XSF Discussion - 2019-05-03


  1. Arc

    *cheers* I finally got it to work, Prosody now reports whether a connection is bosh or websocket in the admin telnet

  2. 靳龙

  3. 靳龙

    hello

  4. flow

    > dwd> If references is counting codepoints (or even bytes) should we be normalizing the text first? Yes, I think you have too

  5. dwd

    flow, Right - if anything *can* normalize, everything *must* normalize - seems the general rule.

  6. flow

    dwd, what's the scope of your "general rule"? It appears to me that you have to normlalize in order to identity codepoints interoperable, but usually, e.g., if you just want to exchange <body/>, you don't have to (and I believe most implementations do not)

  7. Arc

    hey Zash, I sent a bundle to the -dev list. Looking for what to work on next now that I'm getting a handle on lua

  8. dwd

    flow, Well, if you're reliant on counting codepoints or comparing. Normally we don't care for human text.

  9. jonas’

    flow, I think normalizing will be bad

  10. jonas’

    normalizing before sending makes sense, but when receiving, you’ll have to work with exactly the codepoints used in the message

  11. jonas’

    otherwise we’d have to specify which unicode version to use with References so that everyone normalizes in the same way

  12. flow

    jonas’, I think both send and recipient have to normalize if something xep message markup is used

  13. jonas’

    why?

  14. flow

    nothing gurantees that the way the sender emitted the UTF-8 will be preserved

  15. jonas’

    err..... what?

  16. jonas’

    you are saying that there is (aside from intentional modification) no guarantee that the sequence of codepoints I put in an XML element in XMPP arrives as exact that sequence at the receiver?

  17. Kev

    That's certainly true in some cases, yes.

  18. Kev

    e.g. where the element has attributes that are JIDs, an entity might very reasonably normalise en route.

  19. jonas’

    Kev, and <body/>?

  20. Kev

    It wouldn't surprise me, at least.

  21. dwd

    jonas’, Well, while I don't know of any implementations that do normalize, there's absolutely nothing in the specifications that says implementations MUST NOT normalize.

  22. jonas’

    in that case, normalisation needs indeed to be specified, both on the sending (before calculating the offsets) and the receiving side (before matching the offsets)

  23. flow

    jonas’, I thank there is nothing that guarantees that this will *not* happen, so I better be safe than sorry here

  24. flow

    *think

  25. MattJ

    I'm inclined to agree

  26. jubalh

    I suppose ignoring of users in MUCs and of jids that write to you 1:1 is done by clients. one doesnt tell the server to "dont deliver messages from X", right?

  27. Ge0rG

    jubalh: https://xmpp.org/extensions/xep-0191.html for 1:1

  28. jubalh

    thank you Ge0rG

  29. dwd

    Thinking about it, anytime that data is served from a database rather than simply switched through, probably has been normalized.

  30. Ge0rG

    except when the database is a legacy mysql?

  31. jonas’

    dwd, right, archives

  32. dwd

    jonas’, Exactly.

  33. Niraj

    mimi89999

  34. mimi89999

    Yes?

  35. MattJ

    We need to create some kind of inline syntax that allows you to apply arbitrary semantics to certain bits of text

  36. pep.

    That'd be great, but we need to make sure it's not too verbose

  37. edhelas

    I need to mark those ideas down

  38. MattJ

    I was actually thinking of marking them up

  39. Ge0rG

    I prefer to mark them right.

  40. edhelas

    better use a marker for that

  41. mimi89999

    What are you inventing?

  42. MattJ

    Some generic markup language

  43. mimi89999

    What will it be used for?

  44. MattJ

    The stuff listed here: https://xmpp.org/extensions/xep-0372.html#usecases

  45. MattJ

    and much more!

  46. pep.

    I think we need something that looks like a language we all know, but not too close either

  47. pep.

    Javascript seems like a language we would all know. I'm sure that would make a good markup language

  48. pdurbin

    I sure hope someone approves the message I sent last night to https://mail.jabber.org/mailman/listinfo/juser

  49. pep.

    Does that need approval?

  50. MattJ

    juser still exists, wow

  51. pdurbin

    "Your message to JUser awaits moderator approval"

  52. mimi89999

    `<body>But, soft! what light through yonder window breaks? It is the east, and <reference type='mention' uri='xmpp:juliet@capulet.lit'>Juliet</reference> is the sun.</body>`

  53. mimi89999

    Why is this bad?

  54. pdurbin

    Does anyone here know a moderator for that list?

  55. Zash

    mimi89999: body is defined as having only text content

  56. mimi89999

    s***

  57. mimi89999

    How can we be sue that the offset will always be the same?

  58. pep.

    That's what's being discussed

  59. pdurbin

    I guess I'll wait. Maybe a week. Then I'll ask how to get my post approved.

  60. pep.

    I'm not sure who has access, board members?

  61. MattJ

    No

  62. pep.

    iteam?

  63. pep.

    I think Seve had access to some of it?

  64. MattJ

    Some subset of iteam

  65. pdurbin

    Thanks. Is there a URL for iteam? I can't find it.

  66. MattJ

    I don't think so

  67. MattJ

    There is a MUC, but I think everyone in it is also in here

  68. Zash

    https://xmpp.org/about/xsf/infrastructure-team ?

  69. Zash

    all outdated of course

  70. MattJ

    Someone should file an issue about that

  71. MattJ

    (only kidding, please don't)

  72. Zash

    There's an issue tracker for that?

  73. Zash

    I'll file an issue about sending you an email about filing an issue about fixing that.

  74. Ge0rG

    pdurbin: the easiest and fastest way is to subscribe, then re-send

  75. pdurbin

    Thanks for the URL. I'm not sure why resending would help.

  76. MattJ

    I think Ge0rG is suggesting that it only requires manual approval if you're not subscribed to the list when you send

  77. MattJ

    which is the case with most of our lists I think, so probably that one too

  78. pdurbin

    I was already subscribed when I tried to post my message. I guess I'll keep waiting. :(