XSF Discussion - 2019-01-18

  1. Guus

    Does MAM explicitly define what happens to an archive if the address for which the archive is kept disappears (eg: a MUC room gets destroyed)?

  2. Guus

    Deleting the archive seems logical, but might mess with locally stored archives?

  3. Daniel

    > This is going to break PM archival on the user account, right? Muc pm will end up in the user archive not in the group

  4. Ge0rG

    But when you query the user archive, how do you know who was the owner of that nickname at that time?

  5. Daniel

    You don't

  6. Daniel

    Is this bad?

  7. Daniel

    In the scope of how terrible muc pms are in general I mean

  8. Ge0rG


  9. Ge0rG

    From a pragmatic point of view, you can't tell who is the author of any given message in a semi-anon MUC anyway, so scoping LMC and MAM to "identities" is pointless.

  10. Daniel

    If you want to make lmc work in PM (personally I'd say f*ck it who cares about muc pm) I'd relax the rules and match only by full jid and id. Because for PMs it's hard for someone else to guess what id you used

  11. Daniel

    Contrary to group chat messages where the new guys sees the ids

  12. Ge0rG

    Daniel: I'd say "match LCM on full JID and ID in MUCs as well, because you don't know who was the original author, so why should you enforce who the editor may be"

  13. Guus

    Future historians will look back at this point in time and define it to be where the anarchy started.

  14. Ge0rG will just inject a Carbon.

  15. Ge0rG

    I wonder how many clients still didn't fix that.

  16. Daniel

    I think a bot that joins mucs and than randomly changes messages from people who are no longer joined to 😂 is against user expectations

  17. Ge0rG

    what about a bot that joins MUCs with a recently deceased nickname and floods the MUC with garbage? Or with some NLP based random sentence generator trained on MUC logs of that specific nickname?

  18. Ge0rG

    Imagine a bot that be-rants this MUC about how bad everything this, in my name.

  19. Guus

    We assumed that Ge0rG already is such a bot, Ge0rG 😉

  20. Ge0rG

    Guus: no way! BTW, did my trademark-related email reach you under the board address?

  21. Guus


  22. Ge0rG


  23. Ge0rG

    Guus: I assume you received peter's response to my initial email because he's subscribed to board@

  24. Ge0rG

    Communicating with board is kind of hard if everybody is blocked.

  25. Guus

    The last message I received on the subject is Peter's response on Fri, 11 Jan 2019 11:34:48 -0800 (PST)

  26. Guus

    I think that the policy for the board mailing list is in place to allow for private discussion - but I'm actually unsure. ralphm ?

  27. Ge0rG

    Guus: I've responded to that yesterday, also considering the Board feedback.

  28. Guus

    Ge0rG I did not receive that. I'm unsure if anyone did, unless you cc'ed individuals.

  29. Guus

    you didn't get a bounce or any other indication that there was a problem?

  30. Ge0rG

    Guus: I sent it to peter, with ralphm and board in CC, and got no bounces

  31. Guus

    I'm unsure what the intended behavior is, but it worries me that things disappeared in a black hole, without notification.

  32. Guus

    Can someone from iteam comment? ( Kev intosi MattJ )?

  33. MattJ

    iirc that is the case, but I've never had anything to do with the mail server and lists so I'll defer to someone who actually knows

  34. Guus

    Ge0rG as I'm interested in what you wrote, can you forward it to me individually?

  35. Ge0rG

    Guus: bounced to your gmail

  36. Guus

    I received it 0 minutes ago.

  37. jonas’

    it is intended AFAIK because mailing lists get loads of spam

  38. jonas’

    so you drop evevrything from unsubscribed because moderating this manually requires 10-30mins of work each day

  39. Guus

    no bounce?

  40. jonas’


  41. Ge0rG

    the proper way™ to do this is to reject the incoming mail

  42. Ge0rG

    so that the originating server will do the bouncing

  43. Ge0rG

    but email is the most hysterical of infrastructures, so you can't have nice things.

  44. Ge0rG

    anyway, this is not the first time that my attempts to contact board are null-routed. Must be some kind of tarpitting or shadowbanning or howitscalled.

  45. Kev

    It is as Jonas describes it. Everything from a non-member of the list is dropped.

  46. Guus

    can we reject that instead, as Ge0rG describes? A silent drop causes confusion.

  47. Kev

    Not really.

  48. Kev

    That's a great way to get us on blacklists.

  49. Kev

    Because email.

  50. oli

    after 20 years of xmpp, shouldn't there be something like a replacement for mailing lists?

  51. oli

    reading this discussion is weird?

  52. oli


  53. oli

    the xmpp guys are unable to communicate

  54. Zash

    It doesn't have to replace everything

  55. jonas’

    Guus, rejecting email causes backscatter, which is a huge problem when the source address was spoofed in the first place. essentially, you become a more-or-less open relay :(

  56. goffi

    We have XMPP blogging already, and it could be used with email gateway.

  57. Zash

    PubSub discussion groups isn't a thing already?

  58. Seve

    I would prefer to use some kind of XMPP forums instead of mails, indeed

  59. jonas’

    goffi, so yet-another-thing which tries to interface with email and will fail horribly?

  60. jonas’

    just like discourse does

  61. jonas’

    the only thing which reasonably interfaces with email is github, actually

  62. goffi

    jonas’: why would it fail ?

  63. jonas’

    goffi, because interfacing with email isn’t as trivial as people think. e.g. threads

  64. goffi

    we can map threads with XMPP blogging.

  65. goffi

    That's something I have in my head for long, I'm hopping to do a proof of concept in the coming months.

  66. Ge0rG

    jonas’: maybe I'm misunderstanding the tech, but how exactly does *rejecting* email (as opposed to accepting it and then sending a bounce) cause backscatter that anybody can make *you* responsible for?

  67. Kev

    It wouldn't, but that's not possible with mailing list thingies.

  68. Ge0rG

    Wait, a mailing list thingie that's configured to only accept mails from members can do that, right?

  69. Kev

    In theory or in practice?

  70. ralphm

    What we could maybe do is accept members from the members list as allowed to send to board. I seem to remember Mailman supports that.

  71. Ge0rG

    Kev: it depends on whether we are talking about envelope-from or inline from.

  72. ralphm

    With @list syntax, apparently.

  73. ralphm

    “Another way to deal with this is to use the @Listname syntax to add the other sublists to accept_these_nonmembers of each sublist.”

  74. ralphm

    Kev: does that help?

  75. Ge0rG

    Apparently the mailman developers didn't understand the finesse of mail rejection a decade ago.

  76. jonas’

    Ge0rG, no, mailing list delivery happens too late

  77. jonas’

    you’d have to do that while the SMTP connection from the origin is still open, which is not how that happens

  78. Ge0rG

    jonas’: this is an excuse.

  79. Ge0rG

    if you can reject a message based on a heavyweight perl-based bayesian framework check magic, while the smtp connection is still open, you surely can do it based on a small list of whitelisted addresses.

  80. jonas’

    Ge0rG, can you actually do that?

  81. Ge0rG

    it only didn't happen because of lazyness, ignorance, or -EBUSY.

  82. jonas’

    all SA deployments I’ve seen do not reject but only classify

  83. jonas’

    (for the reason that it’s hard)

  84. jonas’

    (and that you might not want to reject based on bayes alone, actually)

  85. Ge0rG

    jonas’: it's only hard because MTA configuration files use arcane syntax that's incomprehensible to mere humans.

  86. ralphm

    I'm sure they accept patches

  87. jonas’

    ITYM because mail is a complex problem, like IM.

  88. Ge0rG

    jonas’: I'm not complaining about prosody syntax. Except for the VirtualHost directive being an implicit scope for everything that follows.

  89. Ge0rG

    jonas’: you won't solve a complex problem by exposing an even more complex configuration file format, and forgetting to document it.

  90. jonas’


  91. jonas’

    which software are you even talking about? :)

  92. oli

    the mail server

  93. Ge0rG

    jonas’: my personal experience is with postfix, but I've horrible memories of sendmail config, and no positive memories of exim

  94. oli

    postfix is not that bad

  95. jonas’

    Ge0rG, postfix is well documented though

  96. ralphm

    Ge0rG: sendmail config with or without m4?

  97. Ge0rG

    jonas’: my master.cf contains three different invocations of `smtpd` with different parameters, which I cargo-culted from the internets and don't have the slightest clue about why they are there. It mostly works.

  98. Ge0rG

    jonas’: so far I didn't find a single HOWTO that tells you *why* you have to paste them in and what they mean, only snippets that you are supposed to copy verbatim.

  99. Ge0rG

    goffi: good luck with FOSDEM! :)

  100. melvo

    I would like to get an account on wiki.xmpp.org. Can someone help me?

  101. Ge0rG

    melvo: please state your desired username in CamelCase, your email for the password transmission and an optional realname; here or via PM

  102. Ge0rG

    melvo: in case you are the melvo who tried to contact me yesterday and today, please use xmpp:georg@yax.im as that other JID is for my private use and not supporting OMENNO encryption.

  103. Ge0rG

    I'm also on the verge of leaving mobile network coverage for the weekend. Sorry.

  104. melvo

    > melvo: in case you are the melvo who tried to contact me yesterday and today, please use xmpp:georg@yax.im as that other JID is for my private use and not supporting OMENNO encryption. Yes, that was me. OK, thanks.

  105. goffi

    Ge0rG: thanks, there is a bunch of things to do :)

  106. Ge0rG

    The summit banner on xmpp.org is covering up the hamburger menu on mobile devices...

  107. Ge0rG


  108. Zash

    Web \o/

  109. pep.

    Ge0rG, no hamburger for you.

  110. Ge0rG hungry!

  111. ta

    Ge0rG: you can copy some marketing phrases for yaxim from that

  112. Ge0rG

    ta: I'd rather not. Maybe they are trademarked...

  113. ta

    Yaxim, mastering the most secure messaging system standard.

  114. oli

    ta: what are the less secure *standards*?

  115. ta

    maybe talk

  116. neshtaxmpp