Modern XMPP project discussion - 2021-08-25


  1. Megaguy32

    Matrix vs XMPP?

  2. MattJ

    XMPP

  3. MattJ

    Next question? 😄

  4. emus

    MattJ: I really really dont understand why you have this opinion here? This sounds out of the blue to me! 😉

  5. raucao

    it would actually be nice to have a page explaining why

  6. raucao

    because this is obviously a frequently asked question

  7. raucao

    and personally, it takes me like 10 minutes to explain it every time someone asks

  8. emus

    But you mean why XMPP or why people XMPP over Matrix?

  9. Holger

    raucao: > it takes me like 10 minutes to explain it every time someone asks My default response is, "from the user perspective, Matrix is better at replacing public/team chat (Slack) while the popular XMPP clients are better at replacing private chat/groups (WhatsApp)".

  10. Holger

    My own reasonings are about the protocol structure and how it's maintained, but users won't care.

  11. raucao

    i'm working on the public/team chat angle :)

  12. raucao

    and i do thing the protocol structure and how it's maintained is actually what people are interested in also, when asking this question

  13. raucao

    at least the developer crowd

  14. Holger

    Well, not sure about our selling points over Matrix there, TBH. Some people criticize Matrix' distributed database design for privacy and/or scalability reasons, others complain about how the specs + reference implementations are maintained by a single team plus how stuff is de facto centralized at matrix.org (identity service and whatnot), but ... users won't case.

  15. raucao

    i know many who do. there's no "the user" who you can just speak for

  16. Holger

    raucao: Yeah, some people might care of course. Most devs will tell you how straightforward a proper REST API is though.

  17. Holger

    raucao: Right. The vast majority of the target audience _I_ am interested in doesn't care about protocols and stuff, though.

  18. raucao

    it's certainly convenient. see e.g. JMAP for how it makes email app development easier

  19. Holger

    Yup.

  20. jonas’

    *triggered*

  21. raucao

    lol

  22. Zash

    Oh no

  23. raucao

    success :)

  24. Holger

    For those who do care, XMPP's modularity (proper IETF core that's not affected by adding new XEPs) is my #1 selling point.

  25. raucao

    > doesn't care about protocols and stuff, though not in the technical sense perhaps, but in more subtle ways a lot also do. and with matrix, you have to learn matrix addresses e.g.

  26. raucao

    that's a protocol thing

  27. raucao

    same as a JID is

  28. raucao

    in decentralized systems you cannot hide some of these things

  29. Holger

    Heh, was going to respond that the Matrix ID is supposed to be hidden, in theory.

  30. raucao

    > For those who do care, XMPP's modularity (proper IETF core that's not affected by adding new XEPs) is my #1 selling point. agree. but it's also a major criticism for client support disparity

  31. Holger

    Yeah. I strongly disagree with this criticism :-)

  32. Holger

    If clients are missing support for important features, the reason is missing manpower, not spec modularity.

  33. raucao

    it sounds like you have some strong opinions on this subject, so it would be especially interesting to read a slightly longer-form explanation of your viewpoints

  34. Holger

    Most Matrix clients are missing important features.

  35. raucao

    to a lot of people i guess

  36. raucao

    > Most Matrix clients are missing important features. is this referring to element being the only full-featured client, or does that include element?

  37. Holger

    The former.

  38. raucao

    right. same goes for servers then

  39. Zash

    Since what's considered an important feature varies from person to person, I'm sure you can find someone who thinks element is missing the most critical feature ever. Or does that only happen to XMPP software?

  40. Holger

    Zash: Element did miss the most critical feature ever until a few weeks ago, audio messages!!!

  41. raucao

    only emojis, stickers, and threads are critical in 2021 🤡️

  42. Holger

    Zash: Now that they have it, Matrix has superseded XMPP. (No kidding, I've read that just yesterday.)

  43. Zash

    Of course

  44. Zash

    I guess we should pack it up then since we clearly lost and everything is pointless

  45. MattJ

    In other news, being able to link the Snikket iOS developer to friendly docs on Modern XMPP instead of having to vaguely explain how things should be implemented... is really nice :)

  46. jonas’

    :)

  47. MattJ

    (currently https://docs.modernxmpp.org/client/design/#names )

  48. raucao

    nice!

  49. southerntofu

    pretty cool

  50. southerntofu

    following the discusions we had in jdev and joinjabber on room name/description/topic it would be nice to have some guidelines about that in there as well

  51. MattJ

    Yeah, I thought the same when I read them. PRs welcome :) (probably to https://docs.modernxmpp.org/client/groupchat/ )

  52. southerntofu

    yeah let's open an issue so people can contribute feedback on how their specific client does UX on that

  53. raucao

    oh wow, i hadn't even realized that there was a subject in addition to the description

  54. raucao

    (only just reading the spec)

  55. raucao

    i like the idea of using the subject as a kind of pinned announcement message in UIs

  56. raucao

    vaguely remember some web client having done that with IRC channel topics

  57. Zash

    The subject is essentially a pinned message, so that makes sense.

  58. MattJ

    Name: "Modern XMPP discussions" Description: "This is where we discuss the Modern XMPP project" Subject: "Our next sprint is coming up soon! Register here: https://..."

  59. raucao

    i'd still prefer names as short as possible, like e.g. "Modern XMPP". perhaps ideally capped at ~15 chars (as best practice, not hard limit)

  60. raucao

    this way they're not cut off in most clients