XSF Discussion - 2017-10-09


  1. vanitasvitae

    > prime audience for the use of end to end encryption (drug addicts) [...] Sad how scared of authorities he seems to be :(

  2. jonasw

    I think in russia it’s a bit healthy to be scared of authorities......

  3. mathieui

    in western countries too, tbh

  4. jonasw

    mathieui, I think there it’s healthy to have respect of authorities. Scared is another level.

  5. mathieui

    well, not "scared", but "expecting the worse of them at all times"

  6. Flow

    Is there a way to retrieve all newer PubSub items after the item with ID X?

  7. jonasw

    Flow, with RSM possibly

  8. jonasw

    Result Set Management, it’s a XEP

  9. jonasw

    apparently some pubsub services support that

  10. Flow

    I don't found anything in xep59, xep60 and xep313

  11. jonasw

    wouldn’t you give the item ID as minimum and then query everything after that?

  12. Flow

    RSM appears to be more about "paging through the resulting set"

  13. Flow

    not about that particular query

  14. Flow

    it is not necessarly clear that the RSM IDs are the same as PubSub item IDs.

  15. jonasw

    yeah

  16. jonasw

    I’ve seen them differing, too

  17. MattJ

    I think that's what https://xmpp.org/extensions/xep-0313.html#sect-idm139955456388768 was for

  18. intosi

    MAM ids != PubSub item IDs. If only because those are not guaranteed to be unique.

  19. Flow

    too many IDs ;/

  20. Ge0rG

    Flow: that's what I said, recently.

  21. Holger

    Flow: https://xmpp.org/extensions/xep-0312.html

  22. Ge0rG

    Flow: but you didn't like it :(

  23. Flow

    Ge0rG: out of context

  24. Holger

    Flow: Not sure (where) that's implemented, though.

  25. intosi

    On any given service, only (node id, item id) needs to uniquely point to an item. But you can republish an item with the same ID, which would defeat the purpose of the id in MAM.

  26. Ge0rG

    Flow: "too many IDs" is true irregardless of context.

  27. Holger

    Flow: Ah nonsense, that only works with a timestamp.

  28. Flow

    So there is no way to catch up on pubsub items while offline, presuambly because the assumption is that you will get the notifications about new items also while offline?

  29. Holger

    Hmm well 0312 does somewhat try to provide that, no?

  30. Flow

    I really start to dislike xep0060. Not only is there no way to reliable get the publisher's JID, you also can't do the simplest queries…

  31. Kev

    That's why someone at the Summit was suggesting using 313 for 60 queries.

  32. Holger

    goffi's PubSub component does MAM, I think.

  33. vanitasvitae

    jonasw, If I want to make changes to JET, shall I make these changes in xep-0391, or inbox/jet?

  34. Ge0rG

    vanitasvitae: inbox/jet should cease to be a thing as soon as it's got a number

  35. vanitasvitae

    okay :)

  36. jonasw

    Ge0rG, no, it still exists for historical purposes

  37. jonasw

    vanitasvitae, updates against xep-0391.xml

  38. Ge0rG

    maybe we could get rid of that, we've got git history after all.

  39. jonasw

    Ge0rG, I think the purpose is to have it on the webserver too

  40. vanitasvitae

    jonasw, I noticed, that the XEP mentions a dependency on 234 (jingle file transfer). Is this necessary? JET should work for other Jingle usecases too, even when a client does not implement 234.

  41. vanitasvitae

    Looks like I added that dependency, when the document was still titled "Jingle Encrypted Transfers"

  42. jonasw

    vanitasvitae, I didn’t add that depedency :)

  43. jonasw

    feel free to modify that

  44. vanitasvitae

    okay

  45. jonasw

    in Experimental, you can as author basically everything

  46. jonasw

    in Experimental, you can do basically everything as author

  47. jonasw

    you just need to preserve compatibility

  48. vanitasvitae

    okay. I should also add revisions, right?

  49. jonasw

    vanitasvitae, you SHOULD, yes. An editor will do that if you don’t.

  50. jonasw

    when you do it, you have the chance to make the revision text as meaningful as possible :)

  51. vanitasvitae

    :D I'll do you the favor :)

  52. jonasw

    if you would follow the spirit of semantic versioning when choosing the next version number, that’d be great

  53. Flow

    So MAM with PubSub only improvement are the query conditions 'start', 'end' and 'with'. But we could specify a PubSub specific (after|before)-pubsub-item-id?

  54. Zash

    Does mam+pubsub mean mam-rsm ids are pubsub ids or?

  55. Flow

    Zash: intosi says no

  56. Zash

    Altho with pubsub ids being user-controlled ... and sometimes always "current", glhf

  57. Flow

    For the same reason I had to invent a new "value" for the "Atomically Compare-And-Publish PubSub Items" XEP

  58. Flow

    (Of course the reasoning is not false, I just think that it is a design flaw that IDs may point to different content, instead of having an <item-retracted/> marker)

  59. Flow

    Zash "always "current""?

  60. jonasw

    Flow, maybe you could first specify your new ID for re-use by different things and then build CAP on top of that?

  61. Flow

    jonasw: I'm not sure if it's of any use

  62. Zash

    Flow: In a bunch of PEP things, the item ID is always the string "current"

  63. Flow

    besides of the compare-and-publish case

  64. Zash

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

  65. intosi

    Zash: indeed. And just the fact that the Item ID is something the publisher can provide, instead of letting the server guarantee its uniqueness across the service, is enough to invalidate it for MAM id.

  66. Zash

    Yeah

  67. Zash

    Like with message@id

  68. Zash

    Hm, does that mean thaw MAM+PubSub is doomed?

  69. Flow

    Zash: You can still use 'start', 'end' and 'with'

  70. Zash

    Yes

  71. Flow

    And specify further fields for MAM+PubSub like after-pubsub-item-id

  72. Zash

    Does the pubsub service have final say over the item id?

  73. intosi

    MAM with Pubsub has semantics that differ from Pubsub retrieval anyway, as you're likely to get the same itemid more than once in some use cases.

  74. MattJ

    I was never convinced about MAM+pubsub

  75. Zash

    MattJ: Weren't you the one who said "Just use MAM?" back then?

  76. MattJ

    niirc :)

  77. Zash

    I REMEMBER EVERYTHING .. almost

  78. intosi

    I use MAM instead of remembering everything.

  79. Ge0rG

    I use a server-side client with logging.

  80. MattJ

    ^

  81. Ge0rG

    MAM is too bumpy.

  82. Zash

    Or you can just forget everything the instant it scrolls off the screen. If it was important then they shouldn't have written it over an ephemeral instant messaging thing! :)

  83. Nathan

    ...

  84. edhelas

    we can change the 0060

  85. edhelas

    it's not like we have many Pubsub + MAM implementation in the wild :p

  86. Nathan

    are there any impacts of MAM?

  87. Zash

    Impacts?