XSF Discussion - 2022-04-16


  1. lovetox

    can someone write a simple status message in pep XEP

  2. lovetox

    i think anyone would migrate to this in an instant

  3. lovetox

    and we could safely ignore presence status message from that point on

  4. lovetox

    and no dont put show in their or anything else

  5. lovetox

    just the status message

  6. MattJ

    lovetox: yes, you can!

  7. lovetox

    i dont want to :/

  8. lovetox

    maybe there is someone who is really bored :D

  9. lovetox

    ok im writing this

  10. MattJ

    👩‍💻

  11. flow

    got for it :)

  12. flow

    go for it :)

  13. pep.

    <no-permanent-store/> should store until delivery to one device? In the XEP it only says where the message should not be stored (0136, 0313, chatroom logs). So what are other possibilities?

  14. Thilo Molitor

    lovetox: i can help with that if you want

  15. MattJ

    pep.: I guess that was the intention when I wrote it, yeah

  16. MattJ

    Don't know where it fits with plans to replace the current mod_offline behaviour

  17. pep_

    https://github.com/xsf/xeps/pull/1177 I have opened this PR fwiw, I should un-draft it shortly

  18. larma

    lovetox: I guess we first need a decent way to solve the issue of getting spammed with PEP notifications of things that haven't changed since last connect. PEP notification versioning anyone?

  19. Zash

    There's https://xmpp.org/extensions/xep-0312.html

  20. lovetox

    larma, of course this would be nice to solve, but i don’t see how its a pre-requirement for any XEP that builds on pubsub

  21. Zash

    move status from presence to pep? I sorta imagined you'd take the user activity xep and remove the activity, leaving the <text> field pretty much

  22. MattJ

    Maybe what should be done is to merge all the "User *" XEPs into one "User Status" XEP

  23. MattJ

    We discussed at the summit how different apps have different status formats - e.g. some just show text, some emoji+text, some a picture or even a video

  24. pep.

    Right when that is removed from poezio :x

  25. pep.

    (and moved off to a plugin)

  26. lovetox

    MattJ, if you mean by one "User XEP" one PEP node

  27. lovetox

    this would really limit the functionality

  28. lovetox

    as i can set different permissions for different pep nodes

  29. lovetox

    which seems important, with stuff like location for example

  30. MattJ

    Do you/users really care to individually set the visibility of what activity, mood and game you are playing?

  31. lovetox

    activity and mood are obsolete if you have a user status xep

  32. MattJ

    Yeah, location may be worth keeping separate. Also tune was one of the reasons that all these XEPs were made in the first place.

  33. lovetox

    what matters is, location and tune

  34. MattJ

    Because it changes frequently, and people were spamming <presence/> updates

  35. lovetox

    while i could see how tune has no special permission, as its also not very privacy sensitive topic

  36. lovetox

    location should be seperate

  37. MattJ

    I was mainly thinking that if you want to stick a specific <activity> payload in the status, that would be good

  38. MattJ

    and mood

  39. MattJ

    and clients can just implement/display that or not

  40. lovetox

    no its not good, its to complicated and from a time when no emojis existed

  41. MattJ

    Okay, just deprecate them and switch to <emoji>?

  42. lovetox

    just a <message>, and user can put in whatever he wants

  43. MattJ

    or don't try for any semantic separation, and just <text> and leave it up to client UIs?

  44. lovetox

    maybe in the future a, <url> to set a picture or whatever

  45. pep.

    If you want a client to interpret things in there, don't use "just" text. Otherwise it probably doesn't matter (but you may be limiting clients then to interpret things)

  46. lovetox

    thats the point, i dont see why i want the client to interpret something

  47. lovetox

    you should display whats in the text

  48. lovetox

    that is waht the user wants to communicate

  49. lovetox

    not a client on the other end that starts to interpret something

  50. pep.

    Maybe you don't have a use-case for it, (I personally don't either), but it's not a reason to limit it for others is it

  51. MattJ

    The problem is that <text> is not currently the most commonly-used status type

  52. MattJ

    It works as a baseline, but many modern apps don't use it

  53. MattJ

    or use it in combination with something else

  54. lovetox

    of course you can always extend with links to videos, pictures whatever

  55. lovetox

    but text is needed

  56. MattJ

    Agreed

  57. pep.

    A whitelist access model with nobody in it still allows one's own account right?

  58. lovetox

    thats what we use for bookmarks

  59. pep.

    Ok

  60. pep.

    I can just redirect to 0223 I guess for the privacy model

  61. pep.

    Is the use of @id='current' specified anywhere?

  62. Zash

    I think it's in '60 under singleton nodes

  63. Zash

    yeah https://xmpp.org/extensions/xep-0060.html#impl-singleton

  64. pep.

    Thanks

  65. mjk

    pep.: nitpick: you refer to pgp e2ee xeps as examples of forward secrecy, but isn't the whole point of having pgp over xmpp is being able to decrypt archives with your one and only key?

  66. pep.

    I prefer?

  67. pep.

    Ah sorry, refer

  68. mjk

    :)

  69. pep.

    I took that from the previous protoXEP, maybe that's worth changing

  70. pep.

    I don't know if it's the whole point, but to pgp users that's definitely a feature yeah, I guess

  71. mjk

    Right 'the whole' is rather a hyperbole

  72. pep.

    If you have an idea how to rephrase this I'm happy to edit it

  73. pep.

    Just removing mentions of PGP there? First paragraph(?)

  74. pep.

    mjk, you can still use PGP and want your messages not to be stored forever on the device right

  75. pep.

    This isn't incompatible

  76. mjk

    Yeah, just remove pgp xep mentions, as they don't provide FS

  77. pep.

    Somebody getting hold of your device could "just" make it fetch everything again and *poof*, but I guess PGP users may be aware of this? (not?)

  78. pep.

    That would be equivalent to trusting the server's privacy policy I guess

  79. mjk

    And depends on archiving settings, and those seem to be usually turned on by default

  80. pep.

    They're enabled almost automatically by clients using MAM

  81. pep.

    (all of them)

  82. pep.

    They're actually turned off by default on servers

  83. pep.

    At least I know it's the case in prosody, and I think ejabberd?

  84. mjk

    Not sure if by clients. I know Gajim and C. don't turn mess with mam settings

  85. pep.

    Yeah but they just need to require MAM archives for prosody to turn them on

  86. pep.

    Yeah but they just need to require MAM for prosody to turn them on

  87. mjk

    Oh, that's some dark xmpp corner

  88. pep.

    (Is 'MAM archives' also a what's it called? ATM Machine thing.)

  89. pep.

    Currently I'm wondering if the minimal timer value should be in PEP so all devices of an account use the same value to prevent loops and all. Or maybe the solution to this is for clients not to send a new value whenever the sender sends something that is lower to the minimal value.

  90. pep.

    Maybe I should just drop the minimal value and set a minimal value in the spec, but then I have to decide of a minimal value and I don't know everybody's use-case.

  91. pep.

    Or I don't, and users will just decide between them.

  92. pep.

    Using PEP on MUC for this would have been nice though as there's an obvious node to go fetch

  93. mjk

    Yea, magic numbers in specs is no good. For some, it can be weeks, for others, minutes

  94. pep.

    I guess that's more of a social issue really and putting this in the spec is too much a hassle

  95. pep.

    If somebody wants their messages not to be readable well then they'll put 1 second and the recipient will complain.

  96. pep.

    (or 0 even, since that's allowed in the type)

  97. pep.

    Yeah I've convinced myself

  98. mjk

    > Is 'MAM archives' also a what's it called? ATM Machine thing. Answering machine?

  99. pep.

    No I meant to point out the repetition. The 'M' in ATM already is 'Machine'

  100. pep.

    The 'A' in MAM already means 'Archive'

  101. Zash

    Time to start using MAMA Archive

  102. mjk

    Idk, MAM is archive 'management', not 'archive' itself

  103. mjk

    I prefer saying just 'server archive'

  104. pep.

    Yeah but that can designate something different. Even though nowadays everybody does MAM

  105. mjk

    And MAM is a xep to manage that :)

  106. mjk

    You won't believe this, but there's another instance of 'Draft' on xmpp.org: > […] production systems are advised to carefully consider whether it is appropriate to deploy implementations of this protocol before it advances to a status of Draft. That's in the red carpet of experimental xeps. This Draft thing really dies hard, perhaps a `grep -R Draft .` over the entire repo is needed?

  107. mjk

    Yeah, that wasn't a very good idea

  108. mjk cleans chunks of xml off their terminal

  109. pep.

    mjk, I updated the spec if you want to have another look :)

  110. mjk

    Too busy repainting the bike shed rn!

  111. pep.

    :)

  112. mjk

    https://http.xmpp.xyz:5281/upload/hJMoebU-CPeZ8QBp/pgCZZS_HQzCHhAtbSKJJvg.patch

  113. mjk

    jonas’: would this ↑ be alright, or are PRs much preferred?

  114. mjk

    For the record, the rest of the Draft occurences are either in xeps' xml, or in a readme (which doesn't seem to be rendered anywhere?)

  115. mjk

    pep.: > even though keys are deleted, message contents is retained both in server and client archives Effectively, when the ephemeral key is destroyed, the server-archived message is destroyed, as it now just consists of random bits. "Spooky deletion at a distance" is what I call it. :) So, when speaking of forward secrecy, server archive becomes irrelevant to privacy, only disk space

  116. mjk

    Or I should say "when _every_ recepient of a message destroys the message key"

  117. mjk

    Well, I think you understand the concepts, I'm just pointing out it's not very clearly expressed in that paragraph :)

  118. jonas’

    > jonas’: would this ↑ be alright, or are PRs much preferred? anything which does not generate an email will be lost this weekend

  119. jonas’

    mjk: ^

  120. pep.

    mjk, hmm, indeed. That's also original phrasing that I kept (just the first sentence in the paragraph). Any suggestion?

  121. mjk

    jonas’: okay then. It won't be lost in my repo, and I'll find time for github later :)

  122. mjk

    pep.: start with deleting 'both in server and ' and see if the rest sticks?

  123. mjk

    I'll need to look at the intro more carefully, can't right now

  124. pep.

    Yeah I just did that and the rest isn't too weird.. we'll see

  125. rion

    Hey guys. Can anyone help me to pay for my xmpp.psi-im.org? I'll make a visa transfer to your card in advance.

  126. rion

    Paypal/stripe do not work in Belarus