jdev - 2020-05-24


  1. lovetox

    hm if im joined a muc, and change my show, do i have to send the presence also directed at the muc? or is it enough to send it for my account

  2. lovetox

    i know i dont have to if i go offline (unavailable)

  3. MattJ

    lovetox: sending it to your account doesn't update the MUC, so yes, you probably want to send those separately

  4. lovetox

    and this is expected? when i send a unavailable presence, all mucs get updated by the server

  5. lovetox

    but not if i only change show and status, because one would probably send different show and status to those mucs?

  6. MattJ

    I don't think the rules were necessarily created with MUCs in mind

  7. MattJ

    The idea is that you can send directed presence to some entities, separate from your account presence. The only special thing is that if you go offline, the server will "cancel" any directed presence for you

  8. Link Mauve

    I wanted to add a way to change this rule, negociated by the client with a supported server, and have any non-directed presence also update directed presences.

  9. Link Mauve

    But never did so yet.

  10. edhelas

    hey, do you know clients that are implementing and using 0012 ? (https://xmpp.org/extensions/xep-0012.html)

  11. mathieui

    we have an (apparently broken) implementation in poezio

  12. edhelas

    i'm wondering how it is actually implemented

  13. edhelas

    like, when are you sending those requests, and how are they refreshed

  14. Link Mauve

    mathieui, didn’t we switch to XEP-0319?

  15. lovetox

    edhelas, for example if you open a information dialog of a contact

  16. lovetox

    or if you hover for some seconds over a tooltip of a contact

  17. lovetox

    but 0319 is better suited for that obviously

  18. edhelas

    yes, i'm working with that one :)

  19. pep.

    We talked about removing it from poezio at all

  20. pep.

    0012

  21. lovetox

    in Gajim i answer to 0012 requests

  22. lovetox

    but i never request it

  23. pep.

    What do people say about the 402 / autojoin thread btw? More specifically my point on <extension/>, if any exists, don't randomly remove the bookmark

  24. pep.

    I can PR that to 402

  25. pep.

    It might be "obvious" for some, but the XEP only mentions preserving when editing

  26. pep.

    > Clients MUST preserve these (particularly preserving unknown elements) when editing items.

  27. pep.

    And it doesn't say anything about known elements

  28. flow

    pep., IIRC I suggested, in a thread where dwd was also parcipating, special container elements: one for client generated extensions, and one for server generated extensions

  29. flow

    this would also solve the cases where clients are unable to replay the elements in the client-generated extensions

  30. pep.

    That seems orthogonal to my case(?)

  31. flow

    potentially, maybe I misinterpreted "my point on <extension/>"

  32. pep.

    https://mail.jabber.org/pipermail/standards/2020-May/037473.html and https://mail.jabber.org/pipermail/standards/2020-May/037471.html for more context

  33. flow

    well <extension/> appears to be such a container for client generated extension elements

  34. flow

    and my idea was to extend the protocl so that if a client does not explicitly include <extension/> on publish, the previous/current content of that container is preserved

  35. pep.

    I'm not sure what definition you're referring to but it's not really defined in 402

  36. flow

    exactly, and IIRC there was a previous thread on standards@ how to extend things like bookmarks with custom annotations

  37. flow

    exactly, and IIRC there was a previous thread on standards@ how to extend things like bookmarks with custom annotations/(meta-)data

  38. pep.

    My point is not about edition but removal. dunno if these two can/should be mixed?

  39. pep.

    Also the fact that extensions seems only to be defined in terms of clients not knowing about stuff inside :p

  40. flow

    https://mail.jabber.org/pipermail/standards/2019-November/036640.html

  41. pep.

    yes I remember this

  42. pep.

    And I know that's where it comes from :x

  43. flow

    pep., I am sorry, then I probably did not understand what "More specifically my point on <extension/>, if any exists, don't randomly remove the bookmark" is asking

  44. pep.

    Asking if anybody is against a PR specifying <extensions/> a bit more in this regard

  45. pep.

    402 has no schema. yay

  46. flow

    pep., what would that additional specification entail?

  47. flow

    I also usally encourage people to just present patches, be it for specs or source code, as those are easier to discuss ;)

  48. pep.

    Something along the lines of "If there is an extension that you didn't add yourself, don't automatically remove the entry (that is, without explicit user input)."

  49. flow

    Isn't that what " Clients MUST preserve these [<extensions/>]" already does?

  50. pep.

    What about the "when editing items"?

  51. pep.

    Because I've heard a few client devs ready to remove these items witout blinking

  52. pep.

    Automatically when the user leaves a room

  53. flow

    that sounds like non-compliant behavior

  54. pep.

    So many things a person can infer from half a specified sentence..

  55. flow

    as the client clearly edited the item while not preserving things

  56. pep.

    Can we not just add a bit more text in there?

  57. flow

    well sure

  58. flow

    clarifications are always good, and it appears your clarification is within the spirit of the xep

  59. flow

    it's just *very* hard to write specifications that are not open to different interpretations

  60. lovetox

    pep., i think you misunderstanding the XEP

  61. pep.

    how?

  62. lovetox

    because you talk about client devs removing items

  63. lovetox

    of course a client can remove a item if it understand what it does

  64. lovetox

    this is not an eternal archive of xml elements

  65. pep.

    Well I'll let the user be the judge

  66. lovetox

    they have a purpose, one client puts an extension in, and another client can remove it

  67. lovetox

    maybe thats the problem, you think this is some kind of User added thing?

  68. pep.

    lovetox, sure, that is not what I'm saying. I just don't want clients randomly removing items when leaving a room or sth (because the whole point of all our "bookmark" XEPs is to be client sync mechanisms)

  69. pep.

    (and I stand by this statement)

  70. lovetox

    random? i dont know what you talking about

  71. pep.

    That's why I'm not comfortable hacking profiles on top of 402

  72. lovetox

    if i unset the autojoin flag on a bookmark is this random for you?

  73. lovetox

    so why would it if i modify an element in the extension node?

  74. pep.

    Ok remove the "random"

  75. pep.

    To me, a user leaving a room doesn't justify removing the item

  76. pep.

    I want the dumb list of bookmarks to be separated from the autojoin mechanism

  77. lovetox

    but why? so we have to query more for the same functionality?

  78. pep.

    hmm?

  79. lovetox

    if you want that make a proposal

  80. lovetox

    but extension elements can be added by clients, and can of course be removed by them, this has nothing to do with profiles

  81. pep.

    Well if I could modify 402 right now to be a dumb list of bookmarks and a separate sync mechanism I would do it. But it's draft and I'm sure I'll get quite a few grumps

  82. pep.

    (so it's not "yet another bookmarks xep")

  83. lovetox

    so you dont want the bookmarks in PEP?

  84. pep.

    No that's not the point

  85. pep.

    I want real bookmarks

  86. pep.

    And maybe a separate syncing mechanism :)

  87. lovetox

    do you think when i talked about removing a item i talked about a bookmark?

  88. pep.

    what else?

  89. lovetox

    nobody removes bookmarks on leave

  90. lovetox

    we set the autojoin flag to false

  91. lovetox

    but dont remove the bookmark

  92. flow

    which sounds sensible IMHO

  93. lovetox

    if you have now profiles, the autojoin flag is not enoguh anymore

  94. pep.

    Well apparently some do, (or would), and I'm just trying to prevent that

  95. lovetox

    so you add a <profile name=mobil> into the extension

  96. lovetox

    which means, all clients that have this profile treat this like a autojoin flag

  97. pep.

    I'm fine with switching autojoin off. It still irks me that 402 is just another client sync mechanism but I can live with it if I have guarantees

  98. lovetox

    if such a client now leaves the MUC, he removes that profile node

  99. lovetox

    not the whole bookmark

  100. lovetox

    so other clients that have the profile dont join the MUC anymore

  101. pep.

    yeah

  102. lovetox

    but client that dont have the profile can still honor the not modified autojoin flag

  103. lovetox

    and that was just an idea out of the top of my head, if you have better proposal please come forward

  104. pep.

    yeah your proposal seemed fine, with the small detail I added in

  105. lovetox

    but i want to state, nobody wants to delete bookmarks :)

  106. lovetox

    but pep. everything you but in PEP is a sync thing or not?

  107. lovetox

    thats why PEP exists to sync shit

  108. pep.

    I guess you're missing my point

  109. pep.

    PEP is fine

  110. lovetox

    yeah i also think that, you want a "real" bookmark list?

  111. pep.

    Just a dumb bookmark list. without autojoin.

  112. lovetox

    dont know what this means, nobody deletes automatically bookmarks, so why would that not be a real bookmark list?

  113. lovetox

    why without autojoin? what do you gain if this is not in the XEP?

  114. lovetox

    can you not simply ignore it?

  115. pep.

    no because it changes the meaning of the XEP

  116. pep.

    I can ignore the huge monster in front of me and live my life as usual, that doesn't remove the huge monster in front of me (ish)

  117. pep.

    Now, as I said above, I'm happy with what you suggested in the thread. It doesn't change the underlying semantics to me, but that works

  118. Link Mauve

    There was a funny manga about that. :D

  119. Link Mauve

    Where a girl who can see monsters tries to ignore them but is scared by them anyway.

  120. lovetox

    pep., i dont even care about that, i dont even own a smartphone, and im joined 10 MUCs, i dont need profiles

  121. lovetox

    i just out of interest want to understand what you think :)

  122. pep.

    I don't especially have the same usage of phones or desktop / console clients

  123. lovetox

    i see autojoin as something you can ignore, other clients will modify it, but it will not change your bookmark list

  124. pep.

    lovetox, I would still have use for autojoin, I just don't think it fits in bookmarks and should be separate

  125. pep.

    Then you can have pointers to these bookmarks etc. and do howmanyxeps you want

  126. lovetox

    but that would be a pain in the ass with race conditions

  127. lovetox

    a bookmark list in pep, and then a autojoin list in pep

  128. lovetox

    i really dont want to think about it ^

  129. lovetox

    i really dont want to think about it ^^

  130. pep.

    Maybe the whole thing is flawed :x

  131. lovetox

    no it works fine, just be fine with it :)

  132. pep.

    I'll "just" ignore this message :p

  133. pep.

    I can't think of a proper name otherwise I'd propose something else for 402

  134. pep.

    To reflect the semantics

  135. lovetox

    great

  136. lovetox

    now is sleep time, good night

  137. pep.

    :)