Gajim - 2016-10-13


  1. bot RSS: Feeds for Gajim • Changeset [16168:bf992be0151d]: Fix height of X[] • Changeset [16169:e1251[…] https://trac.gajim.org/changeset/e12519bd0b38d3698e8e09dd4f5d8e8b60a0424a • Changeset [16170:dd9be0a66aef]: Add additional_[] • Changeset [16171:f5b10[…] https://trac.gajim.org/changeset/f5b10ff9e6eeea5e9e8b4ecfa7ddcaac73e79fd4 • Changeset [16172:86264c5ceecd]: Dont leak prese[]
  2. tmolitor arune: the current implementation of message receipts is not user friendly anyway...showing a red cross suggests the message didn't get through wenn gajim really can not know what happened to the message...the red cross should be displayed when an error stanza for that message is received rather than for a message where no receipt is received...
  3. tmolitor lovetox: we should send no-store hints in stanzas containing only chatstates...mam will save those otherwise and fill up server storage with useless stanzas...and synchronizing those useless information with other clients consumes bandwith and cpu power, too....
  4. tmolitor http://xmpp.org/extensions/xep-0313.html#hints
  5. tmolitor holger: sorry...I should have made this more clear: not smacks hibernated sessions are targeted by this module, but those sessions still having open TCP connections which are "resumed" like 5 minutes later or so...having correct timestamps even in this case is very useful, I think (and by the way: setting accurate timestamps when a message passes the server seems a good idea anyway I think :) )
  6. arune tmolitor: thx
  7. Holger tmolitor: Ah, ejabberd adds a delay timestamp on resume as well, but it seems the XEP doesn't mention this case. I guess it should.
  8. Holger tmolitor: With 'passes' I guess you mean when it's initially received by the server? I mean a stanza might go through multiple CSI/resend-on-smacks-timeout/whatever loops. And it might've come from a remote MUC server at some earlier point in time.
  9. Holger And to me it's still not obvious whether you'd want multiple delay tags or just the first one in these cases.
  10. lovetox Holger, are not MAM usually configured the way that it doesnt save messages that have not body and no explicit store hint?
  11. Holger lovetox, tmolitor: Yes, 0313 suggests to only store messages that contain a <body/>, so standalone chat states usually won't be stored.
  12. lovetox tmolitor, you have a wrong configured server :)
  13. lovetox also it would be waste, if just because of MAM, we hat to add a "no store" hint to 10 other xep stanzas
  14. Holger I'm unsure about that as well though. I'm unsure about everything in XMPP :-)
  15. Holger There's stanzas you might want to have in MAM that don't have a <body/>.
  16. lovetox thats why there is a "store" hint :)
  17. lovetox because thats not the usual case, its an exception
  18. Holger Yes, maybe.
  19. lovetox its a message archive, and not a copy all traffic archive :D
  20. Holger If everyone used OMEMO, no message would have a <body/>, wouldn't it?
  21. Holger An then there's people arguing an "all traffic archive" would have its merits.
  22. lovetox yeah but then they could just configure mam that way or not?
  23. lovetox i dont see as the clients duty to decide for the user whats important and what not
  24. linus I do
  25. linus Well, not necessarily to decide for them, but to provide a sensible default and allow them to change it
  26. Holger lovetox: The advantage is that the client (or server) that generates a stanza knows what sort of stanza its generating.
  27. linus Configuring it on the server makes it a lot less flexible. And it's a lot easier to upgrade/add features to clients than to servers.
  28. Holger lovetox: Right now we're trying to do the right thing by specifying (slightly different) heuristics in XEPs such as 0160, 0280, and 0313. They sometimes do the right thing and sometimes not.
  29. lovetox thats very theoretical discussion
  30. lovetox its makes no sense for smacks stanzas or chatstate stanzas
  31. Holger lovetox: In practice, a MUC invite without body would currently go into offline storage, but it would not be carboned nor would it go into MAM.
  32. Holger lovetox: Same is true for 0333 chat markers.
  33. lovetox there is not even a client out there who could display mam chatstate standards in a way that would make sense
  34. Holger Hm?
  35. lovetox how do you display 20 chatstates received in a second from one contact?
  36. lovetox aktive, pause, composing, inactive, pause, composing, aktive ..
  37. Holger I think it makes sense for chat states, for example. And probably also for MUC invites.
  38. lovetox how should that be displayed?
  39. Holger Your idea of adding hints here and there might work. All I'm saying is that I'm unsure.
  40. Holger And currently those hints are usually not added, so we rely on these heuristics, and that can easily go wrong in practice.
  41. lovetox did you mean it makes sense for not saving chatstates
  42. lovetox or the oppsosite?
  43. Holger Sorry. I meant chat markers.
  44. lovetox i agree, chat markers should be saved, though not all
  45. Holger For chat states it might make sense to be able to display the most-recent chat state as well, but yes, this is probably less important.
  46. lovetox oh no, all should be saved
  47. lovetox :)
  48. Holger Maybe :-)
  49. Holger Then you'd also have MUC joins/leaves, for example.
  50. lovetox i think server admins could agree on sensible defaults
  51. lovetox and for everything else there are hints
  52. lovetox but the default should never be, save all
  53. Holger No these things should be clarified on the protocol level.
  54. lovetox and the client has to add to every stanza a hint
  55. lovetox that seems wasteful
  56. lovetox although who has this in his defaults to save chatstates
  57. lovetox i believe prosody
  58. bot RSS: Feeds for Gajim • Ticket #8448 (tuple index out of range File …) created Bug description Traceback (most recent call last): File "_ctypes/callbacks.c", line 314, in 'calling callback function' File "c:\python27\lib\site-packages\pybonjour.py", line 1452, in _callback File "src\common\zeroconf\zeroconf_bonjour.py", line 139, in service_resolved_callback File "src\common\zeroconf\connection_zeroconf.py", line 156, in […] https://trac.gajim.org/ticket/8448
  59. lovetox i never seen it with a ejabbered server
  60. Holger I wasn't aware that Prosody stores them.
  61. lovetox true Store all messages. This is the default.
  62. lovetox from prosody page
  63. lovetox could mean everything that has the "message" tag
  64. lovetox :)
  65. Holger That's about <https://xmpp.org/extensions/xep-0313.html#config-default>, isn't it. The XEP talks about "all messages" there as well :-)
  66. lovetox probably :)
  67. Holger It would BTW be nice if Gajim had a knob for that.</feature-wish> :-)
  68. lovetox for archiving prefs?
  69. Holger Yes.
  70. lovetox we dont
  71. lovetox hm never saw that
  72. Holger Yes this was a feature wish :-)
  73. lovetox i write it to the list
  74. lovetox im 60% finished with message markers
  75. lovetox though i beleive only conversations and gajim use it or?
  76. Holger Chat markers?
  77. lovetox yes thats what i meant :D
  78. Holger Nice :-)
  79. Holger I'm not aware of other clients either.
  80. Holger (Maybe Movim?)
  81. Holger As per 0160, MAM messages would go into offline storage, BTW ;-)
  82. Holger added a <no-store/> hint to ejabberd's MAM messages just last night.
  83. lovetox Holger i dont get it
  84. lovetox what mam messages?
  85. lovetox you mean i send a message to the server and server saves it to mam and to offline storage
  86. Holger You perform a MAM query and go offline while the server responds.
  87. Holger The rest of the MAM response would go into offline storage.
  88. lovetox ahhh
  89. lovetox this could be indeed why we sometimes get double messages :D
  90. Holger Nah the client should discard the MAM response coming from offline based on the 'queryid'.
  91. lovetox true
  92. Holger So I might agree that MAM responses should not go into MAM :-)
  93. Flow Holger: Something that should be added to the MAM XEP?
  94. Holger Flow: Maybe. The alternative is leaving it to the offline storage modules to not store MAM messages.
  95. Holger Flow: I think there's no prior art of adding 0334 hints to other XEPs, is there? But it might be a good idea.
  96. Flow I think we have to distinguish between MAM query result messages stored in 1. offline storage (for later delivery) and 2. stored in another MAM archive
  97. Holger You want neither, do you?
  98. Flow Correct, I'd say we do not want the results to be stored in offline storage or in other (MAM) archives
  99. Holger 2. won't happen as per the standard MAM rules, as the MAM result messages don't have a body.
  100. Flow Holger: OX-IM uses XEP-0334, or at least recommends them
  101. Holger Ah, nice.
  102. Flow I still would want to see the carbons hint to get removed and xep334 used exclusivly
  103. Flow But that's a different topic
  104. Holger +1
  105. Holger Another alternative for the specific case of MAM responses might be to have 0160 state that only messages with a <body/> should be stored if they don't have a 0334 hint. It probably makes sense to have the same rules in 0160 and 0313.
  106. Holger 0160 currently has a special exceptions for chat states which is probably a bit weird these days.
  107. Holger (Special-casing those but not other things, I mean.)
  108. Flow Right
  109. bot RSS: Feeds for Gajim • gajim_appindicator_integration_plugin.patch attached to Ticket #8447 Patch file https://trac.gajim.org/attachment/ticket/8447/gajim_appindicator_integration_plugin.patch
  110. bot RSS: Feeds for Gajim • Ticket #8374 (Once the Gajim roster window is closed, there is no way to display it …) updated so i just found out we have an appindicator plugin. is this not exactly for displaying the icon in ubuntu? so this should solve the problem of the unrecoverable window, because you could open gajim from the systray. https://trac.gajim.org/ticket/8374#comment:13
  111. tmolitor holger: yes, I meant the initial receipt...if the message has no delay tag at this point, my module adds one...but it's still possible that other modules add a second delay tag afterwards, of course...
  112. tmolitor holger, lovetox: well, I didn't configure much on my server regarding mam...my only settings are: default_archive_policy = true max_archive_query_results = 32 archive_expires_after = "never" muc_log_by_default = true muc_log_all_rooms = true
  113. tmolitor on a prosody 0.10 server, so I doubt I configured my server wrong ;)
  114. tmolitor conversations adds a no-store hint for such messages, too and I think it's good practice not to rely on such XEP suggestions anyway...
  115. lovetox they added it probably because of that prosody behaviour
  116. lovetox though its not smart that every client adds 10% more text to a stanza that is send thousands of time per user
  117. lovetox though that seems to be a problem of prosody also with offline messages
  118. lovetox https://prosody.im/issues/issue/619
  119. tmolitor well...as I said...I wouldn't rely on those classifying rules written in the xep as this is very unfelxible...and it wouldn't hurt to add 10% more text to the stanzas because this saves you from receiving all those chatstates later on when querying the mam archive (which is much more text ;) )
  120. lovetox not me, only you :)
  121. lovetox i am on an ejabbered server
  122. lovetox but i agree aslong prosody doesnt fix that issue
  123. lovetox we should probably add this
  124. lovetox but only in the sense that this is a workaround for a prosody issue, and this shouldnt be a statement to falging every stanza with message hints
  125. tmolitor lovetox: well...it would help many people out there because mam queries would be much quicker without those useless chatstates :D
  126. lovetox yeah i will add this soon
  127. lovetox tough also create a prosody issue about it
  128. Link Mauve Note that EME is meant to fix at least part of the issue of “is that a message containing text or not”, without relying only on the presence of a body.
  129. Link Mauve Once accepted, it should be linked in MAM and offline message, at least.
  130. Holger https://jabber.fu-berlin.de/share/holger/hNhmZDVJUChXiy4k/ad-hoc.png
  131. Holger This looks wrong.
  132. arune Well, default branch works much better on Windows than Ubuntu
  133. arune I just tested the daily deb from apt but bookmarks were gone, it didn't connect to my second account and menus were working strange
  134. linus The copyright dates on the about dialog are still 2003-2014
  135. linus x)
  136. bot RSS: Feeds for Gajim Plugins • Ticket #146 (Upload in OMEMO works, but image is not displayed) closed fixed: sadly not, we plan though to ship maybe some of the plugins per default. i know right now its a lot of configuring to get everything working https://trac-plugins.gajim.org/ticket/146#comment:3
  137. lovetox thanks holger, i will fix this if i have time
  138. lovetox linus, that is the bug i was talking about
  139. lovetox i encounter it on ubuntu also but also on windows
  140. lovetox try to reduce the size of your contact list with groups
  141. lovetox its some strange GTK bug where one operation hangs the whole programm
  142. lovetox i had success with little contact lists < 10
  143. lovetox then it works
  144. lovetox like minimize the groups so contacts are not visible
  145. Link Mauve lovetox, did the up to date GTK+ build fix anything for you?
  146. lovetox hm, we were talking about gtk but i dont recall how i can update ubuntu to up to date GTK
  147. Link Mauve Ah no, I meant for Windows that other day.
  148. Link Mauve For Ubuntu you just have to update to the next release.
  149. lovetox 16.10 ?
  150. lovetox hm i could try that
  151. lovetox im on 16.04