XMPP Service Operators - 2019-09-11


  1. tom

    matrix is just a bunch of marketing

  2. tom

    pretty much every reason they gave for why don't they just use XMPP has been nullified with recent xeps

  3. tom

    rather than fix painpoints and things we already have and sharpen our tools they reinvent everything from scratch and fix the same problems XMPP had to fix years ago.

  4. tom

    https://www.jwz.org/doc/cadt.html

  5. jrmu

    tom_: hm I got the same feeling too

  6. jrmu

    personally I prefer IRC over XMPP, I feel like IRC can do what XMPP does even simpler, but I think

  7. jrmu

    at least XMPP has been proven to work. Matrix to me seems like it's just a disaster waiting to happen

  8. jrmu

    The funny thing is matrix's main feature -- video chats -- is actually done using jitsi meet, IIRC, which I thought was using XMPP

  9. jrmu

    Otherwise I don't see any new compelling feature that matrix offers

  10. jrmu

    So I'm not delusional, matrix is just the latest hype/fad that will die a nasty death soon

  11. jrmu

    well, it'll probably survive, but will suffer some major setbacks in the next few years

  12. tom

    It's not a continuous improvement type thing or permaculture no

  13. jrmu

    ?

  14. mike

    if it mattered how good the protocols were, I wouldn't have lost 99% of my userbase to discord. UX is the only thing most people care about. if matrix looks nice, it'll be around forever.

  15. jrmu

    the xmpp clients have decent UI. But matrix is going to struggle with scalability, spam, and security because of bad design I think

  16. jrmu

    I'd also point out that matrix is fundamentally broken when it comes to the command line last I used it in 2018, and that matters for a lot of programmers

  17. jrmu

    There was only weechat-matrix at that time, and it didn't work well

  18. mike

    which xmpp clients have decent ui? it's a huge gap. Conversations is the only one I'd recommend to a non techy user right now. Need at least one client that's simple and usable and cross platform tbqh.

  19. jrmu

    Conversations and Monal/ChatSecure are decent

  20. jrmu

    Not great but decent

  21. jrmu

    I personally think Monal's UI is better than ChatSecure for iOS

  22. jrmu

    as for linux/windows, gajim was decent

  23. jrmu

    I agree that the clients could use some more polish

  24. mike

    gajim is a great diagnostic tool for me, it's technically great and very complete, but the UI is awful and inconsistent.

  25. mike

    dino is coming along nicely, hope it builds on windows one day.

  26. jrmu

    Personally I'm betting the farm on IRC

  27. jrmu

    I think IRC has a great protocol, just hideous UX, I just need to fix the UX

  28. mike

    still need a client, that's where the gap is.

  29. jrmu

    Hm, for me the existing XMPP clients are decent enough, I just need to submit some minor patches

  30. jrmu

    if I had to write a client from scratch, I'd prefer to focus on IRC though

  31. jrmu

    Right now Monal has issues with MUC support

  32. jrmu

    I hadn't heard of Dino before, what platforms are supported?

  33. mike

    Monal's muc support has been a bit lacking since day one but there's not much else you can use on iOS

  34. mike

    dino is only unix-ish stuff right now.

  35. mike

    https://dino.im/ I'm using it right at this moment

  36. jrmu

    Ah

  37. tom

    >‎mike‎: if it mattered how good the protocols were, I wouldn't have lost 99% of my userbase to discord. UX is the only thing most people care about. if matrix looks nice, it'll be around forever. that's what dino is for. The fad users who don't think about long term effects of their decisions. slap whatever questionable UI design choices into dino. that isolated that userbase from the protocols. The protocol does matter and in fact it matters the most because It's the foundation of all our infrastructure. even if you've got the most solid frontend possible it will fall over on a bad foundation. Also, not everyone is part of the dino userbase. Some people need professional communication tools and protocols, and if the protocol can't support that than were going to continue to have segreated networks

  38. tom

    the situation for email was similar in the 90s

  39. tom

    you had the AOL messenger built for kids and computer illiterates that was built upon email, but was siloed as in federation with other email servers turned off

  40. tom

    eventially the competition of various other small ISPs and orgranizations and the need for interoperable digital communication pervailed

  41. tom

    and AOL was forced to end up turning on federation

  42. jrmu

    I think the bad protocol design will result in some serious scalability and security issues down the line

  43. tom

    If you want to see the kind of immediate returns on userbase as discord does, all you have to do is spend the equiv amount of money bribing popular youtubers to shill XMPP. Just as the advertisement method was for Discord

  44. jrmu

    They won't feel it yet but they will have a very hard time crossing the threshold of getting mainstream

  45. tom

    but keep in mind the kind of userbase that will attract. you could bring about another eternal september

  46. tom

    My friends and colleages still use mumble despite the discord advertising. the majority of discord's userbase is gullible children. the rest is network effect

  47. jrmu

    Well that network effect is very important

  48. jrmu

    That's why I think it's still very important that the XMPP community make it a high priority to bridge seamlessly with the IRC community

  49. tom

    well then bribe some youtubers

  50. jrmu

    No, I think the focus is making sure XMPP has great bridges to IRC

  51. jrmu

    I don't have that kind of money =)

  52. tom

    neither does hammer & chisel (discord) they just loan it from venture firms and sell userdata to pay it back

  53. mike

    the network effect is absolutely everything and discord plays that game brilliantly. they're facebook jr.

  54. jrmu

    Yeah they have some very bad business practices

  55. jrmu

    But yes we can't ignore the network effect. We have to make sure that all the open protocols remain compatible

  56. tom

    well see what happens to them in the near feature. either they get bought out or go bankrupt

  57. jrmu

    remain connected*

  58. tom

    there might also be some new privacy laws coming soon which would throw a wrench in the selling of userdata bussiness model

  59. jrmu

    IMO one area we need to focus on is helping make these open protocols more profitable

  60. jrmu

    How can we let IRC/XMPP/Matrix operators profit without data mining?

  61. jrmu

    Otherwise without funding, our protocols will always remain niche

  62. tom

    there are more bussiness models than selling your user's data jrmu. communications is like a utility jrmu. Everyone needs it but not everyone has the expertise, time, or interest to provide it

  63. mike

    you only need massive funding if you're not decentralised. lots of small services, properly federated, can run on smaller models funded by the communities they serve.

  64. tom

    I still get paid to build and maintain email servers

  65. tom

    and there are lots of companies that provide email services

  66. mike

    "enable IM services for domain" should be a checkbox in a hosting plan.

  67. tom

    like this https://hosted.im/ and this https://www.cisco.com/c/en/us/solutions/hosted-collaboration-solution/index.html

  68. tom

    oh this https://nextcloud.com/

  69. jrmu

    tom_: Yeah. I was thinking of providing something like a one-stop infrastructure service for open source developers

  70. jrmu

    like a google drive + email + slack service for education/open source

  71. tom

    jrmu, like this https://www.linode.com/

  72. jrmu

    Linode only provides VPSes, I thought?

  73. jrmu

    Right now a lot of developers are using freenode and that's not very easy to use

  74. tom

    >loud Hosting for Developers High performance SSD Linux servers for all of your infrastructure needs.

  75. jrmu

    That requires self hosting though, it's too much work for most developers

  76. tom

    it's similar to what you were proposing yes? albeilt more like a hosted pbx than a virtual machine

  77. jrmu

    I was thinking of a github + gmail + dropbox + slack provider where it's just a few mouse clicks away, no programming needed

  78. jrmu

    The difference would be this provider would use open protocols, contribute the source code back for all software, and avoid data mining

  79. tom

    jrmu, oh. well then that exists already. look at nextcloud. it could use a bit of work but provides already most of what you describe with a plugin or two

  80. jrmu

    And probably charge a subscription fee

  81. jrmu

    Nextcloud has some serious issues I think, but that's sort of what I had in mind

  82. tom

    the JSXC javascript implementation of an xmpp client for in-browser use

  83. jrmu

    Nextcloud has the right idea but very bad execution

  84. jrmu

    eg the xmpp client can't connect with freenode or the rest of the IRC community, and the video streaming is sluggish and can't handle 1000s of simultaneous users

  85. tom

    I hate that they currently are screwing around with this nextcloud-talk webrtc thing when JSXC is already in the plugin repo, but you can see where it's coming from by reading the development tracker. people who don't know any better talking about "lets integreate rocketchat" type deal

  86. jrmu

    Ideally what we'd have is a video chat solution that works with IRC, so you could immediately chat with people already on IRC, plus add video on top

  87. jrmu

    Rocketchat looks good but also I struggled with it because it's very non-portable

  88. tom

    jrmu, i'd be interested in looking into the IRC situation. I've tried it myself by registering for transport service at the red hot chilli peppers XMPP server, but I'm not sure if their service is broken or what I could not get it to work properly

  89. jrmu

    I couldn't figure out how to install it on openbsd

  90. tom

    jrmu, I've been forced to use it and host it for a medium sized enterprise

  91. jrmu

    Did you like it?

  92. jrmu

    The UI looked great but administering it looked like a nightmare

  93. tom

    the whole thing is a pile of crap. there isn't even a native client, it's just a embedded web browser type deal, like discord

  94. jrmu

    ah =/

  95. tom

    and the backend is a nodejs blob

  96. jrmu

    That's what I had feared

  97. tom

    javascript and backend servers do not go well together. I

  98. jrmu

    Yep

  99. tom

    there's a reason javascript is isolated (or was supposed to be before nodejs) to web browsers

  100. tom

    but rocketchat offers nothing more than the baseline XMPP implementation, maybe even less. the only thing the sets it up is it's user interface, which is similar to the riot type deal

  101. jrmu

    Yeah Riot has a nice UI

  102. tom

    what you get with rocket chat is HTTP_UPLOAD, mucs, and that is about it

  103. tom

    (functionality wise)

  104. jrmu

    They also have jitsi meet in there somewhere I think?

  105. tom

    no

  106. jrmu

    ah

  107. tom

    it's just a slightly less bad slack

  108. jrmu

    If only we could have a Riot client for IRC =)

  109. jrmu

    then we'd be talking

  110. tom

    who's maintain it though

  111. tom

    *who'd

  112. jrmu

    Yeah that's why we need to solve that profit issue, hehe

  113. tom

    in order for that to work you'd have to alienate powerusers and professionals

  114. jrmu

    IMO, the best way to profit right now is to focus on providing an open source / education platform

  115. jrmu

    Since the main people who'd be willing to pay are open source enthusiasts

  116. tom

    and hobbyist developers aren't going to want to spend time maintaining something they would never want to use

  117. tom

    so this would have to come from a fund

  118. jrmu

    Provide a turn-key service for a slack-like alternative to freenode

  119. jrmu

    That open source developers could use, and charge a monthly subscription fee for storage/bandwidth

  120. tom

    https://dino.im/

  121. jrmu

    There's not a chance we can compete with freenode

  122. jrmu

    err

  123. jrmu

    There's not a chance we can compete with discord*

  124. jrmu

    but there's a chance we could compete with freenode

  125. tom

    I don't going back to irc is a good idea

  126. tom

    briding sure

  127. jrmu

    Dino has a nice UI

  128. jrmu

    This looks good, surprised I didn't notice it before

  129. tom

    well it's certainly subjective

  130. jrmu

    This looks better than gajim

  131. jrmu

    Any serious flaws to dino?

  132. tom

    looks like computer illiterate smartphone-inteface-on-computer cancer toy to me. but I understand the need for a client of such nature

  133. jrmu

    Yeah for normal people

  134. tom

    no

  135. jrmu

    I think the IRC protocol itself can be rescued, I'm trying to work on adding a lot of XMPP's features to IRC, like file upload, message sync

  136. jrmu

    There's no need to even change the protocol, it can be layered on with services and bots

  137. jrmu

    and bouncers

  138. tom

    using the term 'normal people' is misleading

  139. jrmu

    hehe

  140. jrmu

    no worries I understand what you mean. I'm using bitlbee+irssi

  141. jrmu

    I don't like GUIs either

  142. tom

    I like GUIs. got thing against them. I use Hexchat

  143. tom

    but i'm sure you can see theres a major difference in target audience between hexchat, and say...dino

  144. jrmu

    Yeah

  145. tom

    https://hexchat.github.io/img/screenshot-windows.png

  146. tom

    http://mamchenkov.net/wordpress/wp-content/uploads/2016/12/HexChat.png

  147. tom

    such as a major target audience difference in Gajim 0.16.X and Gajim 1.X

  148. jrmu

    For me, the focus needs to be more on the...discord type audience

  149. jrmu

    As mike brought up, the network effect is really important for networking protocols

  150. jrmu

    if we could get 1 million users on open protocols, it'd make a huge difference in terms of how attractive the protocols become

  151. jrmu

    Otherwise, people will say irc and xmpp are dead

  152. tom

    are you really sure about that? because we had that before. everybody and their grandma had a jabber endpoint/jid

  153. Licaon_Kter

    XMPP is used by billions....in some form...but yeah...dead...

  154. tom

    I feel like that kind of target is just going to mosey around to whatever has the most advetising of the time period, regardless of quality, standardization, security, privacy, usability, etc

  155. tom

    back when google exposed jabber federation

  156. tom

    albeit with a complete refusal to do so properly, IE TLS force-turned off

  157. tom

    what tf was up with that anyways?

  158. Licaon_Kter

    Gajim is constantly evolving, try the snapshots, it can be made more like Dino if you wish.

  159. jrmu

    The XMPP that is used by billions isn't the open kind, and it isn't what the open source community is using

  160. tom

    google of all companies surely could figure out how to get a X.509 certificate. and have the manpower to modify any server implementation to spy on their users more

  161. jrmu

    That's the big problem

  162. jrmu

    I can't connect to any of those networks using a foss client =(

  163. tom

    well unless you got silicon valley venture captialist money to blow up online advetising I think we are going to stay stuck in the AOL-email era. Best forget about trying to get the most userbase right now and focus on ourselves and our painpoints

  164. tom

    email was invented in the 1970s and it wasn't until the 1990s we got the proliferation and federation we have today

  165. tom

    right now the markets that need relibility, robustness, and standardization are the ones using XMPP

  166. tom

    https://www.ia.nato.int/niapc/Product/SDoT-Security-Gateway-version-6.0_711

  167. tom

    https://i56578-swl.blogspot.com/2018/12/xmpp-over-hf-radio-using-stanag-5066.html

  168. tom

    https://www.isode.com/markets/military-xmpp.html

  169. jrmu

    hm, so the US military

  170. tom

    not just the US military, NATO's tactical coms

  171. tom

    tactical field communication

  172. jrmu

    what do you think about the open slack alternative for freenode?

  173. jrmu

    IMO that could be quite a big market

  174. tom

    also, on a bit of an unrelated note, something that may be useful to you jjrh https://www.isode.com/markets/military-xmpp.html#irc

  175. jrmu

    they are a market that cares about security and data ownership

  176. tom

    isode has an IRC gateway

  177. jrmu

    ah interesting

  178. jrmu

    although Isode IIRC, is proprietary so won't be of use to me

  179. tom

    maybe that specific isode server implementation

  180. tom

    but isode helps develop a lot of spec and has open source reference clients and libraires

  181. jrmu

    ah

  182. tom

    for one the pubsub forums XEPs were mostly from isode's contributions

  183. tom

    which the military uses for MEDIVAC

  184. jrmu

    Ah interesting

  185. tom

    https://xmpp.org/extensions/xep-0346.html

  186. tom

    if you look at the example stanza it shows isode

  187. jrmu

    ah

  188. tom

    https://www.isode.com/whitepapers/military-forms-using-xmpp.html

  189. Allie

    DoD XMPP is so huge its hard to use, though

  190. Allie

    plus the only client is Transverse

  191. Allie

    I need to check out that other branch of gajim.

  192. Allie

    although I might already be using 1.x...

  193. muppeth

    Meh... no need to cry and complain. Everything we need is already here. Just needs some work here and there. Rather then flamewar (been there done that) everyone shoukd just check where they csn contribute and just go for it.

  194. muppeth

    Conversejs easily by pretty much anyone can be built into rriot like chat even with jirsi as a plugin.

  195. muppeth

    I managed to rebuild my little community after we left matrix pretty much from zero, and i was suprised when people started using xmpp for the first time and they were shock as to how fast, light and easy it is to use it

  196. muppeth

    And there is movim which is great webclient with even social network built in

  197. muppeth

    There is really less and less to complain about. Sure matrix has riot, developed by for profit company so much easier imo, but thats lretty much it. No other client has viop and barely any e2ee

  198. muppeth

    And since they have a for profit company that understands that to succed you need to hype it up (they are in the startup world) so they try to patch things like using jitsi to solve voip or now some library that is pretty much a e2ee bridge beteen your client and server etc

  199. muppeth

    Sounds nice but in long run lots of duct tape

  200. muppeth

    > They have clients that offers features and gives what people wants today. XMPP don't Meh.... its a hype thing. Your client does everything and more. No need to be ashamed of it.

  201. Martin

    Am I missing messages or is muppeth having a long monologue?

  202. muppeth

    You even have reactions!!!!

  203. muppeth

    Martin: was in the metro

  204. MattJ

    Martin, shush, don't stop him, it's great!

  205. Martin

    https://files.mdosch.de:64183/upload/7dBm13VkPv1cctQP/Screenshot_20190911-174958_Conversations.png

  206. muppeth

    So yeah typical muppeth rant

  207. Martin

    Ok 😃

  208. muppeth

    MattJ: hahahaha... just got to my stop so that would be it for now hahababa

  209. Martin

    I'm in the metro too but I don't feel like ranting. 😂

  210. MattJ

    Martin, wow. Which city's metro doesn't make you feel like ranting? :)

  211. Martin

    Munich

  212. Martin

    It's crowded now but it's rush hour so I am used to it. 😃

  213. Allie

    oh and re: discord, ughhhhh... it just irritates me. I set up an account for the first time the other day to get a transport working with spectrum. I don't know why people love the interface because it's cluttered and horrible to me.

  214. Allie

    only got one person to talk to on it, but they refuse to use anything else, hence the transport

  215. Martin

    Ha, Allie takes over ranting. 😃

  216. Allie

    👵📢☁️

  217. Martin

    Ha, I really like Simpsons references. 😃

  218. Allie

    hehehe

  219. Allie

    I'm not old, but it's interesting to be old enough to watch people just keep reinventing the wheel

  220. muppeth

    > Ha, Allie takes over ranting. 😃 Hahahaa should make shifts

  221. Allie

    lol

  222. muppeth

    But seriously.should turn that frustration into positive thing. Just like others do. You know. Gajim doesnt have ugly ui, it a sophisticated tool for specialists. We dont lack good clients for iOS, we are just careful with adding new features to provide the best and the most robust and well tested and handcrafted user experience.

  223. Licaon_Kter

    muppeth: good spin

  224. Allie

    If I'm ever in a place where my programming skills are decent enough I may help out with some existing projects. But I also gotta have time which is in short supply these days 😔

  225. muppeth

    Allie: thats just lame excuse everyone uses. Last week i reskinned conversejs l. Did not take any skills (i'm lame dev) and if you lokm around there is plenty that could be done. Theming, helping with design, writting posts, promoting, troubleshooting omemo on all possible cients, etc

  226. Licaon_Kter

    muppeth: link reskin?

  227. muppeth

    Licaon_Kter: webchat.disroot.org the code will be published this week as we're migrsting to new git instance

  228. Licaon_Kter

    muppeth: 👍

  229. muppeth

    https://disroot.org/upload/shWngElb9IiCrfho/VhfryVMiRx-1vnFxI7w7Cg.jpg

  230. muppeth

    https://disroot.org/upload/654KcwQao5Nc-x-U/-QdUlPNwS4eprn22yixCzw.jpg

  231. Allie

    muppeth: I feel like there's a difference in working with web stuff vs local applications, though. I can do web stuff fine. Plus I'm still trying to wrap my head around XMPP protocol wise. I've been working on a bot for my own server to try and get a better handle on things.

  232. Allie

    that is a pretty nice layout

  233. muppeth

    Allie: but all the others are successful because they dont care about local stuff. Discord, mattermost, rocketchat and riot is just electron

  234. Allie

    I know. I want nice local apps to use myself, though 🙂

  235. muppeth

    And since we have three webchats already there and web stuff is very easy entry point for everyone, why no jump in and improve on it so we have aweome webchat solution for xmpp which solves the issue of multiplatform

  236. Licaon_Kter

    muppeth: great, too bad that concorde theme is kinda abanadoned/broken

  237. muppeth

    And then the non-lame gurus can work on native client improvements

  238. muppeth

    Allie: i use conversations on mobile and dino on desktop and i'm very satisfied with it

  239. Allie

    I still gotta get a web interface to my instance. So maybe I'll play with that then.

  240. muppeth

    Dont need more atm. I have gajim but as someone said it for special admin stuff

  241. Allie

    I have Dino installed but it's a little too simplistic for me. I use Conversations on Android and Gajim on my desktop

  242. muppeth

    > muppeth: great, too bad that concorde theme is kinda abanadoned/broken Initially i was hacking on the main theme but now i want to move it as a proper seperate theme and push it upstream

  243. muppeth

    Allie: what non simplistic features you need for everyday use for text messeging system

  244. Licaon_Kter

    muppeth: better update and fix those CVEs https://github.com/conversejs/converse.js/releases/tag/v5.0.2

  245. muppeth

    Licaon_Kter: thanks will do.

  246. Licaon_Kter

    And Dino, same issues

  247. Allie

    muppeth: if I did nothing but msg people, Dino is probably fine. but running my own server (albeit not with many users) I like having service discovery, being able to manage transports, ad-hoc commands, etc.

  248. muppeth

    Allie: i ise those rarely so every now and then i start gajim just to do those specific things but most of the time dino is just fine

  249. muppeth

    Allie: which tranports are you running?

  250. Allie

    right now spectrum for discord and biboumi for IRC, although might be adding more

  251. muppeth

    Allie: can discordtransport be configured by user (like bridging rooms/joiing rooms etc) or you need admin for that?

  252. Allie

    I think with spectrum it works like pretty much any regular transport. You can msg individuals, join rooms, etc., but not bridge to existing MUCs I don't think. Unfortunately it's limited because stupid discord won't let you login unless you login to the official web app from my server's IP first. So that really limits its usefulness unless spectrum adds the ability to handle captchas

  253. rom1dep

    > Gajim doesnt have ugly ui, it a sophisticated tool for specialists. I have few happy users on gajim, they aren't exactly young or advanced on the computer literacy spectrum.

  254. Allie

    Gajim usually works well for me except for little annoying issues sometimes

  255. Allie

    or it'll be like "Welp... 🤷‍♀️" and crash

  256. rom1dep

    What really kills it (and Dino) is imho the inaptitude of Gtk3 to behave and look well on anything but a gnome env. But that's another topic

  257. Allie

    yeah