XSF Discussion - 2019-10-01


  1. jonas’

    pep., you can set arbitrary headers from javascript

  2. jonas’

    (mostly arbitrary)

  3. jonas’

    so a web client could make good use of that

  4. COM8

    Could somebody pleaser rerun travis for PR #574? (https://github.com/xsf/xmpp.org/pull/574)

  5. jonas’

    COM8, rebase your branch on current master to fix that issue

  6. jonas’

    (that’ll also re-trigger the CI automatically)

  7. COM8

    OK

  8. COM8

    Ok, now I need somebody that reviews the changes again 😀

  9. pep.

    For some reason I can't merge anymore? hmm.

  10. pep.

    Or maybe I've never been able to merge on xmpp.org

  11. Ge0rG

    🎶 I can't code 🎶 I can't merge 🎶

  12. Ge0rG ,oO( next playing: you can keep your patch on )

  13. pep.

    :)

  14. Guus

    what? no XEP-0118?

  15. Ge0rG

    Guus: you can't have them directly in messages.

  16. pep.

    That's fine, juste have the MUC subscribe

  17. pep.

    And display them for who's interested

  18. Ge0rG

    Technically, the MUC could act as a PubSub subscription proxy.

  19. Ge0rG

    like it does for avatars

  20. pep.

    it doesn't? :(

  21. pep.

    Because we're still using that presence hack

  22. pep.

    Maybe if previous council hadn't rejected Link Mauve's PR..

  23. Ge0rG

    current council is almost a previous council now, so it's time to bribe the next council

  24. pep.

    yeah

  25. jonas’

    unrelated: I’m hungry

  26. Ge0rG

    jonas’: there is no XEP for that ;)

  27. jonas’

    we *do* have User Mood

  28. jonas’

    also, I still need to write my council application

  29. Daniel

    unrelated: I’m hungry and suprisingly far away from food that’s not oreos

  30. jonas’

    oreos is at least something

  31. pep.

    We're two on that pack of oreos, it won't last long

  32. jonas’

    then daniel was incorrect about his statement.

  33. jonas’

    or maybe has a rather narrow definition of food :-X

  34. pep.

    I hope things don't get to that

  35. Ge0rG

    never mess with hungry nerds.

  36. COM8

    Other non food related question: Any reason why the URL in the subject for the logs still is http? https works fine for me

  37. Ge0rG

    It used to not have the right cert on the logs host.

  38. COM8

    but now this has been taken care of and we can change it?

  39. jonas’

    cc @ MattJ ^

  40. dwd

    In the MLS WG Interim today.

  41. jonas’

    dwd, will you be able to send a council agenda?

  42. dwd

    .

  43. Daniel

    jonas’, lots of clients will do a get after a receiving 302 to a post

  44. Daniel

    so muclumbus search is currently broken on all Conversations

  45. Daniel

    because the http library i'm using does follow the same 'broken' behaviour on purpose

  46. jonas’

    Daniel, derp, yeah

  47. jonas’

    going to fix that ad-hoc

  48. jonas’

    I didn’t recall that you need a POST for search

  49. jonas’

    Daniel, should be fixed

  50. dwd

    Anyone got any MUC log data? Joins, message, leave? MLS Working Group is asking.

  51. Kev

    xmpp.org and jabber.org both should have some.

  52. dwd

    Kev, Public and machine readable?

  53. Daniel

    jonas’, what change did you make? because it doesn’t appear to be working

  54. Ge0rG

    I have some years worth of poezio logs, but those lack join / part history

  55. pep.

    I have join parts I think

  56. Kev

    Jabber.org should have machine readable stuff, but not public, would have to ask Peter for it.

  57. Ge0rG

    dwd: I have some years worth of poezio logs, but those lack join / part history

  58. Daniel

    jonas’, in fact i can’t see any changes. still redirecting

  59. dwd

    Oh, wait, can I get the join/leave/presence updates off MAM these days?

  60. Ge0rG

    dwd: most probably not

  61. dwd

    It's the joins and leaves of individual clients we could use, here.

  62. pep.

    MR 20191001T10:38:17Z 000 <Kev>  xmpp.org and jabber.org both should have some. MR 20191001T10:38:34Z 000 <dwd>  Kev, Public and machine readable? MI 20191001T10:38:44Z 000 ---> kokonoe joined the room MR 20191001T10:39:32Z 000 <Daniel>  jonas’, what change did you make? because it doesn’t appear to be working MR 20191001T10:39:39Z 000 <Ge0rG>  I have some years worth of poezio logs, but those lack join / part history MR 20191001T10:39:50Z 000 <pep.>  I have join parts I think MR 20191001T10:39:52Z 000 <Kev>  Jabber.org should have machine readable stuff, but not public, would have to ask Peter for it. MR 20191001T10:40:06Z 000 <Daniel>  jonas’, in fact i can’t see any changes. still redirecting

  63. pep.

    I have join/parts

  64. dwd

    pep., Ah, interesting.

  65. pep.

    44M for my xsf@ logs

  66. pep.

    I can upload that somewhere

  67. dwd

    In that format?

  68. pep.

    yep

  69. pep.

    It's machine readable. There's no distinction between presence and messages semantically, but "--->" and "<---" with no nicks should be good enough markers I think

  70. pep.

    Unless a MUC admin wants to troll you

  71. jonas’

    Daniel, but it’s a 301 now

  72. Ge0rG

    Unless somebody joins as `---`

  73. dwd

    Probably better than nothing, but in a perfect world I'd have device data.

  74. pep.

    Ge0rG, no, you'd see <--->

  75. Ge0rG

    Ah, it's a difference between the stored and the displayed format

  76. pep.

    yep

  77. Daniel

    jonas’, that has the same broken properties

  78. Daniel

    can we not redirect at all?

  79. jonas’

    Daniel, okay, working on that

  80. jonas’

    will have to figure out the correct rewritecond for that

  81. jonas’

    give me a minute

  82. Ge0rG

    HTTP is the broken property here...

  83. Ge0rG runs and hides

  84. jonas’

    indeed

  85. Daniel

    Obviously

  86. jonas’

    Daniel, should be fixed

  87. pep.

    oh, actually presence and messages are marked differently

  88. pep.

    MI / MR at the beginning of the line

  89. Daniel

    I mean it's not really broken. They just clarified a must not with a should

  90. jonas’

    "clarified"

  91. Daniel

    Small typo fix basically

  92. pep.

    dwd, ^

  93. jonas’

    Daniel, for 301, I think it’s always been a "keep POSTing", not a "convert to GET"

  94. Ge0rG

    pep.: wasn't there an &nbsp; in the mix, used as a separator?

  95. jonas’

    Daniel, in any case, I removed the redirect for ^/api.+$ URIs

  96. pep.

    hmm, not impossible, I don't recall the exact format

  97. pep.

    Link Mauve, ^

  98. Daniel

    > Daniel, for 301, I think it’s always been a "keep POSTing", not a "convert to GET" Maybe. But nobody was doing this. And "get after post" is actually a pattern in webdev

  99. Daniel

    To avoid the f5 resend isuue

  100. Daniel

    jonas’: fixed indeed. Thanks

  101. jonas’

    Daniel, sorry for messing this up, I could’ve sworn that the search via API was a GET and not a POST

  102. jonas’

    (I actually thought about this when setting up the redirect last night)

  103. jonas’

    should’ve double-checked

  104. Zash

    dwd: MUC? xmpp.org has logs wit join and part info

  105. Daniel

    jonas’: no worries. Thank you for offering this service

  106. Zash

    Search via POST? That's unusual

  107. jonas’

    Zash, yeah, I put it on my mental todo to also allow GET

  108. jonas’

    and in a v2 of the API it’ll be GET only

  109. !XSF_Martin

    debacle here? I wonder if he (aka debian xmpp team) wants to share the newsletter to planet.debian.org 😃

  110. dwd

    Zash, That'd be useful.

  111. Ge0rG

    Everybody is talking of The Newsletter, but has it been tweeted yet?

  112. debacle

    however, the Debian XMPP team blog is also aggregated on Planet Jabber, which probably already has the newsletter - it would be duplicated there https://planet.jabber.org/

  113. MattJ

    I don't think that matters

  114. Ge0rG

    debacle: you would cause infinite recursion

  115. MattJ

    That is, I don't think it's a reason to not broadcast to a wider audience :)

  116. Ge0rG

    The Debian XMPP team could up their blogging.

  117. jonas’

    Daniel, FYI, the search endpoint now supports both GET and POST

  118. jonas’

    nyco, it doesn’t hurt to add a ping for urgent PRs here, multiple channels (email + xmpp) help to remind me when I’ve seen a message one of the channels

  119. Ge0rG

    jonas’: one ping only!</Scottish-Russian-accent>

  120. j.r

    Could the Message Fastening XEP (https://xmpp.org/extensions/inbox/fasten.html) be a way to have proper answering/threading instead of citations?

  121. jonas’

    j.r, the protocol isn’t the problem, how to deal with clients which don’t support it is the problem

  122. jonas’

    XMPP has support for threading in RFC 6121 even, but no client does it because it’s unclear how to handle clients which don’t support or don’t want to support threads

  123. j.r

    jonas’: yeah that's a interesting thought... you could do some legacy citation but how to filter it out for clients that support it 🤔