jdev - 2020-03-14


  1. lovetox

    im getting this error from an ejabberd on entering the wrong password

  2. lovetox

    The response provided by the client doesn't match the one we calculated

  3. lovetox

    seems to me like a awfully developer orientied text message

  4. lovetox

    hm no sorry its prosody of course, ejabberd usually has pretty good error text

  5. lovetox

    so is there consense what the text field of an error actually should contain?

  6. lovetox

    i know from history some server devs treat this like a debug string

  7. defanor

    I think providing such a message at all may be risky and unnecessary: as the RFC mentions, "In order to discourage directory harvest attacks, no differentiation is made between incorrect credentials and a nonexistent username.", while this points at incorrect credentials. Although even if it wasn't for a textual message, the number of challenges (with SCRAM, for instance) would give it up.

  8. lovetox

    i think you misinterpret that security recommendation

  9. lovetox

    its not recommend to send a message like "Password wrong" which means Username is correct, and so you can harvest users

  10. lovetox

    but it does not mean you cant send a message like "Incorrect Credentials" or "Username or Password wrong"

  11. lovetox

    which is exactly what every service i encountered does

  12. lovetox

    ha !

  13. lovetox

    and prosody handles that wrong, its possible to harvest usernames with the prosody sasl impl

  14. defanor

    Indeed, I was talking about Prosody's message. A non-informative textual message should be fine.

  15. lovetox

    it aborts after <auth> if the username is no known

  16. lovetox

    it aborts after <auth> if the username is not known

  17. lovetox

    if it knows the username, it sends a challenge

  18. flow

    lovetox, i'd say that <text/> should be user exposable, while encouraging impls to put detailed debug messages into something like <debug-text/>

  19. lovetox

    flow iq allows only one child or not?

  20. flow

    so? subchild

  21. lovetox

    yeah k :)

  22. lovetox

    i also think it should be user exposable

  23. lovetox

    that does not mean that every user in the world must understand what that message means

  24. flow

    Only very few places in xmpp disallow stuffing another extension element somewhere

  25. lovetox

    but it should be something that a user can easily google or ask for

  26. flow

    yep

  27. flow

    but to not allienate the user, making to text not to technical may also be a good advise

  28. lovetox

    the standard is weird

  29. lovetox

    https://tools.ietf.org/html/rfc6120#section-6.5.10

  30. lovetox

    so not-authorized is allowed to be sent in response to <auth> and <response>

  31. lovetox

    if i send it in repsonse to <auth> its evident that the username is not existent

  32. lovetox

    because thats the only thing i send in a auth

  33. flow

    but you don't have to send it right after auth

  34. flow

    also, it is sasl mechanism specific what is send in auth

  35. lovetox

    yeah but i doubt any sever impl, now does a random challenge

  36. lovetox

    for a non existent user

  37. lovetox

    to fake it

  38. lovetox

    yeah im talking about PLAIN, and SCRAm

  39. lovetox

    scram also puts a bit more in the auth, but nothing that would trigger a not-autorized if i do it wrong

  40. lovetox

    hm after auth was successful

  41. lovetox

    and there is a stream restart, is there a order of events

  42. lovetox

    like must the server first send the new stream header

  43. lovetox

    or does it not matter and i can fire it even if i didnt yet receive the server stream header

  44. raucao

    hi. i requested to get an account for the wiki a while back and was told to wait for someone with admin privileges...

  45. raucao

    just wanted to check in again

  46. pep.

    raucao, hey, you should stick around. Not everybody with rights is there everywhere, and they need your email iirc

  47. pep.

    Ge0rG, Guus ^

  48. raucao

    oh, could it that there's no message archive for this room?

  49. pep.

    There is yeah but I don't know if everybody reads everything :)

  50. pep.

    (I think there is?)

  51. raucao

    looks like i have a hole in my history from when i wasn't joined

  52. raucao

    unless there weren't any messages for 7 days

  53. raucao

    ah, seeing the log link in the topic now. never mind

  54. raucao

    yeah, looks like MAM is not working for me in this room

  55. raucao

    (dino.im)

  56. pep.

    dino doesn't support muc mam

  57. raucao

    are you sure about that? i'm using it daily in many rooms

  58. raucao

    what would even be the difference between muc mam and just mam?

  59. raucao

    is it a different XEP than https://xmpp.org/extensions/xep-0313.html ?

  60. pep.

    muc mam is just mam on muc :)

  61. pep.

    And yes dino doesn't do that

  62. pep.

    it does normal muc history

  63. raucao

    certainly does not do "normal muc history". have that turned off in my rooms and using MAM

  64. raucao

    and dino works with it

  65. raucao

    unless i missed it not working for the last 12 months or so

  66. raucao

    normal history is just receiving a bunch of messages upon announcing presence, correct?

  67. pep.

    normal muc history is probably provided by your MAM module

  68. raucao

    https://github.com/dino/dino/wiki/Supported-XEPs

  69. pep.

    MUC join is you sending a join presence, you receiving all other occupants' presences, your receive a self presence, then you receive historical messages if there is any, then subject, then live messages

  70. pep.

    Ask in dino@ if you want

  71. raucao

    ok, well. i just told you that it works in all of the other many rooms i'm using and that i noticed it only for this room just now

  72. raucao

    but i'll ask there then

  73. raucao

    actually, nothing to ask for. i'll just check it myself

  74. raucao

    they clearly state that MAM is supported ,and they also have it as an option in the room menu

  75. raucao

    pep., are you using dino daily, or where does that knowledge come from?

  76. pep.

    "Message history" in the room details is muc history, not MAM

  77. raucao

    it's not message history

  78. pep.

    it is.

  79. raucao

    it's literally message archiving

  80. larma

    raucao, pep. is right, dino doesn't do MAM in MUCs (dino dev here)

  81. larma

    though if you didn't recognize yet, MUC history isn't that bad it seems 🙂

  82. pep.

    raucao, sorry I'm not trying to play you

  83. raucao

    so it lets you enable it but doesn't do it?

  84. raucao

    that's very not great

  85. raucao

    https://xmpp.kosmos.org:5443/upload/791c7ed148e453f934ef56e1a4acb79a30845f0f/Eu5C2s84i7IGyDNlMGd1W6YYwrRb1TBxaHlih8MH/Screenshot_from_2020-03-14_18-34-54.png

  86. pep.

    raucao, enable?

  87. raucao

    the room options

  88. pep.

    that's not MAM

  89. raucao

    for room settings

  90. larma

    The MUC configuration form is send from the server and just displayed by dino

  91. pep.

    That's confusing settings

  92. raucao

    message archiving is not message archiving?

  93. pep.

    (not dino's choice)

  94. pep.

    raucao, message archiving here is MUC history

  95. raucao

    waaat

  96. raucao

    guys

  97. pep.

    Ah wait

  98. pep.

    No, message archiving is MAM here, you're correct

  99. raucao

    of course it is

  100. pep.

    That doesn't mean dino fetches it

  101. larma

    raucao, servers can add arbitrary settings there, dino just displays them without knowing what they mean

  102. raucao

    we don't allow normal history on our server

  103. Zash

    MUC history is something you get when you join, unless you actively opt-out.

  104. raucao

    yes, i realize that it's the room setting

  105. pep.

    That's just options your servers passes you

  106. raucao

    i know

  107. raucao

    still abysmal ux to show that and then not support it

  108. raucao

    no matter where it comes from

  109. raucao

    so it must be my phone that keeps track of history

  110. larma

    raucao, I do agree to some extend, but it's hard to do anything against that

  111. raucao

    and me being usually joined in the rooms i use

  112. raucao

    larma: what's so hard about implementing mam?

  113. raucao

    that's the right thing to do

  114. raucao

    if it does it for normal conversations anyway

  115. pep.

    raucao, "what's so hard about .." is probably not the way to do :p

  116. raucao

    that's a question in response to someone saying it

  117. raucao

    > but it's hard to do anything against that

  118. raucao

    that's a valid question

  119. raucao

    if someone says it's hard

  120. raucao

    i'm genuinely interested in improving the situation

  121. pep.

    it's slightly different then normal MAM, you have to target a MUC. You also have to special case MUC-PMs I guess

  122. raucao

    because i'm highly technical, so if i run into this, then many people will

  123. pep.

    And.. privacy concerns don't apply at the same points

  124. pep.

    Though I guess that should be solved when configuring the MUC..

  125. raucao

    there are no privacy concerns for local archives

  126. larma

    a) it's hard to implement MAM, especially with MUCs b) it's hard to filter room settings to not display settings that could be confusing because they don't affect dino

  127. pep.

    raucao, "local"?

  128. pep.

    muc mam is stored on the muc

  129. raucao

    yes, but your local history is stored locally

  130. raucao

    what you mean is already the room setting

  131. raucao

    so you can choose it per room

  132. raucao

    larma: so it's not implemented at all? i understood what pep. said as it being implemented for 1:1 chats

  133. raucao

    and it's clearly listed in https://github.com/dino/dino/wiki/Supported-XEPs

  134. larma

    It is implemented for your local server which means it can and does fetch the history of 1:1 chats

  135. raucao

    so for MUCs it would have to ask the MUC server is the main difference, aside from slightly different message, due to sender being the muc jid, right?

  136. raucao

    i added a comment on https://github.com/dino/dino/wiki/Supported-XEPs

  137. raucao

    so it's clear for people looking at that

  138. raucao

    sry for being offtopic in here now. the conversations/dino setup works so well for me that i was certain it must have been implemented :)

  139. larma

    the complicated part about MAM is not fetching messages, it's about fetching the messages you need, keeping track which you already got etc

  140. raucao

    yes, but you already solved that

  141. raucao

    obviously

  142. larma

    it becomes more complicated if you have multiple data sources

  143. raucao

    you only have one, no? the muc server's source

  144. larma

    well for the sync process I have mine and all the MUCs I am joined to

  145. raucao

    yes, of course

  146. raucao

    but that's only one variable

  147. larma

    it's not *that* simple

  148. larma

    I am not saying we are not going to implement it

  149. larma

    it's on the todo for 0.2 😉

  150. raucao

    cool

  151. larma

    (but it took more than 3 years from 0.0 to 0.1, so not sure what that means)

  152. larma

    it's a requirement for reactions which is also planned for 0.2 😉

  153. raucao

    i would say it's a requirement for all usage of a modern chat app

  154. raucao

    message history is a basic feature, which users of other chat apps do expect IMO

  155. raucao

    not just 20 messages "xmpp history", but i mean seamless archives with no holes

  156. larma

    you still have the local history, so it's not like things don't work properly

  157. raucao

    local history doesn't give you missing messages

  158. raucao

    to me it's broken

  159. larma

    it's just that if the server does not provide the necessary muc history to give you the missing messages that you have missing messages

  160. raucao

    no server will give you 1000 messages

  161. larma

    that's probably why you didn't even notice yet that there is no MAM in MUCs

  162. raucao

    especially not as default config

  163. raucao

    for normal history

  164. raucao

    because it's wildly inefficient

  165. larma

    you also usually don't look back 1000 messages in a history

  166. raucao

    no, but more than 20 for sure

  167. raucao

    the reason i didn't notice is that i usually don't leave rooms

  168. larma

    I am not saying it's perfect, but it's good enough for many

  169. larma

    well, if you leave a room you don't get its messages

  170. larma

    you are not supposed to

  171. raucao

    i think that's a very counterproductive opinion if you wants users to switch from telegram et al

  172. raucao

    but you're entitled to it, of course

  173. larma

    if you leave a signal or whatsapp group and join again later, you won't be able to read the messages in between

  174. raucao

    i didn't say signal or whatsapp. those are usually not used with larger groups as chat channels

  175. raucao

    more like small group of friends

  176. larma

    same for IRC

  177. raucao

    hahaha

  178. larma

    or Matrix depending on channel configuration

  179. raucao

    saying it's as bad as IRC is not a good thing

  180. raucao

    discourse, slack, etc. are the competitors in this use case

  181. larma

    slack, the thing where you can only read the latest 5000 messages in the free version?

  182. raucao

    they all have seamless history, because otherwise you can't work with people properly

  183. raucao

    yes, that thing. people do pay for it. that should tell you that it's valuable to have the history

  184. raucao

    people literally pay money for chat history

  185. raucao

    it's hilarious, but that's proving how important it is for work

  186. raucao

    also gitter, mattermost, rocket.chat and all the other ones focused on public rooms

  187. raucao

    or work rooms

  188. larma

    I guess you miss my point. I am not saying we don't want to implement MAM in MUCs, just that there are many occations where it is not wanted the way you are envisioning it