Gajim - 2014-01-20


  1. bot RSS: Feeds for Gajim • Ticket #7599 (cipher specification cleanup) updated Milestone 0.16 https://trac.gajim.org/ticket/7599#comment:3
  2. crow asterix: yes if there is ( D ) then it is debugging enabled. without this '--loglevel .potr=DEBUG' gajim.log is empty but still file is created. i would expect if '--loglevel .potr=DEBUG' then to create log in conf directory '-p portable -c ../DataDir '
  3. bot RSS: Feeds for Gajim • Ticket #7628 (Support SHA-2 for SSL certificate fingerprints) created problem Gajim checks and displays only SHA-1 for SSL certificate fingerprints. There have been reported weaknesses in SHA-1. Upgrade to support SHA-2. analysis In 2005, cryptanalysts found attacks on SHA-1 suggesting that the algorithm might not be secure enough for ongoing use. ​https://www.schneier.com/blog/archives/2005/02/cryptanalysi[…] https://trac.gajim.org/ticket/7628
  4. Dave Cridland Afternoon, folks.
  5. Dave Cridland I was wondering if Gajim supports MAM - it's not in the list, but I was vaguely hoping somebody has slipped it in and forgotten to update the XEP list. :-)
  6. Asterix I have a patch for it, but there is a blocking thing for the moment
  7. Asterix when we wget the logs from a JID, we need to discoverit to know if it's a real JID or a MUC JID. So it's not really usable for the moment
  8. Asterix and doing this disco is quite complexe as it require some asynch handling
  9. Asterix azerttyu : could you stop leaving / joining please?
  10. Asterix Dave Cridland : ^
  11. Dave Cridland Asterix, Thanks. Is your patch anywhere handy?
  12. Maranda Asterix, that shouldn't be so horribly blocking as long as it doesn't screw up the Archive UI, if it's for muc private messages that's easily filtered already if you set prefs to only log messages from contacts.
  13. Maranda That's more problematic for Message Carbons actually.
  14. Asterix Dave Cridland: I have my patch localy, not even sure it still applies to trunk
  15. Asterix Maranda: User (me included) want to also log PMs ...
  16. Asterix for the moment it's hard to make a distinction. we need to disco bare jid to know if it's a room (and hope the the MUC server is not down ...)
  17. Maranda Asterix, not really hard if you set prefs to log for all, they'll be logged (as long as the server doesn't filter 'em...)
  18. Maranda if you retrieve the archives per window prefix (Like in Jappix) it's easy to make queries depending on the jid
  19. Maranda (aka if I send you a pm it'll be gajim@conference.gajim.org/Asterix)
  20. Asterix "per window prefix"?
  21. Maranda If I send you a pm here and open the related tab, Gajim is aware I'm having a conversation with gajim@conference.gajim.org/Asterix
  22. Maranda If I send you a pm here and open the related tab, Gajim is aware I'm having a conversation with gajim@conference.gajim.org/Asterix isn't it ^^?
  23. Maranda I dunno how MAM is implemented UI wise in Gajim but I suppose it should show within the pm tab? (or the related History button?)
  24. Asterix Gajim will retrieve logs on startup to sync them with local logs
  25. Maranda ALL of the logs?
  26. Asterix since last connection yes
  27. Maranda I hope it split requests using RSM or that'll end servers bouncing requests (At least Metronome will if an unbound query has more then 50 results)
  28. Maranda I hope it splits requests using RSM or that'll end servers bouncing requests (At least Metronome will if an unbound query has more then 50 results by default)
  29. Asterix we handle RSM
  30. Maranda Asterix, well if it just syncs with the local history archive, you should have the information a private conversation was ongoing with a muc contact already.
  31. Maranda or should
  32. Maranda don't think you need to actually do any discoing in that occasion
  33. Asterix no if it was on another machine
  34. Maranda Maybe it's a stupid question but why do you need to assert it's a MUC pm anyways? Doesn't
  35. Maranda Gajim just searches per JID?
  36. Maranda I see the possible reasons for perfect sync'ing
  37. Maranda But just be aware that if you just install Gajim on a new machine and the user has a very large archive retrieving it all on the first connection may not be pretty, bbl :)
  38. Asterix Gajim stores with a different ID PMs and normal chats
  39. Maranda "normal chats" as groupchats?
  40. Asterix no, type="chat"
  41. Maranda MUC PMs have the same type, if you're going to need to assert what is what on the initial sync that will be eiree.
  42. Maranda If I may, it could be better to just assume they're normal chats until the user doesn't join the MUC in that case change it
  43. Asterix Gajim needs to do the difference yes
  44. Maranda that will void the hassle of just discoing
  45. Asterix so we have wrong info in logs local logs until we chat again with the same occupant(which may never happen) ... too bad
  46. Maranda Asterix, I know it's not perfect but this whole thing with discoing is potentially much worse and could end in the same way anyways
  47. Maranda (say the muc is down... the room gets destroyed etc)
  48. Asterix those probabilities are quite low
  49. Maranda As much as not joining a room again...
  50. Maranda after all you need to send a disco info to the bare jid anyways :)
  51. Maranda not the full jid
  52. Maranda if the chat is with gajim@conference.gajim.org/Asterix and I join gajim@conference.gajim.org ^^
  53. Asterix yes but even if we don't join the room again, we may want to browse local history
  54. Asterix which will be wrong
  55. Maranda How that will prevent it? And how's that wrong because "gajim@conference.gajim.org" will figure between entries I suppose?
  56. Maranda It's okay to send a disco info when we filter result for the jid e.g. gajim@conference.gajim.org but not during sync'ing that's all I'm saying
  57. Maranda It's okay to send a disco request when we filter result for the jid e.g. gajim@conference.gajim.org but not during sync'ing that's all I'm saying ^^
  58. Maranda (filter as in Open the History and search)
  59. Dave Cridland Maranda, You can search history when you're offline. It's very useful.
  60. Maranda Dave Cridland, yes but sync'ing the whole archive is already something which gives me creeps, sending discoes during that x2
  61. Dave Cridland Maranda, You get the creeps easily. :-)
  62. Dave Cridland Maranda, I don't understand what the problem with doing so is. I understand it's tricky to implement on the client-side, which I have to admit makes me consider giving it a crack - been a while since I tinkered with Gajim - but I don't see what the problem could be from the serer-side.
  63. Maranda Dave Cridland, the problem is that Gajim has on startup/login delay issues already mainly. Maybe I'm naive but I don't think sync'ing (very large) archives will help expecially for those nice "first time occasions".
  64. Dave Cridland Oh, I wouldn't want to block on it.
  65. Dave Cridland I didn't realise Gajim had any startup delay, actually; I always found it fairly quick against M-Link with all the knobs turned on (where M-Link caches presence and things, so eliminating the usual XMPP presence delays etc). I know it syncs a lot of data, but I'm not aware of this blocking.
  66. Dave Cridland FWIW, I used to, when I first joined Council, hold many Council meetings while on the train over GPRS using Gajim. Didn't cause a problem there, either.
  67. Maranda Dave Cridland, there're (as long as not fixed) still some issues with very large rosters (which is more GTK's fault afair)
  68. Dave Cridland Really? How large does the roster need to be to show the problems?
  69. Maranda I seen issues with 400+, it'll take a while the first time but it copes it way better then it used to. But it's still very slow compared to say SwiftIM
  70. Dave Cridland Heh. Kev spent several weeks optimizing huge rosters in order to persuade PSA to use it, to be fair.
  71. Maranda And honestly this whole "I need to know what is what" in both MAM and Carbons is annoying, I know Kev opinion is to just "go on and disco" but I think this should be addressed at protocol level with "better flagging" (either by making a new message type or something else), rather then adding complexity on one side or another.
  72. Asterix that would really simplify clients !!
  73. Dave Cridland Sounds reasonable. Why not sketch out a proposal on the list?
  74. xmt Which farstream plugins/transmitters does gajim need?
  75. Asterix fsrtpconference
  76. xmt So you don't need the raw or msn plugins?
  77. Asterix no reason to have msn plugin
  78. Asterix raw I don't know what that is
  79. xmt This is all plugins: fsrawconference fsrtpconference fsmsnconference fsfunnel fsrtcpfilter fsvideoanyrate
  80. xmt All transmitters: rawudp multicast nice shm
  81. xmt Not all of them will build so I wanna make sure I get the ones we need.
  82. Asterix nice is used
  83. Maranda Dave Cridland, list as standards?
  84. Dave Cridland Yes, I think there's so few MAM/Carbon implementations that some developer feedback would be great.
  85. Maranda Actually only Jappix has a MAM implementation iirc
  86. Maranda client wise ^^
  87. Dave Cridland Maranda, No, OTalk does as well.
  88. Maranda googles.
  89. Dave Cridland Maranda, otalk.im - &Yet's skype-onna-web clone.
  90. Maranda found. ^^
  91. Maranda Dave Cridland, but more then MAM/Carbons this should go for either Core/MUC XEP
  92. Dave Cridland Oh, marking PMs as some new type? I don't think that'd be possible.
  93. Maranda Dave Cridland, but more then MAM/Carbons this should go for either IM/MUC XEP
  94. Maranda Dave Cridland, reason? That'd be the less expensive and simplest solution
  95. Maranda of course you can mark PMs in other ways
  96. Dave Cridland I don't think it'd be possible to add a new enumeration value to the type attribute of a stanza.
  97. Dave Cridland Servers would block them.
  98. Maranda (just by changing the MUC xep)
  99. Maranda But that'd have a good number of issues as well
  100. Maranda This was discussed in jdev a few weeks ago.
  101. Maranda Dave Cridland, a good idea could be using Message Processing Hints but then we incurr in the same issue, people implementing (expecially in clients)
  102. Maranda So I'm not sure what kind of feedback I could propose on this.
  103. Maranda And if there isn't a wide implementation the discoing clientwise is still the most reliable, brb.
  104. Maranda And if there isn't a wide implementation then discoing clientwise is still the most reliable, brb.
  105. Dave Cridland Right, but the server could do the disco as needed.
  106. Maranda As a server developer that adds scenarios which I wouldn't like honestly... How should it be done? The server itself should send out a disco info queries to the bare jids and wait for an answer to tag out the results? What happens if there's no answers, also should we wait for the answers?
  107. Dave Cridland Servers send out a load of disco queries anyway. It'd just mean they could tag things in MAM and Carbon.
  108. Dave Cridland As for what happens if they never reply - there's a lot of cases like that anyway, such as PEP.
  109. Maranda Well this is a bit different then PEP... MAM queries results can have multiple jids to assert. PEP is very limited, and as Kev rightly pointed out what do we do then keep a cache of every entity features. I'm not sure but at this rate having a session stored map of joined mucs (obtained from directed presences) sounds saner.
  110. Maranda (That's what I currently do to prevent the mess caused by muc pms being carboned)