XSF Discussion - 2018-08-08


  1. moparisthebest

    flow: the idea behind xmpp and xmpps in the same bucket is to allow server admins to define priority+weight between them

  2. moparisthebest

    In practice you can do whatever

  3. SamWhited

    It should probably not be a 'should' then. I don't understand why server admins would ever want to do that

  4. flow

    moparisthebest, I get that, but why should server admins want that?

  5. flow

    FWIW, i've recently implemented "multiple SRV service/proto ordering" in MiniDNS. But I still haven't heard a reason why the XEP specifies it

  6. edhelas

    another question related to RSM, because it's not clear to me https://xmpp.org/extensions/xep-0059.html#jump, the <index>, it's starting from the end of the results (for MAM the older messages) or the beginning (the most recent) ?

  7. edhelas

    > It does that by including in its request the index of the first item to be returned I seems to be from the beginning

  8. jonasw

    edhelas, the index zero refers to the first item in the full query reuslt

  9. edhelas

    so for MAM it's starting from the oldest messages, and for a Pubsub node starting from the most recent

  10. jonasw

    maybe

  11. goffi

    edhelas: I have no time to check right now (I'm at work), but I think I've experienced some inconsistencies on your blogs posts on ejabberd, and on mine with SàT Pubsub. I wanted to check that before reporting it (I'm not sure who's wrong), but the result is that when I use RSM I get oldest publications from your node.

  12. edhelas

    ok so there's definitly unclear things in the XEPs

  13. goffi

    not sure if it's in the XEP on just an implementation bug, I'll definitely have to check that this week, and I'll check with you then.

  14. goffi

    but right now I can't

  15. edhelas

    to me Pubsub node always list items from the most recent to the oldest, with RSM the <index> is then starting from the most recent (order by created_at desc offset :index)

  16. edhelas

    for MAM it seems (but again I'm not sure) to be the opposite

  17. goffi

    I kind of remember having asked something about that at a XMPP summit also

  18. edhelas

    ejabberd implements RSM for Pubsub but not MAM from what Holger told me

  19. goffi

    SàT Pubsub implements both

  20. MattJ

    What?

  21. MattJ

    RSM is required for MAM, it's not required for pubsub

  22. edhelas

    MattJ sorry, I was talking about the <index> option only

  23. MattJ

    Oh, ok

  24. edhelas

    ejabberd does supports RSM for MAM but not the <index> parameter :)

  25. edhelas

    MattJ but I'm curious to know your point of view regarding this ordering thing between Pubsub items and MAM messages

  26. MattJ

    I'm considering an order parameter in MAM anyway

  27. edhelas

    I think it would be the best :)

  28. goffi

    edhelas: http://jabber.996255.n3.nabble.com/XEP-0313-why-it-is-really-not-a-good-idea-to-use-MAM-with-Pubsub-td36116.html ==> I have mentioned chronological issue there.

  29. MattJ

    Oh, is this about Pubsub + MAM + RSM?

  30. edhelas

    no

  31. MattJ

    Or just Pubsub + RSM?

  32. edhelas

    for me it was just Pubsub + RSM

  33. edhelas

    but MAM + RSM + Pubsub is another issue

  34. MattJ

    Ok, good

  35. goffi

    I do use MAM + RSM + Pubsub

  36. pep.

    The following might be worth watching, to apply to the XMPP community: https://debconf18.debconf.org/talks/9-ignoring-negativity/ https://debconf18.debconf.org/talks/174-ignoring-negativity-followup/ "Dealing with negativity in [large] projects"

  37. pep.

    (or any big enough community, for that matter)