XSF Discussion - 2022-10-14


  1. emus

    👍

  2. lovetox

    We need to get some author onto xep-0292

  3. lovetox

    the XEP is in my opinion not even remotely doing what we need

  4. lovetox

    it moves vCard to pubsub, but invents new IQ semantic for get/set, instead of using pubsub

  5. lovetox

    maybe reminder why we need this in pubsub

  6. lovetox

    1. we need privacy settings, which pubsub gives us (XEP actively says it the community decided AGAINST privacy settings)

  7. lovetox

    2. We dont need servers to implement custom vCard4 support for this to work

  8. lovetox

    which makes adoption *really* easy

  9. singpolyma

    Yes, we just use pubsub for vcard4 and ignore the special iq

  10. lovetox

    yeah problem is people moving to that, and i dont want them to waste there effort

  11. MattJ

    lovetox: happy authoring 🙂

  12. MattJ

    You sound sufficiently motivated

  13. lovetox

    the custom IQ semantic can only be optional

  14. lovetox

    its insane to me that this iq semantic was introduced 12 years ago

  15. lovetox

    talk about slow moving eco system

  16. singpolyma

    I think the custom IQ should be cut from the xep and it should specifically say that pep works

  17. lovetox

    MattJ, i know i could take the time to do a MR

  18. lovetox

    I always feel im not so good with words when trying to write such texts

  19. nicomuc

    Does that mean that the full vcard will be part of the pubsub#event, or will the ‘empty item as update notification’ still be valid?

  20. lovetox

    we should hire technical writers

  21. MattJ

    This is something on my list for Board to discuss

  22. MattJ

    We used to have one, and they were the Editor

  23. MattJ

    Then the Editor role became... what it became

  24. moparisthebest

    lovetox, just do it, any text is better than nothing

  25. moparisthebest

    besides if you write some incorrect grammar 15 people will trip over each other trying to fix it :D

  26. singpolyma

    lovetox: if you want I could draft something. The process is just send a PR to the xep in some GitHub repo?

  27. moparisthebest

    yep!

  28. singpolyma

    And I guess this is allowed for anything pre-stable?

  29. moparisthebest

    getting it merged is someone else's problem hehe

  30. moparisthebest

    iirc the shortcut is asking the author if it's ok, otherwise council can force it through or something? been awhile

  31. lovetox

    nicomuc, the eventnotification thing has to be also looked again

  32. lovetox

    it makes not sense like it is now

  33. lovetox

    what a client wants is to know when a vcard has new content it doesnt know

  34. lovetox

    the event like it is described in the xeps just tells you on coming online "there is some vcard"

  35. lovetox

    means its not really useful, so yes if you want to do this right, you probably would need some kind of metadata node

  36. lovetox

    that tells you the timestamp or hash of the vcard

  37. singpolyma

    That's a generic pubsub/pep issue, not specific to vcard

  38. singpolyma

    The xep mostly just needs to define the pep node name and say "use vcard4 XML in this pep node"

  39. lovetox

    yes we could try something like, maxitems=1, then always publish with a itemid=hash of vcard

  40. singpolyma

    "how to hash XML" is a whole thing, but yeah, not specific to vcard problem IMO

  41. lovetox

    yeah hash is maybe to complicated

  42. lovetox

    but everything else is tricky

  43. lovetox

    hm, actually not, just put a UUID in there

  44. lovetox

    we just need to make sure that subsequent publishes dont have the same id

  45. singpolyma

    Right

  46. lovetox

    then a sentence about node config so events dont send payload, and i think we achieved greatness

  47. singpolyma

    Sounds fine. I don't really see why you don't want payload sent, but doesn't affect my current use case either way so fine

  48. lovetox

    because the use case is that you send +notfiy in presence

  49. lovetox

    this means on coming online the server will send you the last event for every user

  50. lovetox

    if the payload is in, it means you get for every user in your roster a full vcard that you often already know

  51. lovetox

    its the same reason we use a devicelist node for omemo, or metadata node for avatars

  52. singpolyma

    +notify sends even if it sent to you before? Gross, but ok

  53. Zash

    no device tracking, especially not by remote servers, so they can't know if they sent you this thing already or if you're a new device with the same caps as another

  54. singpolyma

    Zash: well, I expected just like normal pubsub they send to you when the publish happens and that's it

  55. singpolyma

    You have to know what you follow and fetch to catch-up later otherwise

  56. MattJ

    Then clients have to make a request to every node of every contact they are interested in

  57. Zash

    Or we could do what ejabberd did, except negotiate it, and send notifications to the account and use MAM

  58. singpolyma

    Zash: right. Stop using headline if people actually want the content always

  59. Zash

    https://xmpp.org/extensions/xep-0312.html is another possibility

  60. Zash

    more stateless

  61. singpolyma

    Hmm, yeah