/irc-logs / w3c / #testing / 2013-01-30 / end

Options:

  1. # Session Start: Wed Jan 30 00:00:00 2013
  2. # Session Ident: #testing
  3. # [00:04] * Quits: paul_irish (~paul_irish@public.cloak) (Client closed connection)
  4. # [00:26] * Joins: paul_irish (~paul_irish@public.cloak)
  5. # [01:49] * Quits: paul_irish (~paul_irish@public.cloak) (Client closed connection)
  6. # [01:56] * Quits: darobin (rberjon@public.cloak) (Client closed connection)
  7. # [02:00] * Quits: plh (plehegar@public.cloak) (Ping timeout: 60 seconds)
  8. # [02:37] * Joins: plh (plehegar@public.cloak)
  9. # [02:42] * Quits: jhammel (~jhammel@public.cloak) ("leaving")
  10. # [02:45] * Quits: plh (plehegar@public.cloak) ("Leaving")
  11. # [03:11] * Quits: ArtB (~abarsto@public.cloak) ("Leaving.")
  12. # [03:27] * heycam|away|away is now known as heycam|away
  13. # [06:28] * heycam|away is now known as heycam|away|away
  14. # [06:45] * Joins: darobin (rberjon@public.cloak)
  15. # [07:35] * Quits: glenn (~gadams@public.cloak) (Client closed connection)
  16. # [07:57] * Joins: sstewart6 (~simons@public.cloak)
  17. # [08:15] * Quits: darobin (rberjon@public.cloak) (Client closed connection)
  18. # [10:34] * Joins: AutomatedTester (~davidburns@public.cloak)
  19. # [10:48] <AutomatedTester> jgraham: hey, is there a standardised HTTP server for serving files for testing?
  20. # [11:35] <jgraham> No, but that's something we totally need to do
  21. # [11:35] <jgraham> AutomatedTester: ^
  22. # [11:36] <AutomatedTester> yea, something has come up in Mozilla about a test httpd we use 'isn't good enough'
  23. # [11:36] <AutomatedTester> jgraham: in Opera you guys use OperaDriver to launch Testharness.js tests iirc?
  24. # [11:36] <jgraham> Well the "standard" is Apache + PHP, but that doesn't really work for everyone
  25. # [11:36] <jgraham> Well
  26. # [11:36] <jgraham> In Opera we have the testharness tests on apache
  27. # [11:37] <jgraham> with a central server
  28. # [11:37] <jgraham> We don't distibute the tests as part of the build package
  29. # [11:37] <jgraham> So it's a bit of a different situation
  30. # [11:38] <jgraham> (then we use webdriver to load the test and a local web server just to recieve the test results via HTTP POST to localhost)
  31. # [11:38] <jgraham> (there is quite a lot wrong with this model)
  32. # [11:39] <AutomatedTester> jgraham: for a discussion I am having the "then we use webdriver to load the test" part will do me fine :)
  33. # [11:40] <jgraham> (I have just been told by andreastt that it's webdrivewr, although I thought that we used our low level API directly)
  34. # [11:40] <AutomatedTester> jgraham: its more a case of a high level API in the test that then does all the low level api stuff
  35. # [11:41] <AutomatedTester> andreastt: is that fair?
  36. # [11:50] <AutomatedTester> jgraham: is testharness.js the standardised runner for in the browser?
  37. # [11:52] <MikeSmith> AutomatedTester: yes for W3C specs, testharness.js is what all tests that can be automated in-browser should be using
  38. # [11:52] <AutomatedTester> MikeSmith: brilliant
  39. # [11:52] * AutomatedTester goes back to his "discussion"
  40. # [12:32] <andreastt> It used to be the case that we used Scope directly, I think.
  41. # [12:32] <andreastt> But for the past couple of years we've used webdriver to launch the browser and the HTML documents.
  42. # [12:34] <AutomatedTester> thanks andreastt :)
  43. # [12:37] <andreastt> I should note that test results reporting is something entirely different, we don't currently use webdriver for that.
  44. # [12:37] <andreastt> Well, actually for some tests we do. Screenshot and reftests come to mind.
  45. # [12:37] <andreastt> But not for, say, JS tests.
  46. # [13:35] * Joins: abarsto (~abarsto@public.cloak)
  47. # [13:35] * abarsto is now known as ArtB
  48. # [13:55] * Joins: glenn (~gadams@public.cloak)
  49. # [13:56] * Joins: plh (plehegar@public.cloak)
  50. # [14:53] <ArtB> plh, would you please change the Permissions on https://www.w3.org/2013/01/29-webtest-minutes.html to Public? and the 30-webtest-minute.html too?
  51. # [14:55] * Joins: sstewart6_ (~simons@public.cloak)
  52. # [14:59] * Quits: sstewart6 (~simons@public.cloak) (Ping timeout: 60 seconds)
  53. # [14:59] * Joins: sstewart6 (~simons@public.cloak)
  54. # [14:59] * Quits: sstewart6_ (~simons@public.cloak) (Ping timeout: 60 seconds)
  55. # [15:01] <plh> Art, done
  56. # [15:01] * Quits: sstewart6 (~simons@public.cloak) (sstewart6)
  57. # [15:01] <AutomatedTester> andreastt: thanks for that, sorry I missed your reply
  58. # [15:01] <AutomatedTester> I am more interested in test running
  59. # [15:03] <ArtB> thanks PLH!
  60. # [15:36] * Joins: sstewart6 (~simons@public.cloak)
  61. # [15:53] <sstewart6> AutomatedTester: I'd love it if I could check out the tests and start running them
  62. # [15:53] <sstewart6> So "apache" is a little heavy
  63. # [15:53] <sstewart6> A lightweight httpd in python would be okay
  64. # [16:00] * Joins: wolfram (~cain@public.cloak)
  65. # [16:11] * Joins: darobin (rberjon@public.cloak)
  66. # [16:17] <AutomatedTester> sstewart6: agreed
  67. # [16:18] <AutomatedTester> sstewart6: plus apache, while it works on multiple OSes, is not very easy to deploy on multiple servers
  68. # [16:19] <sstewart6> So, lightweight python server
  69. # [16:20] * AutomatedTester thinks that sstewart6 missed the "agreed" statement...
  70. # [16:20] <sstewart6> I didn't. I repeated for emphasis :)
  71. # [16:20] <sstewart6> Or emPHAsis, I could say
  72. # [16:20] <sstewart6> arf arf
  73. # [16:21] <jgraham> So I don't know about your specific use cases, but in general for testing web standards we need a *very* flexible server
  74. # [16:22] <sstewart6> jgraham: what kinds of things are you thinking of?
  75. # [16:22] <sstewart6> BASIC/Digest auth springs to mind
  76. # [16:22] <sstewart6> setting custom headers
  77. # [16:22] <jgraham> Basically it should be possible for a given test to return an arbitary stream of bytes
  78. # [16:22] <sstewart6> controlling how long content takes to return
  79. # [16:22] <sstewart6> (simulated latency)
  80. # [16:22] <jgraham> Yup, simulated latency is one
  81. # [16:22] <jgraham> other thing
  82. # [16:23] <jgraham> Also, sometimes having server side state is useful (e.g. testing progress events)
  83. # [16:23] <sstewart6> Right.
  84. # [16:23] <jgraham> I think I wrote a mail on this to public-test0infra
  85. # [16:23] <sstewart6> Not sure I'm on that list (yet)
  86. # [16:24] <sstewart6> So python's not a bad choice. For the common case we could use any WSGI server
  87. # [16:24] <sstewart6> And for those cases where we need the additional control we could just find something a little more sophisticated
  88. # [16:25] <sstewart6> It's entirely possible I'm missing something
  89. # [16:25] <sstewart6> So far for the OSS webdriver implementation we've not needed to be terribly sophisticated when serving pages :)
  90. # [16:27] <andreastt> As jgraham pointed out, there's not really any standard.
  91. # [16:27] <andreastt> We use Apache.
  92. # [16:27] <andreastt> For Python, I'm not sure if simplewebserver is enough?
  93. # [16:27] <AutomatedTester> we use httpd.js currently for some tests
  94. # [16:27] <andreastt> Least it's biult in.
  95. # [16:27] * Quits: darobin (rberjon@public.cloak) (Client closed connection)
  96. # [16:28] <AutomatedTester> but there is a lot of work required to get it running
  97. # [16:28] <jgraham> httpd.js is pretty neat, but it's very Mozilla-specific
  98. # [16:28] <AutomatedTester> agreed
  99. # [16:28] <AutomatedTester> and it has a number of downsides too
  100. # [16:28] <sstewart6> Jetty and Webbit are my two favourite java servers
  101. # [16:28] <jgraham> And assumes only one client will access the server at a time
  102. # [16:28] <sstewart6> s/java/java httpd/
  103. # [16:28] <AutomatedTester> rather stay clear of Java
  104. # [16:28] <andreastt> sstewart6: I'm quite sure I don't want to pull in Java in order to run the conformance tests.
  105. # [16:29] <sstewart6> And that's why I suggested python :)
  106. # [16:29] <jgraham> (I can't see the message I thought I wrote)
  107. # [16:29] <jgraham> Yes, AFAIK all of (Opera, Google, Mozilla) are happy with Python. Microsoft hate change and Apple don't talk to me.
  108. # [16:30] <AutomatedTester> s/don't talk to me/don't talk to anyone/
  109. # [16:30] <sstewart6> Python: the least politically encumbered language
  110. # [16:30] <jgraham> Pretty much :)
  111. # [16:34] <andreastt> They will talk to sstewart6 if you supply two bottles of Côte du Rohne.
  112. # [16:35] <AutomatedTester> and tell them thy dont really have to do any work
  113. # [16:35] <sstewart6> sstewart6: only slightly politically encumbered
  114. # [17:25] * Joins: darobin (rberjon@public.cloak)
  115. # [18:09] * Quits: wolfram (~cain@public.cloak) (wolfram)
  116. # [18:32] * Joins: jhammel (~jhammel@public.cloak)
  117. # [19:46] * Quits: darobin (rberjon@public.cloak) (Client closed connection)
  118. # [19:50] * Quits: sstewart6 (~simons@public.cloak) (sstewart6)
  119. # [19:57] * Quits: plh (plehegar@public.cloak) ("Leaving")
  120. # [19:57] * Joins: darobin (rberjon@public.cloak)
  121. # [20:00] * Quits: darobin (rberjon@public.cloak) (Client closed connection)
  122. # [20:59] * Quits: AutomatedTester (~davidburns@public.cloak) (Client closed connection)
  123. # [21:00] * Joins: AutomatedTester (~davidburns@public.cloak)
  124. # [22:52] * Quits: AutomatedTester (~davidburns@public.cloak) (Client closed connection)
  125. # [22:54] * Joins: AutomatedTester (~davidburns@public.cloak)
  126. # [23:35] * heycam|away|away is now known as heycam|away
  127. # Session Close: Thu Jan 31 00:00:00 2013

The end :)