XSF Discussion - 2018-03-19

  1. Dave Cridland

    For what it's worth, we copied the password text straight from the original, and the security considerations are really a first cut, but I think passwords are fine here, it's just that they're not real security.

  2. SaltyBones

    Passwords are not real security? :)

  3. jonasw

    Ge0rG’s implementing at-least-once semantics :)

  4. Ge0rG

    SaltyBones: MUC passwords aren't

  5. SaltyBones

    Ge0rG, why not? I have no clue how those work...

  6. jonasw


  7. jonasw

    seen by your server

  8. Zash

    I haven't really seen many password-protected MUCs

  9. jonasw

    yeah, members-only feels more effective and useful anyways

  10. Zash

    A one-time-use password that grats membership would have been nice. Probably could hack it serverside, but are clients going to keep sending the password?

  11. jonasw

    Zash, that’s the MUC invitation thing I asked for a few months ago when the whole PARS stuff was going on

  12. Zash

    How often are people changing their bookmarks from two clients at once?

  13. Zash

    I had the same question about MAM settings IIRC

  14. MattJ

    As I said on the list, the same applies to just about every operation we have

  15. Zash

    -xep 0395

  16. Bunneh

    Zash: Atomically Compare-And-Publish PubSub Items (Standards Track, Experimental, 2017-11-29) See: https://xmpp.org/extensions/xep-0395.html

  17. Zash

    Oh dear publish-options

  18. Zash

    MattJ: Oh, you said that now? I was still reading the one before yours :)

  19. flow

    xep395 was written with things like groupchat subject nodes in mind, FWIW

  20. flow

    i.e. items that could be potentially modified by multiple entities

  21. pep.

    Fun fact, related to the groupchat terminology thread, https://docs.mattermost.com/help/getting-started/organizing-conversations.html, Mattermost has "private channels" _and_ "group messages", that are literally the same thing from what I understand, apart that group messages are limited to 7 members.

  22. pep.


  23. pep.

    "Group message channels are useful for fluid/ad-hoc conversations among users. Private channels are more useful when there's a concrete topic to discuss and you want to preserve the message history, or at least have an easy way to collect and refer to it later. You can also add more people to an existing private channel conversation and make it public later." From a mattermost person.

  24. jonasw

    so for group messages there’s no history?

  25. pep.


  26. jonasw


  27. pep.

    I don't like this split personally

  28. pep.

    I want history, everywhere, all the time

  29. jonasw

    that solves the "what about history in an ad-hoc group discussion?" issue clearly :D

  30. jonasw

    pep., was discussed at summit, it’s not trivial

  31. pep.

    how so

  32. jonasw

    for example, group conversation between Alice, Bob and Carol. At some point, Bob and Carol talk about Dianne, maybe planning an Intervention for her weird behaviour regarding hats. Then the discussion evolves and they need to invite Dianne to discuss some plans next week.

  33. jonasw

    if Dianne has access to the history, that’s bad

  34. jonasw

    if Alice, Bob and Carol need to do UI dances to prevent her from doing so, that’s also bad.

  35. pep.

    they create another channel and move on?

  36. Kev

    I like Slack's approach here, personally.

  37. jonasw

    Kev, how does slack handle this?

  38. Kev

    "Would you like to preserve history? If you do, Dianne will be able to see it. If you don't, it will be removed for everyone"

  39. jonasw

    (also, I have no idea how I came up with the hats thing and now I kinda want to know what Dianne does with hats.)

  40. jonasw

    Kev, when inviting a new person or when first creating the channel?

  41. Kev

    It's not perfect, obviously, but it's functional enough and not surprising.

  42. Kev

    When inviting a new person to a private channel.

  43. jonasw

    that’s neat

  44. Zash

    It's possible to restrict history to only those present to see it

  45. jonasw

    Zash, with MUC, that’s not great either, because you drop out temporarily during connectivity issues.

  46. Zash

    jonasw: Well, you can base it on affiliation, not presence.

  47. jonasw

    Zash, right

  48. Kev

    Most people have no affiliation

  49. Kev

    But yes.

  50. Zash

    Depends on the room

  51. jonasw

    Kev, in private channels, you’d typically need member affiliation

  52. jonasw

    because you want them to be members-only

  53. Zash

    If it's for private team chat then they probably do

  54. jonasw

    so that makes sense.

  55. pep.

    I usually set affiliations on my channels

  56. Kev

    It's not hugely straightforward to limit per-message history based on affilation at that time, though.

  57. pep.

    But that could be automated anyway

  58. Kev

    Possible, obviously, but not hugely straightforward.

  59. jonasw

    Kev, implementation-wise?

  60. Kev


  61. pep.

    jonasw> if Dianne has access to the history, that’s bad jonasw> if Alice, Bob and Carol need to do UI dances to prevent her from doing so, that’s also bad. pep.> they create another channel and move on? jonasw ^, probably what's happening internally in mattermost already

  62. pep.

    When inviting a new person

  63. Zash

    I imagine it gets complicated if you want newly invited persons to see some history from before they were invited, but not all

  64. Zash

    Where on the metaphorical scale from 'actual private room' to 'written notes on a public board' scale do you wanna be?

  65. Zash

    "bulletin board" was the term

  66. pep.

    Everybody's got different use cases, so trying to please everyone is hard

  67. pep.

    I think we should just give up already

  68. jonasw

    rm -rf xmpp.org

  69. pep.

    git push

  70. Zash

    The life of a potato-farming hermit is the ultimate solution

  71. pep.

    Is there any "goal" defined by the XSF as to what they're trying to achieve. What public they're targetting

  72. edhelas

    ln -s xmpp.org matrix.org

  73. jonasw

    edhelas, :(

  74. jonasw

    pep., no

  75. Zash

    pep.: XEP-hearding

  76. pep.

    If not I thought that should be on the list

  77. jonasw

    yeah, that

  78. Zash


  79. Zash


  80. pep.

    I think.

  81. jonasw

    the XSF isn’t targeting any public. the folks authoring XEPs and developing software are.

  82. pep.

    Yeah, that's a bit too broad

  83. jonasw

    the subgroup of that which is interested in making a good IM system should probably come up with something though.

  84. Zash

    I do think it'd be nice if Council or Board wrote some kind of vision statement.

  85. edhelas

    the issue is that lots of app are also using XMPP for non-IM stuff

  86. pep.

    Then I can just read the statement and say "Ok I want in", or "It's not for me", and not try hard to move it my way when it's never going to go where I want

  87. edhelas

    I fully understand that it's the core thing but sometime it's a bit too focused

  88. edhelas


  89. edhelas

    also with the Markdown/XHTML-IM thing

  90. jonasw

    I wish we had a way to link to/show in the XEP list different versions of the same XEP easily.

  91. jonasw

    in the context of the compliance suites

  92. vanitasvitae

    that would be great indeed

  93. jonasw

    it would be great to have a current version which is shown by default when accessing the link

  94. jonasw

    and a staging version where development of the new release takes place

  95. vanitasvitae

    Its frustrating to find out what changed from one version to another without using git

  96. vanitasvitae

    also the attic is often missing versions which complicates the situation even more

  97. jonasw

    yeah, I’m sorry

  98. jonasw

    attic is a manual process

  99. jonasw

    gotta run, see you later

  100. vanitasvitae

    it shouldn't be though

  101. vanitasvitae


  102. Zash

    "historical reasons"

  103. Kev

    We used to have the difftool, but history wasn't kind to it.

  104. Zash

    I do have a half-working markdown based comparison tool

  105. Zash

    Just needs motivation and time

  106. pep.

    There's no standard place for a server to advertise privacy policy, EULA, etc., from what I understand. It would be good to have one

  107. jonasw


  108. pep.

    Would it make sense to incorporate that in an existing XEP? A New one?

  109. jonasw

    pep., cp xep-template.xml inbox/eula.xml && $EDITOR inbox/eula.xml

  110. pep.


  111. pep.

    Something à la {xep contact}?

  112. Bunneh

    pep.: Multiple matches: Contact Addresses for XMPP Services https://xmpp.org/extensions/xep-0157.html Metacontacts https://xmpp.org/extensions/xep-0209.html

  113. pep.


  114. Zash

    Something in IBR(2?) probably

  115. pep.

    Would make sense. I guess you can already do that with forms? Or just redirect to a web page for the whole thing, but I do prefer the "in-band" part of IBR.

  116. pep.

    Though admittedly, EULA would most likely be an http link

  117. jonasw

    it would be good to hvae the common things as structured data so that clients can display a summary

  118. jonasw

    like: [ ] encrypted storage data automatically deleted after [ ] days …

  119. Zash

    It would be good if this could be negotiated

  120. Zash

    As in, that the client can say "I understand these things"

  121. Zash

    Or you end up like if you try to use extended registration forms now, with nothing working and no way to indicate why

  122. jonasw

    Zash, yeah sure

  123. pep.

    yeah, having data forms support for IBR in clients would help

  124. moparisthebest

    "XEP-XXXX Standardized list of things server admins can lie about" ?

  125. moparisthebest

    keeps logs, encrypted storage, we promise to try SUPER HARD not to look at your data

  126. pep.

    moparisthebest, better than non standardized list of things that server admins can lie about? :)

  127. jonasw

    moparisthebest, sure, they can lie about, but if they make false statements they’re liable for that

  128. Zash

    Can't just go on the internet and tell lies

  129. jonasw

    but statements are required as per EU-GDPR

  130. moparisthebest

    just seems super useless

  131. jonasw

    so better have some standardised way to make it easy for everyone

  132. moparisthebest

    oh who would have guessed govt regulation would turn out to be useless :)

  133. Zash


  134. pep.

    moparisthebest, you trust or you don't trust statements of your server admin, that's your issue

  135. pep.

    But let them tell their lies

  136. moparisthebest

    pep., I'd rather avoid the false sense of security and foster a healthy distrust of server admins

  137. Zash

    Civilized society needs its privacy statements and agreements.

  138. pep.

    moparisthebest, I want my users to be aware of how I operate

  139. pep.

    Otherwise they don't get to use my service

  140. moparisthebest

    meh I don't think it does Zash , I'd prefer to just solve the problem with technology

  141. moparisthebest

    otherwise why even bother with things like TLS ? just ask intermediaries to promise not to look at your traffic?

  142. Zash

    You know what they say about technical solutions to social problems?

  143. Zash

    Why bother with locks. It's pretty easy to pick them anyways.

  144. Zash

    Locks aren't entirely a techical thing. It's part social signal, part technical.

  145. Zash

    And then things like the legaly system to deal with people who break it. And insurance to reduce the damages.

  146. Zash

    Main reason why TLS needs to basically be perfect is that those civilization things don't scale to Internet-sized groups

  147. moparisthebest

    I guess the scaling thing is the concern, if I run a server for friends/family, we don't need any statements/agreements, and if I run a server for the public, statements/agreements are useless because they are unenforceable anyway, and they don't trust me

  148. Zash

    I do wonder how GDPR relates to self-/small-group-of-friends hosting

  149. jonasw

    Zash, tricky, I’m not sure if third parties can hold you liable.

  150. Zash

    moparisthebest: Myeah, we haven't completely figured out how society works with Internet-scale communications yet.

  151. jonasw

    moparisthebest, let’s talk about unenforceable again when the privacy regulator comes knocking on your door because there’s evidence that your public service stored my messages without my consent :)

  152. jonasw

    (of course, you can point at your records and say "but you enabled MAM" and then I’m like "wtf are you talking about" and then we figure out that my client did that behind my back and now nobody knows who the f* is actually liable for that)

  153. Zash

    We can't have 100% perfect enforcement. But most people are mostly honest most of the time, so usually things work out fine.

  154. jonasw

    (alternatively, you figure out that prosody has been enabling MAM without explicit consent since forever and you’re screwed because you didn’t properly vet the software you’re using)

  155. Zash


  156. jonasw

    pep., speaking of things, one probably also needs versioning for the privacy policy when we’re doing that

  157. jonasw

    Zash, that’s why I said "you’re screwed" and not "the prosody folks are screwed" :)

  158. jonasw

    pep., so that servers can keep track of the version of the policy accepted by the user and re-ask them when things ch ange

  159. Zash

    The balancing act between consent of the user, intent of the admin, UX ...

  160. moparisthebest

    also how do they expect to enforce this over the 90% of internet they have 0 control over?

  161. moparisthebest

    I'm not even sure if, being a US citizen, this applies to me if my server is in germany...

  162. Zash

    Yeah, how do these things work with federation?

  163. jonasw

    moparisthebest, it obviously only affects entities offering services in the EU.

  164. jonasw

    moparisthebest, doesn’t matter, it applies to you if you have EU customers.

  165. jonasw

    (or users)

  166. moparisthebest

    jonasw, citizens of EU, servers of EU, or users in EU

  167. moparisthebest

    ok, so users in EU, and if I don't comply, how do they expect to force me to?

  168. jonasw

    I have no idea

  169. jonasw

    but users may prefer EU services over US services for this reason.

  170. moparisthebest

    if I visit the EU one day they arrest me? :P

  171. Zash

    Extradition agreements are fun.

  172. moparisthebest

    I'll just never come to EU then I guess

  173. jonasw

    just like I’ll never come to the US :-)

  174. jonasw

    or russia for that matter.

  175. moparisthebest

    Zash, I can't imagine those would apply, that'd be kind of crazy

  176. moparisthebest

    oops an EU user accessed the server you run in your house in USA, we are gonna send you to EU prison now...

  177. jonasw

    moparisthebest, EU is taking data protection rather seriously nowadays, I’m not sure what the punishments are though.

  178. Zash

    moparisthebest: Uh, I'd rather imagine that the EU isn't insane like that.

  179. Zash

    Glob help you if you share some copyrighted files tho

  180. jonasw

    having the GDPR stuff pre-IBR via stream feature magic would be great, it could be incorporated into xmpp.net

  181. jonasw

    if anybody dares to touch the code that is.

  182. moparisthebest

    so speaking of what Zash said, bob.com promises no logs, but bob@bob.com messages tom@tom.com and tom.com logs *everything*

  183. moparisthebest

    how does this work?

  184. jonasw

    moparisthebest, no idea.

  185. moparisthebest

    did the administrator of bob.com just break a law

  186. jonasw

    probably not

  187. Zash

    moparisthebest: As I said, clarity on how these things relate to non-commercial self-hosting would be good.

  188. moparisthebest

    ah that gives you a warm and fuzzy feeling

  189. jonasw

    can the XSF sponsor a lawyer to figure out those use-cases?

  190. moparisthebest

    I'm probably not going to jail for running a public xmpp server :)

  191. Zash

    jonasw: and/or the IETF?

  192. jonasw

    Zash, maybe

  193. jonasw

    should put that on boards agenda

  194. moparisthebest

    everyone run their own xmpp server! you might not even go to jail for it in the EU! :)

  195. Zash

    Operators of email and other federated things are probably interested as well

  196. moparisthebest

    yea the answer would probably be identical for email

  197. Zash

    moparisthebest: It depends!

  198. Zash

    Email is store-and-forward.

  199. Zash

    IM is ... not?

  200. Zash


  201. Zash

    Is now, with MAM :/

  202. Zash

    Data at rest is considered differently from data in flight.

  203. Zash

    Sometimes? IANAL.

  204. moparisthebest

    well smacks is kinda store and forward, so is offline messaging, muc backlog thing

  205. moparisthebest

    I think it's safe to say 99% of xmpp messages today are store and forward, or at least you can't tell when sending them so you have to treat them as such?

  206. Zash

    Technically, it's all store and forward

  207. Zash

    Down to the packet routing

  208. moparisthebest


  209. moparisthebest

    seems odd to treat them differently

  210. Zash

    Legally ... hrrrr

  211. moparisthebest

    I mean, this is what happens when you get politicians dictating technology, nothing but bad things

  212. Zash

    > A series of tubes

  213. pep.

    jonasw, re versioning, yes that'd be cool

  214. pep.

    Also keep track of acks?

  215. Zash

    Re that, you could check how it's done in ACME

  216. Zash

    IIRC you reply with a hash of the legalstuff.pdf

  217. pep.

    Zash, I'll have a look thanks

  218. jonasw

    gonna send board@ an email

  219. jonasw


  220. pep.

    http://logs.xmpp.org/xsf/ not available on https?

  221. moparisthebest

    I guess the disconnect makes sense, I'm a programmer, I like technical solutions, politicians are lawyers, they like legal solutions :P

  222. pep.

    domain not in SANs

  223. moparisthebest

    and of course only 1 is the correct way... :)

  224. pep.

    Who do I need to ping to add it?

  225. jonasw

    at least Romeo Montague and Juliet Capulet are actually useful examples in this case (regarding Article 9, "Processing of […] data concerning a natural person’s sex life […] shall be prohibited.")

  226. jonasw

    pep., probably someone from iteam. intosi maybe.

  227. Ge0rG

    jonasw: I'm probably half in jail already for running a public xmpp server in the EU

  228. jonasw

    Ge0rG, \o/

  229. jonasw

    you’ll be interested in next board meeting then ;-)

  230. pep.

    I'll watch closely as well

  231. Ge0rG

    Regarding that Sex life thing, now I'm supposed to check all http upload files and immediately delete dick pics?

  232. jonasw

    Ge0rG, no, you just need consent.

  233. jonasw

    Article 9 (2) is a long list of exceptions to teh general "shall be prohibited", one of which is "the data subject has given explicit consent to the processing of those personal data for one or more specified purposes, except where Union or Member State law provide that the prohibition referred to in paragraph 1 may not be lifted by the data subject;"

  234. Ge0rG

    jonasw: I've asked a GDPR specialist recently, and he ran away crying after seeing my server deployment

  235. jonasw


  236. jonasw

    I bet.

  237. pep.


  238. moparisthebest

    explicit consent like "By continuing to use this service, you explicitly consent to..." ?

  239. jonasw


  240. pep.

    I'm going to https://www.cambridgenetwork.co.uk/events/gdpr-itgovernance-march2018/ this week. Let's see if I gather anything interesting

  241. Ge0rG

    I need to convince my boss that writing a policy for yax.im will be a nice exercise for our younger colleagues

  242. jonasw

    pep., neat.

  243. jonasw

    I’ll dump the things I threw at board here so you can mention it there, pep.: There was some discussion in xsf@ today (actually, is right now). Some of the points which were mentioned: General question: Are IM messages to be considered "personal data revealing racial or ethnic origin, political opinions, religious or philosophical beliefs, or trade union membership, and the processing of genetic data, biometric data for the purpose of uniquely identifying a natural person, data concerning health or data concerning a natural person’s sex life or sexual orientation" in general (article 9)? (I suspect so, IANAL.) If not, I think most of the following points are moot-ish. Situation A: romeo@montague.lit talks to juliet@capulet.lit. While romeo is aware of the privacy policy of montague.lit (he acknowledged it when registering), he is not aware of the privacy policy of capulet.lit. capulet.lit decided to store all IM messages forever, which is probably(? IANAL) something they need explicit consent for even from other domains. Situation B: capulet.lit has a MAM service, but it is opt-in to ensure consent from the users. (Suppose here that we have protocol to actually show a privacy policy when users opt-in to MAM.) juliet uses a client which turns on MAM by default. Who is liable when juliet complains that capulet.lit is storing messages? And how to avoid this? Situation C: coven@chat.shakespeare.lit is a private MUC with MUC MAM enabled. Is this covered by Article 9 (2) (e) ("processing relates to personal data which are manifestly made public by the data subject;")? I suspect not, and then we’d need ways to convey the terms of archival and to express consent when joining such a MUC. Is this situation different if the MUC is public? I suspect that this will have to do a lot with how the UI presents it.

  244. Ge0rG

    moparisthebest: like with the EU cookie warning...

  245. SamWhited

    I've been working on GDPR compliance stuff for weeks now… I'm starting to get chills whenever someone mentions it. Opened this room and thought I'd accidentally started work chat instead.

  246. jonasw

    SamWhited, :)

  247. pep.

    SamWhited, :D

  248. MattJ

    Does anyone actually have the answers to these questions?

  249. Ge0rG

    MattJ: do you consider "pay a €100k compliance violation fee and stop the offending behavior" a valid answer?

  250. moparisthebest

    it seems to me the law was specifically crafted to target walled gardens, not federated systems, and it basically makes it impossible to run federated systems...

  251. jonasw

    MattJ, you’ll find out thursday! (board@xmpp.org is the right adress to dump board agenda at, isn’t it?)

  252. MattJ

    jonasw, I don't know... I haven't received any email, so I don't know where it went to

  253. moparisthebest

    which, politicians ignoring xmpp, fair, but they ignored email too? surely they know about email

  254. jonasw

    moparisthebest, they might not know how email works

  255. moparisthebest

    true, if they think of email as gmail...

  256. pep.

    jonasw, their technical team *might*

  257. jonasw

    MattJ, I can’t add an agendum to the board trello, can you do that for me when I forward you my email?

  258. MattJ

    jonasw, shall do

  259. MattJ

    Ge0rG, I mean, I understand a lot of people are making money from GDPR consulting, but has anyone to date received a €100k compliance violation fee?

  260. moparisthebest

    what if everyone just pulls what I pull on my IRC server, put a statement like "Due to GDPR, citizens of EU are forbidden from using this server" up

  261. moparisthebest

    and then just not enforce it in practice?

  262. Ge0rG

    MattJ: no, because the GDPR isn't in effect yet

  263. MattJ


  264. jonasw

    MattJ, enforcement afaik only starts on may 25th

  265. MattJ

    So nobody knows how the legislation will be interpreted by the courts

  266. MattJ

    I find it unlikely that they would conclude that a non-commercial XMPP service that does not make any money would be forced to pay a €100k fine because they stored someone's groupchat message in an archive

  267. Ge0rG

    MattJ: the first step will be for the data protection offices to ask companies for their policy documents

  268. Ge0rG

    MattJ: unlikely isn't impossible

  269. moparisthebest

    MattJ, and what about a commercial xmpp service that charges $2 per month or something

  270. MattJ

    No, nothing is impossible

  271. MattJ

    moparisthebest, fines are usually proportional to company revenue

  272. Ge0rG

    MattJ: I'm not sure if you would bet your private possessions on that low probability

  273. MattJ

    IANAL, I'm not telling anyone they shouldn't worry about GDPR, I'm just questioning how much you can take a lawyers word today about whether e.g. storing chatroom messages in an archive is legal or not

  274. Ge0rG

    MattJ: the lawyers don't know either, so they predict the worst case

  275. MattJ

    of course

  276. MattJ

    but we already know the worst case, without paying the lawyers anything

  277. MattJ

    If you have to ask, you probably shouldn't be storing it if you want to be 100% safe

  278. jonasw

    I tried to ask very fundamental questions though. questions to which we should already have an answer before the first xmpp service is sued.

  279. jonasw

    like the federation thing

  280. jonasw

    and with answer I mean a technical way to achieve what’s needed to comply.

  281. jonasw

    like what we were discussing earlier with the potential privacy policy XEP

  282. Ge0rG

    There was a nice court ruling in Germany recently, regarding WhatsApp. A WhatsApp User requires written consent from all contacts to put their phone number into the cloud.

  283. Ge0rG

    That sounds like we need consent from each MUC participant

  284. jonasw

    oddly, people haven’t stopped using whatsapp :(

  285. Ge0rG

    jonasw: yes, probably out of ignorance.

  286. jonasw

    Ge0rG, for public MUCs probably not due to Article 9 (2) (e), I guess.

  287. jonasw

    Ge0rG, no, probably because nobody sues their friends and relatives or people with whom they do business over a phone number upload.

  288. jonasw

    I wish $unlovedRelative was using whatsapp. that’s a perfect way to break off contact

  289. jonasw

    I wish $unlovedRelative was using whatsapp. that’s a perfect way to break off contact *and* get some money out ouf it :>

  290. tim@boese-ban.de

    jonasw, but only if the relative is unloved by the whole family :-)

  291. jonasw

    tim@boese-ban.de, true :)

  292. Ge0rG

    jonasw: you will have a hard time getting money out. The best thing you can hope for is a fine, and you need to tell the relative in advance that you don't consent with sharing of your information with third parties

  293. jonasw

    Ge0rG, do I? isn’t it default that I don’t consent?

  294. Ge0rG

    jonasw: maybe, but you need a willful violation to provoke a fine

  295. jonasw


  296. Ge0rG

    jonasw: I see what you did here.

  297. jonasw

    fine :)

  298. pep.

    Ge0rG, "willful violation"?

  299. Ge0rG


  300. Ge0rG

    pep.: knowing that your behavior is illegal and still continuing. IANAL

  301. pep.

    I see

  302. pep.

    "But but, I didn't know"

  303. moparisthebest

    wait, are you saying ignorance of the law IS an excuse?

  304. Ge0rG

    moparisthebest: only regarding the level of fines you expose yourself to.

  305. Ge0rG

    moparisthebest: if you are not a commercial entity, you are not required to understand and implement all of the GDPR requirements.

  306. Ge0rG


  307. Ge0rG

    At least it is rather improbable that you will be sued for uploading your grandma's cookie receipt to AWS

  308. jonasw

    how about your grandmas erotic friend fictions?

  309. Ge0rG

    jonasw: it depends whether those are real or imaginary friends. With fiction you are subject to copyright, where the civil liability depends on the number of potential readers, with non fiction you are subject to GDPR, and you know the fines there.

  310. jonasw

    well your grandma would be a real person and thus at least one subject in that fiction story would be real.

  311. jonasw

    (at least that’s the limited understanding I got on erotic friend fictions)

  312. Ge0rG

    jonasw: I would argue that fiction doesn't count as sensitive PII, but probably only if it's clearly labeled as fiction.

  313. moparisthebest

    also a possibly upcoming EU law would require disabling e2e and scanning/filtering all stanzas sent https://blog.github.com/2018-03-14-eu-proposal-upload-filters-code/ :'(

  314. Ge0rG

    I'd be the first one to deploy an OMEMO block filter...

  315. SamWhited

    *sigh* I can never decide which I hate more, how much we under-regulate the tech industry, or how much Europe overregulates it.

  316. SamWhited

    ("we" being the U.S.)

  317. pep.

    Depends on the regulations?

  318. moparisthebest

    I'm in a different camp, I think all the regulations are bad

  319. pep.

    jonasw, you mentioned "encryption" when talking about server policies. Disk encryption? Protecting against the hosting provider? They have do have full control over the equipment, I guess paranoïa can go pretty far, how would you deploy that?

  320. jonasw

    pep., I have no idea. I was desperately trying to think of a second thing :)

  321. pep.


  322. andrey.g

    moparisthebest‎, I'm wondering, how the world would look, if not only artificial regulations but also the natural one "only the fittest will survive" would disappear...

  323. jonasw


  324. moparisthebest

    andrey.g, not really sure what you mean, but I'm fine with natural regulations, the artificial ones are the problem

  325. moparisthebest

    also wouldn't call them 'natural regulations' but meh :)

  326. andrey.g

    moparisthebest‎, so we have different meanings of "all" regulations.

  327. moparisthebest

    like I said I wouldn't call like natural laws regulations

  328. pep.

    jonasw, a bit more thinking tells me I can't be technically sure the hosting provider doesn't have access to my system. Best is to be the provider.. I guess that works for small deployments but that's about it

  329. MattJ

    jonasw, https://trello.com/c/t79C3Yds/307-gdpr-advice added

  330. Ge0rG

    pep.: Intel SGX attempts to work around that, with limited success

  331. moparisthebest

    I thought intel SGX was completely broken

  332. pep.

    Ge0rG, if it was my hardware in the DC that would be a bit easier to do encryption I guess?. That still doesn't prevent DC people from fiddling with it. Is that what SGX is for?

  333. pep.

    If it's just a question of liability then I guess I don't need encryption at all, if a leak was caused by a hardware issues, or software issues at the virtualization level, I was told I could probably take it to the hosting provider.

  334. pep.

    Otherwise, if it's mistrust towards the provider, first I'm in a bad position, second, if I still want to do something about it, I guess LUKS on my rootfs with dropbear-in-initramfs would prevent "casual snooping". But protects in no way against a bit more elaborated "attacks"

  335. pep.

    (They have access to the virtualization software after all)

  336. moparisthebest

    pep., yea that's how my dedicated server in germany is set up, but it's really just to protect against the 'hard drives re-used without wiping' attack

  337. jonasw

    MattJ, thank you very much

  338. jonasw

    moparisthebest, did you get around to set up your XEP-0368 test setup?

  339. moparisthebest

    nope, also need to revive that thread and try to get some type of consensus

  340. moparisthebest

    dino is still doing it wrong (imho), gajim just released 368 support but not sure if it's right or wrong :)

  341. jonasw

    what is "wrong"?

  342. moparisthebest

    trying first xep-368 record, failing to connect, not trying any subsequent record

  343. jonasw


  344. Zash

    Does it work?

  345. moparisthebest

    in my specific case, the error it encountered was not-valid-xml, it gets HTTP back

  346. jonasw

    we have a PR for ALPN for aioxmpp, but I’m hesitant to merge it without testing.

  347. Kev

    'It compiles, ship it'.

  348. jonasw

    Kev, that’s a very very very bad idea for python code ;-)

  349. Kev

    Or in the case of Python, 'It commits, ship it'.

  350. moparisthebest

    jonasw, I can give you an account on my server, which requires alpn for ipv4 as the first SRV record, for informal testing

  351. jonasw

    moparisthebest, that would already be a good start.

  352. Zash

    'It turns into .pyc, ship it'

  353. jonasw

    send credentials to xmpp:jonas@wielicki.name. but don’t forget your privacy policy, I’m in the EU! ;-)

  354. moparisthebest

    oh right, well just tell me you aren't in the EU and I'll send you one :)

  355. jonasw

    I may or may not be in the EU.

  356. moparisthebest

    good enough for me, will send you one in a few :)

  357. jjrh

    google talk's xmpp support doesn't support message carbons does it?

  358. moparisthebest

    jjrh, I thought google completely turned off xmpp a couple months ago?

  359. moparisthebest

    but it never supported carbons anyway I think

  360. jjrh

    Nah you can still connect with username @ gmail.com

  361. jjrh

    (I just tested it today)

  362. Zash

    Federation is gone tho

  363. moparisthebest

    oh, so they just killed federation

  364. fippo

    zash: they closed port 5269?

  365. Zash

    fippo: Yup

  366. Zash

    Connection refused on all SRV targets

  367. Zash

    IIRC they gave out not-authorized errors just before that

  368. fippo

    so how long will it take them to remove the dns records...

  369. Zash


  370. Andrew Nenakhov

    moparisthebest, > oh, so they just killed federation Curse their sudden but inevitable betrayal!

  371. moparisthebest

    well it hasn't worked acceptably for years so, meh

  372. pep.

    Andrew Nenakhov, it's not sudden, they announced it at the beginning of 2017, for late June 2017 iirc

  373. Zash

    Hasn't it basically been outdated since 2006?

  374. fippo

    zash: well, someone said "the future is jingle" in 2011

  375. fippo

    but these days the future is stun2, turn2 and rtp3

  376. Andrew Nenakhov

    pep., > Andrew Nenakhov, it's not sudden, they announced it at the beginning of 2017, for late June 2017 iirc It's actually dates much earlier. After Google announced Hangouts, they began gradually chopping off parts of xmpp functionality one by one in a period of over 2 years.

  377. pep.

    Andrew Nenakhov: yeah but they officially announced it then

  378. Andrew Nenakhov

    Not really. They announced that 'nothing changes for current users', but it did, gradually. I'd even call it death by 1000 cuts, because it was clearly done so not to have another uproar like when they killed RSS Reader

  379. Ge0rG

    Maybe the responsible project lead was just promoted to greener pastures and the project fell victim to bit rot?

  380. moparisthebest

    that's the less cynical view

  381. moparisthebest

    I think what really happened is they wanted to lock users into their walled garden :P

  382. Zash

    Probably a bit of both.

  383. Ge0rG

    moparisthebest: yes, Google Management stated to lock in users some time around 2005. But I think there is still a large portion of CADT involved.

  384. jonasw

    you like that acronym, don’t you?

  385. fippo

    ge0rg: pah, getting rid of xmpp was clearly a technical decision because xmpp is based on http!

  386. Ge0rG

    jonasw: it perfectly fits how Google does IM.

  387. jonasw

    Ge0rG, to me, it feels more like what I’m hearing peripherially (I don’t follow sports, at all) about german football. Team didn’t perform for three weeks? Replace all training personnel.

  388. jonasw

    Ge0rG, to me, it feels more like what I’m hearing peripherially (I don’t follow sports, at all) about german football. Team didn’t perform for three weeks? Replace all training personnel and start over!

  389. Ge0rG

    jonasw: CADT as well.

  390. Ge0rG

    Except maybe for the higher age of the involved functionaries

  391. Zash

    define CADT?

  392. fippo

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

  393. fippo

    zash: you might also want to read up on the kevlar-shitting spiders

  394. Zash

    Ah, yes

  395. Zash


  396. moparisthebest

    ah hadn't seen CADT before but I like it

  397. jjrh

    I don't think what google turning off federation was to lock their users in - google doesn't have any issue with that.

  398. jjrh

    I think they mostly just didn't want to support XMPP. Probably turning off federation made sense since they didn't need to deal with that UI aspect.

  399. jjrh

    I'm guessing the majority of users didn't really use any of the federation stuff either.

  400. jjrh

    I never understood the google news reader thing though - ALOT of people used it, there were tons of apps that took advantage of the fact all your RSS subscriptions were on a account just about every android user has.

  401. Andrew Nenakhov

    Google Reader was good, but current Feedly is better. Though RSS seems to be on decline too, so many websites opt for this stupid telegram channels thing, locking themselves into yet another proprietary service

  402. jjrh

    The thing that was nice about google reader was you had a dozen or so apps that connected to google reader so you had a good amount of choice.

  403. SamWhited

    ooh, I haven't seen that one I don't think. I looked desperately for another feed reader that I actually liked after Google Reader shut down, but never found one and eventually gave up.

  404. moparisthebest


  405. moparisthebest

    good web ui, and good android client

  406. SamWhited

    That would require that I do work.

  407. moparisthebest

    on the other hand, no one else can turn it off on a whim SamWhited :)

  408. SamWhited

    Don't care since I can export an OPML bundle

  409. Zash

    I used liferea back in the day

  410. SamWhited

    Also, even if I wanted to self host I'm not running PHP on my server.

  411. jonasw

    good choice.

  412. jjrh

    Yeah there are a few other 'self hosted' choices https://github.com/Kickball/awesome-selfhosted#feed-readers

  413. Zash

    Now I just randomly go to sites when I remember them. Or I hear about stuff because people link to things.

  414. SamWhited

    Can liferea sync to some sort of backend and stay in sync with a mobile version? That's basically my only requirement (that and I don't want to host whatever that backend is)

  415. Zash

    Never got why it had to be a fkn web service

  416. Zash

    SamWhited: I have no idea that was even a thing people did

  417. waqas mumbles something about webscale

  418. jjrh

    Feedly is probably what you want - it has a web reader and a android app

  419. SamWhited

    Oh yes; I don't care if it's a desktop app or a webapp as long as I can read stuff on the bus and not have to figure out what I'd already read later.

  420. SamWhited

    Feedly does look like waht I was looking for at the time; I might give it a shot.

  421. jjrh

    What drives me nuts is so many sites don't actually post the whole article in the RSS feed.

  422. SamWhited

    Ooh yah, that always annoyed me

  423. Zash

    Reading on a bus seems like a recipie for feeling sick

  424. jjrh

    it's like a 2 line sentence with a link to the website - and I mean the whole point is I want to read the article in the rss reader optionally offline.

  425. SamWhited

    Doesn't bother me unless it's one of the big commuter busses

  426. moparisthebest

    tt-rss lets you write plugins to go to the website and grab the whole article anyway jjrh

  427. moparisthebest

    because yes, that's obnoxious

  428. jjrh

    That's nice.

  429. Zash

    It's probably all just fake news anyways!

  430. SamWhited

    Liferea looks nice, but doesn't appear to sync to anything, sadly :(

  431. jjrh

    Fake news is still news because people believe it's news and that's relevant

  432. SamWhited

    That's why I only subscribe to The Onion.

  433. jjrh

    I mostly read about Canadian politics and no cares about Canada enough about us to create a fake news conspiracy

  434. Andrew Nenakhov

    > it's like a 2 line sentence with a link to the website - and I mean the whole point is I want to read the article in the rss reader optionally offline. Websites need eyeballs to show ads. So it's understandable, but is still a nuisance

  435. jjrh

    In some cases - in many others I think they just don't have a clue.

  436. jjrh

    but unfortunately for them, google doesn't care about their ad dollars and scrapes their site with their 'newsstand' app or whatever it's called.

  437. jjrh

    some newspaper sites that have the 'you can read 2 articles for free then you gotta pay' thing are totally defeated by this.

  438. Ge0rG

    Some of the paywall sites also allow you to read stuff if you come from a social network referrer

  439. SaltyBones

    dafuq xmpp? somebody just sent a message without a username to a muc ..by accident!