XSF Discussion - 2011-09-21


  1. Neustradamus

    stpeter: XEP-0045 1.25 rc6 is released or not?

  2. Jehan

    Hello.

  3. Jehan

    Hmmm... I am sad. I think I will miss again the meeting. :-(

  4. Jehan

    Damn timezones! I really wanted to come for today's meeting but I have to leave.

  5. Jehan

    Have a nice meeting, all!

  6. stpeter

    thanks, Jehan

  7. Jehan

    Yep stpeter. I was hoping I could participate to the discussion (if any) for my XEP proposal.

  8. Jehan

    But I am currently in Tokyo and a friend is leaving Japan tomorrow. So there is a goodbye party.

  9. Jehan

    Anyway bye.

  10. stpeter

    T-2 minutes?

  11. stpeter

    brb

  12. Florian

    hi everyone

  13. Florian

    Board Meeting time, right?

  14. stpeter

    that's my understanding :)

  15. Florian

    Will, Nyco and Jack missing?

  16. stpeter

    jack is online, let me invite him

  17. Florian

    just realized, Swift can't invite to MUC?

  18. stpeter

    Kev said he's working on Swift's MUC support :)

  19. Florian

    :)

  20. stpeter

    he wants the ability to kick people, at least

  21. Kev

    I'll do invites once Tobias's FT support is integrated.

  22. Kev

    Don't want to create too many merge conflicts for him.

  23. Florian

    :)

  24. Florian

    Agenda: http://typewith.me/xsf

  25. jack

    hello all

  26. Kev

    I like the way you've filled in who's going to be turning up beforehand :D

  27. stpeter

    sometimes direct IM works better than room invites

  28. Florian

    haha, it was there from the last meeting :)

  29. Florian

    Saved June 29, 2011

  30. Florian

    :)

  31. Florian

    any other topics?

  32. stpeter

    I don't have any other topics

  33. bear

    Server Certification - cool topic

  34. stpeter

    the big question is, are we serious about server certification and if so how do we make it happen?

  35. Florian

    yup

  36. bear

    before we can even consider that, we have to have a test suite in place

  37. stpeter

    as I said, probably I can raise some money for this (e.g., to help pay for travel to get the right people in one place), but I think the bigger challenge is to organize and motivate the work

  38. Florian

    well, I think we should have a spec

  39. Florian

    i.e. what needs to be tested, what needs to be supported for different certifications

  40. bear

    also we would need to discuss if we do levels of certification

  41. Florian

    right

  42. Kev

    Perhaps it might be interesting to find some organisation that wants to test server compliance or interop, and find out what their requirements are. Try to build something for them first, and go from there?

  43. stpeter

    Florian: how is that different from http://xmpp.org/extensions/xep-0302.html and the more detailed features listed in RFC 6120 etc?

  44. jack

    PDF has a series of test suites that people can run

  45. jack

    i wonder how those are done

  46. jack

    it might be useful to look at what other test suites do before embarking on our own

  47. bear

    I'm sure somewhere in the bowels of the IETF someone has done certifying

  48. Florian

    yeah

  49. stpeter

    "Certify" might be a strong term

  50. bear

    compliance suite?

  51. stpeter

    (semi-)automated interop testing is probably the right place to start

  52. bear

    *nods*

  53. Florian

    yeah

  54. Florian

    something along the lines of the W3C Validator

  55. stpeter

    right

  56. stpeter

    that's what I have in mind

  57. Florian

    looking at XEP-302 ... maybe even add federation to that?

  58. bear

    then xsf would need to host the service and allow people to point it at their servers

  59. Florian

    right

  60. stpeter

    Florian: you might be right that we need a more focused spec that describes the s2s scope more precisly

  61. Kev

    I think there's much more value in having sometheng people can run themselves off the 'Net.

  62. Florian

    stpeter: yeah

  63. Kev

    'Net-based ones are great for devs testing their own systems, but less so for some organisation wanting to validate a bunch of services internally.

  64. Florian

    hmm, another question is .... do we "certify" server software or server deployments?

  65. Kev

    So both modes would be ideal.

  66. bear

    having a validation service we then could also start advertising the scores/results of people who run against it

  67. Florian

    bear: indeed

  68. bear

    the source for a validation service would allow for in-house use by vendors

  69. Florian

    right

  70. Florian

    I think there's two ways of seeing this ...

  71. stpeter

    bear: makes sense

  72. Florian

    one is to test compliance of a running server (does federation work, valid certs installed....) and the other one is more internal, i.e. does PubSub work the way it's supposed to

  73. stpeter

    I think that most serious XMPP server developers have such test suites internally, the challenge is working on something common for the greater good

  74. Florian

    for the first one, a W3C online tool would be awesome (target: sysadmins)

  75. Kev

    stpeter: Knowing your own stuff works, is different from knowing someone else's works, though.

  76. Florian

    the other one, the target is the server vendors

  77. Will

    shit - we changed the time didn't we

  78. Florian

    Hi Will :)

  79. stpeter

    Will: yeah

  80. Will

    :-(

  81. stpeter

    Will: but at least Jack's able to join us at this time :)

  82. Will

    so, what have I missed?

  83. Florian

    so what do you guys think of maybe splitting this up into 2 things ... a test suite for developers, and a test site for deployments/sysadmins

  84. bear

    talking about server certification/validation and running it as a system for people to point at their servers

  85. Florian

    Will: Server Certification

  86. Tobias

    this is planned to be purely functionality based testing, right?

  87. bear

    if the tool is written to have suites of tests and to output to a file, then the web service part becomes just an html display of that output

  88. bear

    and the file is available for downloading

  89. stpeter

    yeah

  90. stpeter

    Tobias: yes!

  91. stpeter

    Tobias: no scalability testing

  92. Tobias

    stpeter, yeah..there are enough tools for that ;)

  93. Florian

    right

  94. stpeter

    Tobias: well, we've always avoided scalability testing because there are more variables, set up matters a lot, etc.

  95. Tobias

    stpeter, i know/just getting to know :)

  96. stpeter

    and we're supposed to know something about the protocol :)

  97. Florian

    hehe

  98. stpeter

    sorry, interruption here, bbiaf

  99. bear

    seems that we need to 1) gather up tests that could be in said suite 2) start working on python/lua code to test said items and 3) rinse/repeat

  100. Florian

    yup

  101. dwd

    Who's doing the work?

  102. bear

    I mention python/lua code as those are common enough languages that have modern xmpp support libs

  103. Florian

    so you guys don't think there should be 2 seperate projects?

  104. bear

    this is something i've always been interested in - I would love to have a chance to work on this

  105. Florian

    one for deployments a la W3C and one for the actual code?

  106. bear

    I think we should start small and simple and iterate

  107. dwd

    Florian, I think a "Verified by XSF" thing for a particular service seems sensible.

  108. bear

    a project like this could get mired down in politics and/or apathy if the goal is too broad

  109. Florian

    right

  110. stpeter

    back

  111. bear likes "Verified by XSF"

  112. stpeter

    bear: totally agreed on politics or apathy

  113. Will

    as long as we make sure that people don't read "verified" as "approved"

  114. stpeter

    indeed

  115. bear

    heck, the first steps would be a spreadsheet of what to test with a column pointing to source snippets on how to test for that

  116. stpeter

    Florian: I think bear is right that the validator would just be a public interface to the verifier

  117. stpeter

    (if that makes sense)

  118. bear

    it's how the Atom folks did their service

  119. bear

    a test suite that can be driven by a fancier web interface

  120. stpeter

    yeah

  121. Florian

    yup

  122. bear

    probably a good start would be to get a wiki page of what to test and then just announce it to the list and start iterating on that

  123. stpeter glances at the "Programming in Lua" book on his desk

  124. Florian

    +1

  125. stpeter

    WFM

  126. bear

    having it as a lua and python lib would be a great success

  127. Florian

    yeah

  128. stpeter

    I'm language agnostic really, just want to use something that will draw in or keep contributors

  129. Florian

    might be something for the hackfest in February?

  130. stpeter

    Florian: that's the idea, yes

  131. stpeter

    Florian: and I could probably get funding to pay for travel costs of the top contributors :)

  132. Florian

    :)

  133. bear

    I will ask Mozilla to see if they will sponsor me on this

  134. stpeter logs into wiki.xmpp.org

  135. bear

    http://wiki.xmpp.org/web/Verify

  136. Will

    succinct

  137. bear

    I liked the short/simple one for Interop

  138. stpeter

    heh

  139. bear

    so went with that flavour

  140. bear

    I had a good set of talks with the moz folks who are working on backend systems last week

  141. stpeter

    BTW we have an interop@xmpp.org list if we want to use it http://mail.jabber.org/mailman/listinfo/interop

  142. bear

    I think that mailing list would be perfect

  143. Florian

    +1

  144. Will

    this is terribly rude but i have to go in a minute or two, have we decided and is there anything more to discuss?

  145. stpeter

    don't want to spam standards@

  146. stpeter

    Will: I think we're done for now

  147. bear

    not rude at all Will - perfectly timed as I think we are close to being done officially

  148. Florian

    yup :)

  149. bear

    so +1 to a verification suite to be driven by XSF?

  150. Will

    well, it is kind of cheeky as i forgot to turn up on time :-)

  151. Will

    +1

  152. stpeter

    :)

  153. stpeter

    easy enough to set up a separate git repository once we have code

  154. bear

    cool

  155. bear

    I would love to take point on this along with anyone else

  156. dwd

    Is that possible? Surely you either take point, or you don't? Otherwise it's not a "point", as such...

  157. Florian

    I'd be happy to help

  158. stpeter

    bear: I might be able to interest some folks in Cisco to help

  159. bear loves having folks like dwd in his life

  160. stpeter

    and I'll commit to putting time into this, too

  161. Kev

    dwd: That'd be to 'take blob', wouldn't it?

  162. Kev

    Or 'take disorganised heap' or such.

  163. dwd

    Kev, Or "take blunt".

  164. Kev

    Or that.

  165. stpeter

    maybe I could learn some lua (or re-learn some Python) along the way :)

  166. Florian

    lol

  167. bear

    well, I fubar'd GSoC this season so I want to redeem myself

  168. dwd

    Kev, But this is a "pointless" discussion.

  169. dwd is particularly pleased with that one.

  170. stpeter

    :P

  171. Florian

    :p

  172. bear

    well done!

  173. bear oozes his blob over dwd and kev in hopes of blunting any further puns

  174. stpeter must be missing some context

  175. bear

    ok now :) - we have agreed on that

  176. bear

    any further items to discuss?

  177. Florian

    not from my end

  178. stpeter

    bear: once we have a basic wiki page describing the project, I can send a message to standards@xmpp.org and jdev@jabber.org (plus the members list)

  179. bear

    cool

  180. Florian

    awesome

  181. bear

    deadline of next board meeting to have wiki page done?

  182. Florian

    sounds good

  183. bear

    or do you want 2 meetings worth of time?

  184. stpeter

    sure

  185. stpeter

    I think a week is enough

  186. bear

    k

  187. stpeter

    at least for the basic description

  188. bear

    thanks

  189. stpeter

    bear: we also (still) have xmpp:interop@muc.xmpp.org

  190. dwd

    Is there a plan for verifying the verifier? Like, having leave to appeal to Council to review any discrepencies, or something?

  191. bear

    I think we would have to have the council certify it

  192. Florian

    indeed

  193. bear

    (this also will give the next council something very interesting to look forward to)

  194. dwd

    I think you'd need to have them on-hand to check results, especially in the case where the verifier complains and the implementor disputes it.

  195. bear

    +1

  196. Florian

    +1

  197. bear

    they would be the final authority - because that would also drive changes to the specs

  198. dwd

    bear, Well, I'm not sure - mostly, it'd probably find bugs in the verifier or its libraries.

  199. stpeter

    bear: yes, I think this will drive some fixes to the specs, better lists of testable features (along the lines of what's in RFC 6120), etc.

  200. stpeter

    dwd: true

  201. bear

    :) - I'm being optimistic that the verifier will be perfect!

  202. bear coughs and returns to reality

  203. stpeter

    who's the room owner for interop@muc.xmpp.org?

  204. Florian

    good question

  205. dwd

    stpeter, Kev?

  206. dwd

    stpeter, He more or less ran the last Interop.

  207. Kev

    Possibly me, I guess.

  208. stpeter

    yeah

  209. Kev

    Nope.

  210. stpeter

    bear: I've edited the minutes at http://typewith.me/xsf

  211. stpeter

    I think they're good to go now

  212. bear

    yep, been watching

  213. stpeter

    heh ok :)

  214. bear

    cool - i'll send them out now

  215. stpeter

    thanks!

  216. stpeter

    I'm psyched

  217. bear

    thanks

  218. bear

    me also - will be a nice fall/winter project

  219. stpeter

    heh ... winterop? ;-)

  220. Florian

    :D

  221. stpeter goes back to reviewing Internet-Drafts for tomorrow's IESG telechat

  222. bear goes back to herding Tegra250 boards

  223. stpeter

    :)

  224. stpeter

    you and I are heading in opposite directions :)

  225. bear

    yea, I'm in very barren lands it seems

  226. stpeter

    Neustradamus: I fixed the "www.xmpp.org" stuff in the XEP Python scripts

  227. Neustradamus

    stpeter: ok :)

  228. Neustradamus

    for emails or proposal XEPs ?

  229. stpeter

    Neustradamus: I will soon be AFK for a few hours but I will be back later

  230. Neustradamus

    it is possible to modify the theme of xmpp.org and jabber.org ?

  231. Neustradamus

    In the theme folder / misc/nav-posts.php <div class="pagination"> <span class="previous"><?php previous_posts_link(__('&laquo; Previous', 'carrington-blog')) ?></span> <span class="next"><?php next_posts_link(__('Next &raquo;', 'carrington-blog')) ?></span> </div> change to <div class="pagination"> <span class="previous"><?php previous_posts_link(__('&laquo; Next', 'carrington-blog')) ?></span> <span class="next"><?php next_posts_link(__('Previous &raquo;', 'carrington-blog')) ?></span> </div>

  232. Neustradamus

    stpeter: ok :)

  233. stpeter

    Neustradamus: for inbox proposals, last calls, and deferred specs

  234. Neustradamus

    :)

  235. stpeter

    Neustradamus: those were the only scripts that still had www

  236. stpeter

    Neustradamus: I think that might be fixed in a more recent version of the theme, so the first step is to upgrade

  237. Neustradamus

    an example of pages: http://xmpp.org/extensions/inbox/bind.html

  238. Neustradamus

    the redirection is not updated (not alone)

  239. stpeter

    some of those do not redirect for historical reasons -- we want to keep the old versions online

  240. stpeter

    ah, I see

  241. Neustradamus

    you see :)

  242. stpeter

    yeah, we can fix that in the lighttpd config

  243. Neustradamus

    not directly in the page ?

  244. stpeter

    about http://xmpp.org/extensions/tmp/ -- yes we keep those old versions around (or don't put the time into fixing them)

  245. Neustradamus

    I found new old addresses: http://www.jabber.org/tmp/ (redirect to http://www.jabber.org/ ?) http://www.jabber.org/muc-logs/ (http://logs.jabber.org/ now?)

  246. Neustradamus

    ok

  247. stpeter

    no, easier to fix that in the server config -- I don't want to change every one of the old .html files

  248. stpeter

    bbiab

  249. Neustradamus

    ok

  250. Neustradamus

    I found a "very" old page: http://www.jabber.org/service-policy/ Personal Eventing Protocol has been removed... Can you add a XMPP URI for jabber@conference.jabber.org ? Change the part with 5223 port... Change "<http://logs.jabber.org/>" to http://logs.jabber.org/ Change "JabberID" to "Jabber ID" Change "policy will be posted at www.jabber.org and" to "policy will be posted at jabber.org and t" (remove www.) In "How to Contact Us" part : there is a link to http://www.jabberforum.org/forumdisplay.php?f=21 (dead) Update old xmpp.org links to new links On http://www.jabber.org/about/ http://www.jabber.org/index.php/network/ to http://www.jabber.org/network/ http://www.jabber.org/index.php/download-a-client/ to http://www.jabber.org/download-a-client/ Update http://xmpp.org/services/ to http://xmpp.org/resources/public-services/ Update other xmpp.org links on other pages. http://xmpp.org/extensions/all.shtml is dead ?

  251. Neustradamus

    Kev: http://logs.jabber.org/ it is possible to have same logs for all? old and new?

  252. stpeter

    Neustradamus: why does it matter that we have old links with index.php in them? they redirect just fine!

  253. stpeter

    that's why we set up redirects :)

  254. stpeter

    time for lunch here...

  255. Neustradamus

    stpeter: ok ok for logs

  256. Neustradamus

    stpeter: you can close http://tracker.xmpp.org/browse/ITEAM-12 ;)

  257. Neustradamus

    about logs, there are double... http://logs.jabber.org/ and http://logs.jabber.org/old/ there are same.

  258. Neustradamus

    and http://logs.jabber.org/new/ too :D

  259. stpeter

    ok

  260. stpeter

    now I really need to be away for a while

  261. stpeter

    Neustradamus: jabber.org is *not* an XSF service, so please post about that in the jabber@conference.jabber.org room

  262. stpeter

    bbl

  263. Neustradamus

    stpeter: yes

  264. Astro

    hi

  265. Astro

    will all validator-relevant stuff be posted on interop@?

  266. bear

    astro - most likely yes

  267. bear

    and please do poke us if you notice anything not on interop@

  268. Astro

    great

  269. Astro

    maybe the node.js xmpp-server is going to be ready for interop testing at next fosdem

  270. bear

    sweet!