are runtime options.

So far the only supporting document I found is: https://xmpp.org/extensions/xep-0307.html#support > If a MUC service supports the protocol specified herein, it MUST advertise that fact by returning a feature of "http://jabber.org/protocol/muc#unique" in response to Service Discovery (XEP-0030)

Just that the first three http... features are 'Support for...', wile the others are descriptive.

But happy to be wrong, just push it as-is.

Kev: oh, I feel like I'm in a position to actually improve 0045 now, I won't abuse my powers to do a half-assed job.

> Support for reflecting the original message 'id' in 'groupchat' messages. Sounds kind of weird. I'll leave the original wording.

Kev: https://github.com/xsf/xeps/pull/600/commits/b017284eea20070b7982d6fb58fe6c68c8c6dc54

Ta. I'm more comfortable with this, whether I should be or not.

Do we have all council votes together then?

Think so.

From the MUC logs, I haven't explicitly voted on that one.

jonasw: you can merge https://github.com/xsf/xeps/pull/600 now, thanks :)

Ge0rG, there you go

> Pull request successfully merged and closed 👏

where did my mail about xep-0020 go though

I see one titled [Standards] DEPRECATED: XEP-0020 (Feature Negotiation)

ah, just took a while

(weirdly, it appeared several minutes after the UPDATED: XEP-0153 one)

maybe funny lag between the mailservers :)

jonasw, Konstantin noted that "Deprecating is not a serious problem" for XEP-0071 and that got me thinking - didn't Council vote to Deprecate, not Obsolete? Or am I misremembering?

huh

you are indeed right

that’s my fault, I’m going to rectify this immediately

well, editors fault at least

The references to XHTML-IM in other XEPs should be removed, I know there is at least one in XEP-0277

(and anyway current implementations i.e. Movim and SàT, use full XHTML)

jonasw, Thanks.

this is a busy day for the xeps repository :)

https://news.ycombinator.com/item?id=16539857 (slack is closing XMPP gateway, which was shitty anyway)

goffi: Can you comment on how bad the gateway was?

waqas: it's really basic, not implemeting any advanced feature, it's hard to join or discover rooms.

goffi: How does it handle multi-user PMs? Temp MUC invites? And what about threads?

waqas: you have MUC room named from pseudos, something like mpdm-[nick_1]--[nick_2]--[nick_3]--[your_nick]@your_company.xmpp.slack.com

waqas: to give you an idea, here is the disco of the server:

http://jabber.org/protocol/disco#info http://jabber.org/protocol/disco#items jabber:iq:last jabber:iq:privacy jabber:iq:version urn:xmpp:ping vcard-temp

Somebody should write a proper bridge now.

I'm mainly surprised at jabber:iq:privacy

Ge0rG: I think there was one for libpurple, but unmaintained

So spectrum again. That software stack provides me with nightmares

> I sooooo regret even having the IRC gateway. Slack CEO https://twitter.com/stewart/status/817483176687718400

The thread is awesome

Handicapped users will appreciate that slack "is working on" accessibility.

jonasw, “Note that I intend to overhaul XEP-0394 and I don’t know any implementations.” about Message Markup, well, slixmpp has had one since day one. ^^'

is that used in a client?

I think poezio will use it if there is no XHTML-IM alongside.

I see

Otherwise it will prefer XHTML-IM.

Link Mauve: what does "alongside" mean?

On the receiving end?

Yes.

On the displaying end, more specifically.

Ah, so it's not sending it.

Nope.

What about displaying *Styling* instead of ~markup~?

I couldn’t find a good way to mangle our composing format into either of these new XEPs yet.

They are way too poor to express this format.

The good thing about Styling is that you don't need to mangle anything, you can just live-apply the styling to the input box

That means changing our input format for something that can’t be disabled.

Currently all styling we have is opt-in by using keybinds.

I can write *stars and bold stars*, and these aren’t coupled in any way.

With styling you can’t disable that.

If you want to put the emphasis on a specific word without displaying the stars around, for instance.

But people have been listing these issues for a long time on the mailing list and here, yet people didn’t care in any way about them, so I don’t think I have much else to add.

Link Mauve: I'm aware of the arguments, and I even used to be a proponent of the explicit approach before I tested Styling. It's an easy and compelling way to markup things, and it even works over IRC transports!

XHTML-IM also works over IRC transports. ^^'

Link Mauve: I'd like to hear your definition of "work" bent sufficiently to support your statement.

Ge0rG, biboumi converts XHTML-IM to actual IRC markup

and vice versa

there’s no bending there

it’s actually done, including colors

it’s insane

(and incredibly cool)

IRC has approximately the same limitations as a console client, it can only change boldness, italics, invert and colours, but it works quite well for those parts.

I agree with that statement

So I don’t buy the argument that biboumi should convert my ~~words or **words into IRC colours, that’s removing formatting fun and adding complexity when the recipient doesn’t see what you wanted to send.

I was rather thinking about clients that already support ** and __

And those that will have to be changed because they assigned another meaning to them before.

Further increasing confusion.

jonasw, I though biboumi only did one-way converting?

mathieui, hm, dunno; I have seen coloured things already.

I think at least.

yeah, from IRC it works but I don’t think the xhtml-im → IRC is even implemented

Reading the xhtml-im deprecation thread, I realized that people are not even aware of 0001. Changing Draft to something else wasn't bad but probably not enough

pep., until someone invents a device that can slap another person over the internet what else can be done?

(I *really* want one of those devices)

"stab someone into the face over standard tcp/ip"

moparisthebest the (in)famous slapper

well it'd need to bypass firewalls and NAT so I'd vote over HTTPS :P

pep., tbf, I wasn’t aware of 0001 until I started authoring XEPs either

👨‍💻🖐️🤚👊

/kickban moparisthebest stop making people sad :(

jonasw, sure, same for me, it took a while

see jonasw , you could have used the slapping device on me right then...

s/slapping/stabbing/ pls

need to be more rigorous here ;-)

jonasw, we're not criminals

Or I'm out!

ohkay ohkay

pep. lies.

board meeting in 5?

Ge0rG, you'll love this: http://friedberger-nachhilfe.de/ a website that requires you to fill out a captcha to VIEW it :)

SaltyBones: wfm

pep., FWIW, the website portal for the XEPs (ie, https://xmpp.org/extensions/) could probably use a link to it, even though it's the first one listed.

Also it says you can show/hide the various *types* of XEP, when it then shows Statuses.

SaltyBones, that's what cloudflare does as a company

SaltyBones, You're using Tor, I assume?

nyco, Also, yes. Although I was distracting Guus, so I may have made him late.

it's now

I'm here.

Martin excused himself

Ralphm, MattJ?

Matthew and Ralph are around

Here

nyco, did you talk to Ralph just now, or were you only observing that he's in this MUC?

talked

5 min ago

oops, no

that was me

the other handsome Dutch guy :)

let's still start? and end on time?

agreed

https://trello.com/b/Dn6IQOu0/board-meetings

Do you want to take this one?

moparisthebest, Dave Cridland no, without Tor! Indeed it's not funny otherwise. :p

1. Role Call and agenda

We've established who's here. Anything else for the agenda?

none from me

Nor me

2. Confirm minute taker

Who would be so kind?

>awkward silence<

I'll type them based on the logs then

thx

3. Topics for decision

3.1: Board Priorities 2018

we're holding off for that meeting, right?

yep

still waiting for answers

regarding preferred dates and times

nyco is working on that, I think, given that you asked me about availability earlier

correct

anything we can add to that here and now?

probably in two weeks, afaik

ok, thanks

3.2: Bus factor / bank account

apart from availabilities, if you got to push something regarding the organisation, feel free

As promised last week, I've reached out to Peter for an update, But have not heard back yet. I'll prod again.

(nyco, re prio meeting - I don't but I'll include that invitation in the minutes for others)

4. Reviewing commitment list

we've covered the prio meeting

Ralph expressed earlier that he wants to hold off the ED replacement search until after the prio meeting

wfm

do we have to wait?

No, but I think we'd prefer to

nyco, I'm not sure if we _have_ to. I'm still unsure exactly what entails that role - which might tie into the prio meeting.

Part of the problem is that we can't find people for the role until we can exactly define what the role is :)

got it

the third card, adding commits - I think we'll cover that in that meeting too. Nothing much to do there now, I think?

agree

archive?

maybe put in a different lane for prio meeting?

sure

or just leave it for now, meh. I don't have a strong preference.

Last card, drafting a membership survey - Mattj, did you get around to prepare something?

I didn't yet, sorry

if you need help, please ask

Can I tempt you for an ETA? :)

Next week? :)

awesome :)

it's not that I'm good at it, I'd rather have someone really competent to do it, but I have a few insights that I'd to have confirmed/invalidated

+1W

nyco, you did the last one, didn't you? That gave good results.

nyco, I'm not claiming to be especially competent at it... if you have some ideas, feel free to share, or take it on yourself

I just feel it should be done

ok then I'll push you the few questions/doubts I have, you do what you do of them, ok?

Perhaps you guys discuss this between the two of you, out of band?

ok

wfm

5. Items for discussion

5.1 Fundraising and financing

if we do that conference, we may be able to generate some money?

I'd actually would like us to look at that in more detail, but perhaps that should be prepared for better.

can be a board prio, btw

this may mean we'll discuss that during the prio meeting

I think addressing the money issue is something that we're responsible for anyway, wether or not it's a prio.

totally

I think it might be good to not heap to much on that already very broad meeting.

makes sense

Let me prepare for specific financial / fund-related talking points for next meeting - unless someone wants to add something now.

I'll take that as an 'ok'

6. AOB

anyone?

nope

None here

from the floor perhaps?

7. Time / date of next

I'll be unavailable next week

+1W

Next week works for me

but if you guys are available, I'm ok to skip one.

ok, two out of four will be there. It's likely that you'll get at least one more, I think

Ralph and Marint may like to join

I think we should continue unless others also indicate they can't make it

lets do +1W

we are in agreement

thx MattJ and Guus! ;-)

thank you guys

Thanks Guus

Sorry I didn't see there was official stuff going on at first and didn't want to interrupt (isn't this the wrong MUC?). Anways what I wanted to tell you: OH MY GOD I just successfully signed and verified messages using my XEdDSA python implementation! Prepare for a new OMEMO lib guys :)

Syndace, great news :)

no pb Syndace

\o/

I'd say that's relevant to this MUC, it's on average used for official meetings for 30 minutes every week

oh memo!

24 this week, we were being efficient ;)

Syndace, wow nice!

It would be interesting to make java port at some point, to create smack-omemo-syndace 😀

Haha thanks guys! 😁

Syndace: great, any idea when we could test a public repos ?

Syndace, amazing!

congrats

Well most of it is already prepared and I just need to switch the flag from private to public on GitHub. I hope it'll be a thing of one or two weeks (but I tend to underestimate developement time :D )

why don’t you switch it to public right now?

tack a huge "UNAUDITED CODE" warning on it and go for it

Syndace: will Python 2 be supported ?

I hope not.

python 2 won’t be supported at all anymore in 2 years time

jonasw, I thought about this but I really want it at least to work before I make it public

jonasw: yes I know

Syndace, I thought it does now?

otherwise it makes sense :)

Yeah Python 2 is supported (actually I don't know whether Python 3 is supported)

aand I lost interest.

;)

The last piece is working but not integrated yet

seriously though: python2 is going to die soon-ish. if there are issues with py3, I’ll be happy to help out

jonasw: not all software have switched yet, so python 2 support is good news for now

Syndace: for me Python 2 support is interesting only for a couple of months (until I fully switch to Python 3), but the support is really good news as it means I can start implementation quickly.

Good to hear. When I started working with Python about two years ago I was forced to use Python 2 because one of our dependencies did not support version 3 and I kind of stuck to it. I know of most of the differences though and I think I wrote the code to work in both versions equally.

Syndace, tip: set up travis CI (or another CI) tests for each version you’d like to support

it’s actually quite trivial to do with github

Syndace: I was also stuck by dependency (Twisted), but the port is done now.

Re billion laughs, https://modules.prosody.im/mod_conformance_restricted.html

still need to figure out how to integrate that in my test suite.

> Send “PI”, “comment”, “DTD” or “entity” to xmpp:example.com/conformance, while directly connected to the Prosody instance.

as <body/> in a <message/>?

Yes

jonasw, is python 2 going to die soonish though?

because it seems like 2 and 3 will live on all computers forever at this point

moparisthebest, once upstream support is done, I’ll start to bug every project still using it.

python 3 is like MIX except there are actual users

I wish distros would stop encouraging pidgin (aka not installing it by default) because pidgin - least last time I used it - was really really bad for XMPP

all the "why does this happen???" xmpp issues were resolved when I switched to gajim

Jack of all trades

But what does it actually do besides XMPP and IRC?

Ceterum Censeo Pidgin Delendam Esse.

They don't ship Pidgin for its XMPP capabilities

jjrh, ICQ, Facebook(?), …

Does anyone actually use ICQ anymore? Facebook is broken for sure.

dunno

like don't get me wrong 10 years ago gaim was great

but these days everything is mostly "use a browser and be lucky if it even sorta works with a native client"

Like I used to use bitlbee for facebook chat for a long time but eventually it stopped working and a whole bunch of features like group chats didn't work so I gave up and just open messenger.com

do non-tech people even use desktops/laptops anymore?

my mom and wife haven't used one in years at this point

(I sure hope those are distinct people)

they just do everything on their phones

haha yes jonasw

ah yeah, plural

the first four words are not unambiguous

Well I have been trying to convince folks at my work to use gajim instead of pidgin - it's slow going.

one person is all "it doesn't have IRC so I don't want it" Like comeon - IRC in pidgin really really sucks.... just use xchat if you want a gui

jjrh, deploy a biboumi

amazing IRC<->XMPP gateway

How does that work? is it just a irc bouncer?

jjrh: it allows you to use all your xmpp clients on IRC, at the same time

jjrh, https://biboumi.louiz.org/

it maps IRC to MUCs

pretty great

Yeah so it's just a bouncer - neat.

How does it work for the authentication? You configure that all through your XMPP client or do you need to configure the biboumi gateway?

you configure it through ad-hock commands, so you need like gajim

jjrh, yeah, ad-hoc commands.

wow extra k where did that come from

Ah cool. So what is it SASL? works.

but once configured, it works great with conversations and such too

it doesn’t do SASL AFAIK

but you can add a PRIVMSG to nickserv on startup if that helps.

https://lab.louiz.org/louiz/biboumi/issues/3103#note_516

https://lab.louiz.org/louiz/biboumi/issues/3103#note_941

yeah the main one is having freenode work

(freenode should just run a xmpp gateway! )

most IRC servers, I think freenode too, just lets you send your nickserv password as the server password

yeah so it's not the end of the world.

I think i'll give it a go - would make one aspect of my life a little easier. Tunneling the weechat remote over ssh on my phone is not great.

oh yes

just make sure you don’t allow remotes to (ab-)use your biboumi

Yeah I can see spammers going to down with that :P

*town

I started reading xep-0369 - has anyone implemented it yet?

iirc there is 1 proprietary server implementation and 1 open source almost-implementation

but no it's basically vapourware

I guess SWIFT client has support

I see a request on prosody - whats the open source server implementation? ejabberd?

it tries to solve every use-case on the planet, far too complicated if you just want good multi-user chat, in my opinion of course

I want to say openfire...

haven't finished reading the spec yet but I'll agree it is rather complicated. It however gets much much closer to being like 'slack' where you're a 'member' of a room but not necessarily active.

IMO MIX is rather sane by now, feature-set wise

it solves some nastiness of MUC

(speaking as a client developer)

there are other things to do this https://xmpp.org/extensions/inbox/muc-light.html https://docs.ejabberd.im/developer/xmpp-clients-bots/proposed-extensions/muc-sub/

my main issue with it is that it integrates weirdly with the roster.

I have no comment as to whether they are better/worse than MIX

That functionality would be useful for certain types of deployments - aka a company XMPP server where new users get automatically setup with all the varying rooms - so they /see/ what's there not necessarily actively participating.

don’t link things in /inbox/ as "there *are* other things"

it is a thing isn't it jonasw ? :P

depends

jjrh, you can do that with autojoin-bookmarks (many clients will follow them) for MUC too

MIX mainly has the advantage that an *account* is joined and not an individual client. this makes things simpler because it takes the load off the client regarding keeping state in sync.

Yep, but you still will get the "Someone has mentioned your name" pop up

yeah, that’s true for both MIX and MUC?

Don't you need to be /in/ the room with a client for MUC?

not just bookmarked

you need to

but most clients will follow autojoin-bookmarks

so if you put them in their account on sign up, they will be joined right away

pidgin will of course not, because pidgin is stupid

Right but if I close the window i'm sol

I don’t know what "sol" means

shit out of luck

but if you close the window and your client takes that as "leave this chat" it might very well make you leave from the MIX.

that’s not a protocol issue

I read this idea as I have jonasw as a member of xsf - with the theory that I should be able to signal your attention to that room without your client actively listening / receiving all the junk going on. You may also want your phone to only have the alerts / important information open and not all the other channels you participate in - but you still want to be queried on your phone when someone says "jonasw did you forget we have a meeting today?" in xsf

Guus, others, apologies, I needed to play taxi unexpectedly ✏

jjrh, the client will always receive all messages from a MIX room your account is joined to

independent on whether you are mentioned or whatnot

I thought the whole idea of keeping you as a 'member' of a room was for situations where you might not want your phone sitting in a really busy support channel because it eats up a lot of data but on your desktop client you don't care

no

the idea is that you don’t get removed from a room just because right now none of your clients is online

Ah. That would be a neat feature to have - having your phone only subscribe to the pubsub node and when you say my name or 'signal' me somehow from the context of the chat room my client would alert me i'm needed. Also be a handy feature for things like I want to get important notifications about prosody - and may want to weigh in - but the majority of the time not looking at that window. Stuff like a broadcast that there is a majority vulnerability and you should upgrade asap, or a new version release.

there is no pubsub node which does that

(in MIX)

of course, using pubsub nodes for such announcements would be an interesting, but also totally separate concept.

In the little free time I have at the moment i'm trying to get to the point of understanding the gajim codebase enough to implement better pubsub support to do stuff like alerts/notifications. Instead of having a MUC bot dump git commits that would be a pubsub node - and ideally sub nodes for different branches, so I can subscribe to say a feature branch that interests me or conflicts with my work. (but majority of people aren't concerned with that)

there is subscribe-options which could do that -- if you’re running a specialized pubsub-service

kind of like a mailing list type thing?

which should be trivial to do; a pubsub service can be run on a resource IIRC

could be interesting

I need to implement a skeleton for a pubsub service in aioxmpp, so that I can deploy bots which act as pubsub services.

My thought is more instead of getting my inbox spammed with nagios alerts I would get pubsub notifications. Ideally doing something smart like having different nodes for various servers. I got the idea from moparisthebest 's sendmail xmpp thingy - have a email account you add to receive alerts to and a bot thingy that converts the emails to xmpp pubsub nodes or msg/muc. Everything already supports email so the overhead to support this is just adding another email to receive the notification.

I did something sorta like this for email alerts to SIP SIMPLE.

that would be interesting for mailing lists like all the xsf ones

Yep. Just about everything supports at least email so you're not writting a plugin for all the different platforms.

'mailing lists in xmpp' seems to be something that someone else would have thought of and implemented already

I wonder if there isn't already a XEP

The thing is I haven't found any clients that really have good pubsub support

What does "Really good pubsub support" mean in a client?

Pubsub is used in the context of things, rather than as a thing on its own.

Kev, I think actually just subscribing to a node and receiving messages from it.

I would see - for this case - a window that I can browse. Like movim does

Sorta like a RSS reader I guess

Neither of those sound like a generic client thing to me.

You subscribe to the node in the context of something specific (e.g. Swift will soon support subscribing to FDP nodes)

What is a generic client? :)

A think a client should generically support.

I think IM client would be how I'd describe what you're describing

Since an XMPP pubsub feed reader is also a client (yes, I'm being picky, but just trying to clarify the conversation)

Fair.

https://i.imgur.com/BU0T7s3.jpg

something like that - a window that lets you navigate the hierarchy and new nodes would obviously ping you.

jjrh, but is it like a mailing list, can you reply to specific nodes :)

> For example we may rename XEP-0393 to "Markdown" That made my day.

I really like it either way

Not sure - it's just the idea I have in my head. Just being able to view this stuff more easily in a client would be useful.

> Also if we'd do that, we'd have "Message Markup" and "Message Markdown"... Where can I vote on that? 🤣

jjrh, I'm not a UI guy at all, that seems great for desktop, but how would it look in mobile?

Not a UI guy either - so no clue, I would probably look at how email mobile apps do threading

moparisthebest, as for XEP-0393, I'm sure its implementation will be awful both on desktop and mobile.

but probably something like showing the top level stuff, then you click 'security' and then it shows all the sub nodes for that and so on. For new alerts you would just get a notification showing you what it is (kinda like a new email notification - you get the first 15 words or whatever in the notification)

Yagiza, do you mean https://xmpp.org/extensions/xep-0369.html ? 393 is the styling one and it's already great everywhere :)

moparisthebest, I'm sure not so great.

moparisthebest, XEP-0369 is interesting but seems somewhat complicated, so needs further investigations.

Yagiza, gajim and most other desktop clients, including IRC and email implemented basically 393 before it existed, conversations implemented it immediatly

I don't see the problem at all

I'll probably initially - specifically for email alerts - just dump them to a MUC channel (or private msg) which will work with every XMPP client. This would work /today/ with all the XMPP clients. (the main users I have in mind are on a proprietary client)

kinda stole your sendmail idea moparisthebest and did a hack job with a libstrophe example to send me a private message from a git hook. The server with git on it doesn't have python 3 and all that stuff and I couldn't get rust to cross compile to 32bit so C was easiest.

it wasn't really my idea, perl sendmail is super ancient and didn't work with my server, I copied/hacked that python one from someone else to do PGP and such, then an openssl upgrade broke python and I hacked the rust one together :P

*perl sendxmpp

Well I got the idea from /your/ code. :)

still if you wrote a C sendxmpp you should put that code someplace public :)

wait, 32 bit server in 2018? interesting :)

I wonder if I still have my sendxmpp-curl somewhere

I don't want to talk about it haha

I was as surprised as you were

moparisthebest: Hey, you leave my supercomputer alone!

It was cool in 2003

Zash, you added xmpp to curl?

moparisthebest: No, I wrote a plugin for Prosody that received stuff over HTTP and a curl wrapper compatible with sendxmpp

https://modules.prosody.im/mod_post_msg.html 8 years ago!

But 'wrote' is a bit of a stretch, I found some example from a PR and glued it together. I'll throw it up somewhere. Right now it's just a make target for libstrophe since I couldn't be bothered to figure out how to correctly link it.

ah ok, neat

adding xmpp to curl would also be neat

it does imap and smtp and such, why not xmpp

$ q ~/share/bin/sendxmpp-curl https://q.zash.se/76c690bcad19.txt

moparisthebest: Do it!

but it works, I get a private message when someone pushes :)

I switched most of my cronjobs/alerts to sendxmpp over sendmail years ago, but it has the downside of some messages are obnoxiously big for xmpp

I broke conversations once, now it has a character limit and messages get truncated

I'd like to add a bit of stuff so it uses ~/.config/sendxmpp.toml and is just a C version of sendxmpp

jjrh, did you try https://github.com/thepowersgang/mrustc

in theory it'd work for you but I never tried it (compiles rust to C)

well cross compiling to 32 bit linux should work too...

Nah they only show x86-64 targets - I didn't actually get libstrophe to cross compile, I got the same issue as I did trying with rust some issue with SSL. The reason C worked well is because I could just compile it on the 32bit machine without installing anything new.

jjrh, 'rustup target list' shows both i686-unknown-linux-gnu and i686-unknown-linux-musl

should be 1 invocation of 'rustup target add i686-unknown-linux-gnu' and then 'cargo build --target i686-unknown-linux-gnu'

Yeah I did exactly that - I wanted a static binary so I needed uh musl? ,got some issue with SSL - probably something to do with my system. I didn't feel like investigating so I tried with libstrophe and got a similar issue and just gave up and compiled it on the 32 bit server - doing it with C was less painful since the machine had GCC. I didn't want to install all the rust stuff

yea makes sense

I've only cross-compiled rust to 64-bit windows so far, it worked well though

Yeah I have done this before without issue but anything involving SSL tends to cause headaches

My C knowledge is basic at best though which doesn't help.

yea today I'll jump through a good number of hoops to write rust if it means avoiding C

I don't mind C really I just never end up in situations where I need it.

Last project I did in C was some embedded stuff on a TI arm M4F chip. Not a lot of work doing C these days really.

embedded rust seems to be going fairly well http://blog.japaric.io/ haven't gotten a chance to try it yet

The real issue is just about all the chip manufacturers support C and ASM and provide ALL their examples in that and all their build tools expect you're doing that so you have to jump through more hoops to use something else.

But i'm glad people are working on it.

Doing stuff with this TI board - their environment which is a fork of eclipse - as frustrating and crappy it is does have some really really nice and helpful stuff built in.