/irc-logs / freenode / #whatwg / 2009-02-23 / end

Options:

  1. # Session Start: Mon Feb 23 00:00:00 2009
  2. # Session Ident: #whatwg
  3. # [00:06] * Quits: myakura (n=myakura@p3020-ipbf505marunouchi.tokyo.ocn.ne.jp) ("Leaving...")
  4. # [00:10] * Quits: mpt (n=mpt@canonical/launchpad/mpt) ("Ex-Chat")
  5. # [00:10] <Philip`> Hmm, seems like the idea is to design a browser that separates each origin into its own process, with security enforced by a kernel process, only allowing cross-origin access to scripts and stylesheets (based on Content-Type)
  6. # [00:10] <Philip`> and cross-origin frames/images/plugins are handled by delegating a rectangle of the screen to another process
  7. # [00:12] <Philip`> with the idea being that same-origin-policy enforcement in current browsers is spread throughout the code and thus vulnerable to bugs, so it's more secure to have it enforced by a relatively simple kernel process instead
  8. # [00:14] <Philip`> and plugins are isolated into separate processes per origin too, and can only access resources through the kernel
  9. # [00:15] <Dashiva> So you need separate plugin instances for each origin?
  10. # [00:17] * dglazkov_ is now known as dglazkov
  11. # [00:17] <Philip`> Yes, I think
  12. # [00:17] <gsnedders> Hixie: Oh dear lord that CSS is horrible…
  13. # [00:17] <Philip`> (Also you need to modify the plugins so they communicate through the browser kernel)
  14. # [00:18] <gsnedders> Hixie: Do we not really for sane authors need CSS support? :P
  15. # [00:18] <gsnedders> (I'm of course not implying you aren't sane…)
  16. # [00:18] * Quits: dglazkov (n=dglazkov@72.14.224.1)
  17. # [00:20] <Philip`> (Strict compatibility is not a requirement - the idea is to sacrifice some compatibility for security)
  18. # [00:23] * Quits: heycam (n=cam@vpn-118-138-224-31.its.monash.edu.au) (Remote closed the connection)
  19. # [00:25] * Joins: mpt (n=mpt@canonical/launchpad/mpt)
  20. # [00:26] <jwalden> gsnedders: and x6 for h2-h6! :-D
  21. # [00:27] <jwalden> not to mention that h1 is really hx for the lowest x in a sectioning content element
  22. # [00:27] * gsnedders does know this
  23. # [00:27] * gsnedders has implemented the creating an outline algorithm
  24. # [00:27] <jwalden> at least it compresses well!
  25. # [00:28] <gsnedders> Probably the only implementation of the creating an outline algorithm that is actually used :)
  26. # [00:29] <jwalden> would be cool if html5.validator.nu had that implemented, for parity with the w3 headline generator for html4/xhtml1
  27. # [00:42] <gsnedders> (The TOC in the spec itself is made via the outline algorithm)
  28. # [00:45] * Quits: svl (n=me@69.Red-88-2-205.staticIP.rima-tde.net) ("And back he spurred like a madman, shrieking a curse to the sky.")
  29. # [00:53] * Quits: yecril71 (n=giecrilj@piekna-gts.2a.pl)
  30. # [01:04] * Joins: heycam (n=cam@vpn-118-138-224-34.its.monash.edu.au)
  31. # [01:17] * Joins: tantek (n=tantek@adsl-69-106-241-178.dsl.pltn13.pacbell.net)
  32. # [01:18] * Joins: weinig (n=weinig@99.140.60.47)
  33. # [01:23] * Quits: danbri (n=danbri@unaffiliated/danbri)
  34. # [01:34] * Parts: erlehmann (n=erlehman@86.59.25.121)
  35. # [01:40] <Hixie> jwalden: per spec h2-h6 don't actually do the magic sizing, only h1 does
  36. # [01:46] <jwalden> huh
  37. # [01:46] <jwalden> haven't seen that change
  38. # [01:46] <jwalden> in the spec
  39. # [01:46] <jwalden> maybe it's part of the rendering section
  40. # [01:48] <Hixie> change?
  41. # [01:49] <Hixie> h2-h6 have never done magic sizing, it's always only been <h1>
  42. # [01:49] <jwalden> bad assumptions on my part, I guess
  43. # [01:49] <Hixie> requiring h2-h6 to do magic sizing would be orders of magnitude more complexity, since then we'd have to do the entire outline algorithm each time
  44. # [01:54] <jwalden> hm, my mis-memory
  45. # [01:55] * Joins: danbri (n=danbri@77.167.54.81)
  46. # [01:59] * Quits: doublec (n=Chris_Do@118-93-186-208.dsl.dyn.ihug.co.nz) ("ChatZilla 0.9.79-rdmsoft [XULRunner 1.8.0.9/2006120508]")
  47. # [02:05] * Quits: danbri (n=danbri@unaffiliated/danbri)
  48. # [02:21] * Quits: hdh (n=hdh@118.71.79.203) (Remote closed the connection)
  49. # [02:35] * Quits: tndH (n=Rob@james-baillie-pc083-014.student-halls.leeds.ac.uk) ("ChatZilla 0.9.84-rdmsoft [XULRunner 1.9.0.1/2008072406]")
  50. # [02:50] * Quits: heycam (n=cam@vpn-118-138-224-34.its.monash.edu.au) (Read error: 110 (Connection timed out))
  51. # [02:52] * Joins: doublec (n=chris@202.0.36.64)
  52. # [03:31] * Joins: MikeSmith (n=MikeSmit@EM114-48-144-146.pool.e-mobile.ne.jp)
  53. # [03:32] * Joins: heycam (n=cam@vpn-118-138-224-16.its.monash.edu.au)
  54. # [03:52] * Joins: karlcow (n=karl@modemcable202.32-81-70.mc.videotron.ca)
  55. # [04:05] * Quits: tantek (n=tantek@adsl-69-106-241-178.dsl.pltn13.pacbell.net)
  56. # [04:12] * Quits: sayrer (n=chatzill@user-160va8b.cable.mindspring.com) (Read error: 110 (Connection timed out))
  57. # [04:15] * Quits: grimboy (n=grimboy@78-86-152-156.zone2.bethere.co.uk) (Read error: 110 (Connection timed out))
  58. # [04:22] * Joins: enki (n=enki@c-67-188-42-45.hsd1.ca.comcast.net)
  59. # [04:31] * Quits: weinig (n=weinig@99.140.60.47)
  60. # [04:56] * Joins: dglazkov (n=dglazkov@c-98-207-88-44.hsd1.ca.comcast.net)
  61. # [05:12] * Joins: sayrer (n=chatzill@user-160va8b.cable.mindspring.com)
  62. # [05:32] * Quits: doublec (n=chris@202.0.36.64) ("Leaving")
  63. # [06:00] * Joins: zcorpan (n=zcorpan@pat.se.opera.com)
  64. # [06:06] * Joins: dave_levin (n=dave_lev@c-98-203-247-78.hsd1.wa.comcast.net)
  65. # [06:26] * Joins: harig (n=opera@59.90.71.35)
  66. # [06:34] * Quits: dglazkov (n=dglazkov@c-98-207-88-44.hsd1.ca.comcast.net)
  67. # [06:39] * Quits: heycam (n=cam@vpn-118-138-224-16.its.monash.edu.au) ("bye")
  68. # [06:45] * Joins: weinig (n=weinig@99.140.60.47)
  69. # [06:46] * Quits: weinig (n=weinig@99.140.60.47) (Client Quit)
  70. # [06:47] * Joins: annevk (n=opera@EM114-48-144-146.pool.e-mobile.ne.jp)
  71. # [06:49] * Quits: MikeSmith (n=MikeSmit@EM114-48-144-146.pool.e-mobile.ne.jp) ("Tomorrow to fresh woods, and pastures new.")
  72. # [06:51] * Quits: annevk (n=opera@EM114-48-144-146.pool.e-mobile.ne.jp) (Client Quit)
  73. # [07:10] * Joins: tantek (n=tantek@adsl-63-195-114-133.dsl.snfc21.pacbell.net)
  74. # [07:27] * Joins: shepazu (n=schepers@cpe-76-168-214-103.socal.res.rr.com)
  75. # [07:30] * Quits: tantek (n=tantek@adsl-63-195-114-133.dsl.snfc21.pacbell.net)
  76. # [07:36] <zcorpan> Hixie: "not appliciable" in the status boxes probably shouldn't use the same icons as "excellent support"
  77. # [07:37] <Hixie> fixed
  78. # [07:37] <Hixie> now it's the same as none
  79. # [07:40] <zcorpan> anyone aware of bugs in browsers with <param>?
  80. # [07:42] <zcorpan> hmm what's the impl status of "Offsets into the media resource" (i.e. .duration, .currentTime, .loop)?
  81. # [07:43] <zcorpan> "incomplete" in firefox and safari?
  82. # [07:44] <Hixie> haven't tested either recently
  83. # [07:45] <zcorpan> what about canvas transformations?
  84. # [07:45] <zcorpan> Philip`?
  85. # [07:46] * zcorpan skips the canvas-related sections
  86. # [07:53] * Joins: maikmerten (n=merten@ls5dhcp196.cs.uni-dortmund.de)
  87. # [08:08] * Quits: mpt (n=mpt@canonical/launchpad/mpt) (calvino.freenode.net irc.freenode.net)
  88. # [08:08] * Quits: nessy (n=nessy@203-158-43-109.dyn.iinet.net.au) (calvino.freenode.net irc.freenode.net)
  89. # [08:08] * Quits: gavin (n=gavin@firefox/developer/gavin) (calvino.freenode.net irc.freenode.net)
  90. # [08:08] * Quits: YaaL (i=yaal@hell.pl) (calvino.freenode.net irc.freenode.net)
  91. # [08:08] * Quits: hsivonen (n=hsivonen@kekkonen.cs.hut.fi) (calvino.freenode.net irc.freenode.net)
  92. # [08:09] * Joins: mpt (n=mpt@canonical/launchpad/mpt)
  93. # [08:09] * Joins: nessy (n=nessy@203-158-43-109.dyn.iinet.net.au)
  94. # [08:09] * Joins: gavin (n=gavin@firefox/developer/gavin)
  95. # [08:09] * Joins: YaaL (i=yaal@hell.pl)
  96. # [08:09] * Joins: hsivonen (n=hsivonen@kekkonen.cs.hut.fi)
  97. # [08:10] * Joins: MikeSmith (n=MikeSmit@EM114-48-39-212.pool.e-mobile.ne.jp)
  98. # [08:11] * Joins: pesla (n=retep@procurios.xs4all.nl)
  99. # [08:18] * Quits: sayrer (n=chatzill@user-160va8b.cable.mindspring.com) (Read error: 110 (Connection timed out))
  100. # [08:20] <jwalden> currentTime works in Gecko, modulo bugs of course
  101. # [08:20] <jwalden> duration also seems to work
  102. # [08:23] * Joins: heycam (n=cam@210.84.48.144)
  103. # [08:23] * Joins: zalan (n=kvirc@catv-89-133-232-199.catv.broadband.hu)
  104. # [08:24] <zcorpan> Hixie: "The rendering section will define this in more detail." says the spec about size=""
  105. # [08:25] * Quits: heycam (n=cam@210.84.48.144) (Client Quit)
  106. # [08:34] <hsivonen> jwalden: the outline algorithm is a known feature request for v.nu
  107. # [08:35] <jwalden> so I figured :-)
  108. # [08:35] <hsivonen> jwalden: right now I'm focusing on the Gecko side of things, tyranny of email permitting
  109. # [08:36] <jwalden> all good, not like I had a meaningful reason to need it :-)
  110. # [08:39] <Hixie> zcorpan: i'll clean that up at some point
  111. # [08:41] * Joins: Maurice (n=ano@a80-101-46-164.adsl.xs4all.nl)
  112. # [08:51] * Joins: heycam (n=cam@210-84-48-144.dyn.iinet.net.au)
  113. # [08:54] * Quits: dbaron (n=dbaron@c-98-234-51-190.hsd1.ca.comcast.net) ("8403864 bytes have been tenured, next gc will be global.")
  114. # [09:02] <zcorpan> uh... why is there an annotation box at 5.9.4 The Location interface that seems to annotate something about <video>?
  115. # [09:04] * Joins: ap (n=ap@194.154.88.46)
  116. # [09:08] <Hixie> probably the IDs changed
  117. # [09:09] <Hixie> it happens occasionally
  118. # [09:12] * Joins: roc (n=roc@121-72-173-196.dsl.telstraclear.net)
  119. # [09:14] <zcorpan> is it fixable?
  120. # [09:14] <Hixie> only by making IDs never get reused
  121. # [09:16] * Joins: tantek (n=tantek@adsl-63-195-114-133.dsl.snfc21.pacbell.net)
  122. # [09:19] <zcorpan> which browsers support rel=prefetch?
  123. # [09:20] <Hixie> firefox is the only one i know of
  124. # [09:22] <Philip`> zcorpan: What about canvas transformations?
  125. # [09:22] * Joins: pesla\work (n=retep@procurios.xs4all.nl)
  126. # [09:22] <zcorpan> Philip`: which browsers support them (for the purpose of spec status annotation)?
  127. # [09:23] <zcorpan> Philip`: maybe you could go through the canvas section and annotate appropriately? :)
  128. # [09:25] <Philip`> http://philip.html5.org/tests/canvas/suite/tests/results.html#2d.transformation.transform.identity says all supported transform, and all except WebKit supported setTransform
  129. # [09:25] <Philip`> Hmm, it's possible but I don't think I really care enough to do that :-p
  130. # [09:30] <hsivonen> ooh. YouTube automatic caption translation is impressive
  131. # [09:31] <hsivonen> nessy: what capture software do you use to generate the green circles indicating clicks?
  132. # [09:32] * Quits: zalan (n=kvirc@catv-89-133-232-199.catv.broadband.hu) (Read error: 60 (Operation timed out))
  133. # [09:38] * Quits: harig (n=opera@59.90.71.35) (Read error: 110 (Connection timed out))
  134. # [09:42] * Joins: annevk (n=opera@fnttkyo029007.tkyo.fnt.ftth2.ppp.infoweb.ne.jp)
  135. # [09:42] * Joins: zalan (n=kvirc@catv-89-133-232-199.catv.broadband.hu)
  136. # [09:45] <hsivonen> aargh. I already have a validator bug related to xmlns:foo
  137. # [09:49] * hsivonen gestures angrily in the general direction of Namespaces
  138. # [09:50] * Quits: pesla (n=retep@procurios.xs4all.nl) (Read error: 110 (Connection timed out))
  139. # [09:51] <Philip`> If you would just stop writing bugs, it wouldn't be a problem at all
  140. # [09:53] * Joins: svl (n=me@69.Red-88-2-205.staticIP.rima-tde.net)
  141. # [09:53] <zcorpan> Hixie: the first "ABC Company" example should be marked as "bad"
  142. # [09:54] * Joins: danbri (n=danbri@77.167.54.81)
  143. # [10:01] * Quits: MikeSmith (n=MikeSmit@EM114-48-39-212.pool.e-mobile.ne.jp) (Read error: 110 (Connection timed out))
  144. # [10:06] * Quits: svl (n=me@69.Red-88-2-205.staticIP.rima-tde.net) ("And back he spurred like a madman, shrieking a curse to the sky.")
  145. # [10:09] * Quits: Lachy (n=Lachlan@85.196.122.246) ("Leaving")
  146. # [10:10] * Joins: Lachy (n=Lachlan@85.196.122.246)
  147. # [10:10] * Quits: Lachy (n=Lachlan@85.196.122.246) (Client Quit)
  148. # [10:10] * Joins: Lachy (n=Lachlan@85.196.122.246)
  149. # [10:13] <Hixie> zcorpan: k, thx
  150. # [10:16] <zcorpan> are webkit's and firefox's impl of canvas text "incomplete" or "complete but buggy"?
  151. # [10:18] * Joins: pauld (n=pauld@92.40.186.226.sub.mbb.three.co.uk)
  152. # [10:19] * Quits: tantek (n=tantek@adsl-63-195-114-133.dsl.snfc21.pacbell.net)
  153. # [10:19] <zcorpan> ok i'm done annotating now. they were rough guesses but it's better than nothing, i hope
  154. # [10:19] <hsivonen> argh. the isindex bug persists
  155. # [10:19] <annevk> Hixie, one issue Ruby raised a few times was making <meta charset> conforming in XHTML if it matched the charset of the document
  156. # [10:20] <hsivonen> annevk: I categorize that under violations of the DOM Consistency principle being bad
  157. # [10:23] <Lachy> we could allow <meta charset> in XHTML if it was set to either UTF-8, UTF-16, UTF-16LE or UTF-16BE; or if it matched the encoding specified by the XML declaration or higher level protocol, like the HTTP content-type header
  158. # [10:24] * Philip` presumes it also falls under the Metadata That Looks Like It'll Do Something But Actually It'll Be Totally Ignored Thus Wasting Authors' Time principle (like @profile)
  159. # [10:24] <annevk> @profile is different
  160. # [10:24] <hsivonen> Philip`: indeed. Which is why what Lachy says would only work for authors who validate often but would cause trouble for everyone else
  161. # [10:25] <Philip`> hsivonen: It would also work for authors who don't validate at all but send everything as UTF-8 all the time and use <meta charset="utf-8"/>
  162. # [10:25] <Philip`> which is probably quite a high proportion of the people who serve XHTML
  163. # [10:25] <hsivonen> @profile is an epic waste of time no matter what. However, if the WG doesn't spent its time on it, the aggregate waste over time will be much larger.
  164. # [10:26] <Lachy> at least allowing <meta charset> in XHTML would actually be addressing a real use case
  165. # [10:26] <Lachy> unlike @profile
  166. # [10:27] * jgraham forgets the usecase
  167. # [10:27] <Lachy> jgraham, allowing polyglot documents to be served as either HTML or XHTML without modification and without relying on HTTP content type headers
  168. # [10:27] <jgraham> Yes
  169. # [10:28] <Lachy> rubys said that's what Venus does
  170. # [10:28] <Philip`> jgraham: Software that produces bytes that someone else (who doesn't want to do any fancy reserialiation) might serve as HTML or XHTML, that will retain the correct charset even when it's e.g. saved to disk and separated from its HTTP charset
  171. # [10:29] <jgraham> I suppose polyglot documents are a valid usecase
  172. # [10:30] * Joins: webben (n=webben@dip5-fw.corp.ukl.yahoo.com)
  173. # [10:30] <Philip`> They're a use case which HTML5 already goes to some lengths to support, so it'd be inconsistent to not continue supporting them
  174. # [10:30] <Lachy> hmm, in that case, maybe we would only need to allow it to be specified as UTF-8/UTF-16*, because if you're specifying the charset with the xml decl. then you're not making a polyglot document
  175. # [10:31] <Lachy> ... and doing anything else doesn't work too well when separated from the HTTP content-type headers
  176. # [10:31] <Philip`> Lachy: Can't Content-Type still set the charset to anything (not just UTF), if there's no XML declaration?
  177. # [10:31] <Lachy> Philip`, yeah, but see what I just wrote
  178. # [10:31] <zcorpan> Philip`: if you can set content-type you don't need <meta charset>
  179. # [10:32] <Philip`> I suppose that makes sense
  180. # [10:32] <Lachy> we may not even need to allow UTF-16, since that has the BOM anyway
  181. # [10:33] <hsivonen> zcorpan: though being able to vary the content-type without being able to also set charset is a very fringe case
  182. # [10:34] <Philip`> hsivonen: Not really - just upload an .html vs .xhtml file, and that'll change the content-type
  183. # [10:34] <Philip`> (or have an .html and .xhtml file on your local disk)
  184. # [10:35] <hsivonen> ok
  185. # [10:35] * hsivonen notes that this is can already be done by using the UTF-8 BOM
  186. # [10:36] <Lachy> hsivonen, sometimes, including the UTF-8 BOM isn't always easy and can create problems with server side tools
  187. # [10:36] * Philip` would never trust a text editor to not silently strip the BOM
  188. # [10:37] <Lachy> e.g. if you begin a PHP file with the UTF-8 BOM, then it sends it out and prevents you from calling header() to set any other headers.
  189. # [10:37] * hsivonen would trust TextWrangler
  190. # [10:37] <hsivonen> Lachy: so you'd need to produce the BOM by other means than literal inclusion in PHP
  191. # [10:38] <Lachy> the only safe way to get a BOM output from such a system is to do: echo "\EF\BB\BF"
  192. # [10:38] <hsivonen> right
  193. # [10:38] <Lachy> (if I got that syntax right)
  194. # [10:38] * Joins: pauld_ (n=pauld@92.40.28.229.sub.mbb.three.co.uk)
  195. # [10:38] <Philip`> I guess any server-side templating system which reads a template, parses it as XML, does some substitutions, then prints it out again, would be hard to make output a BOM if you're the person who's just editing the templates
  196. # [10:39] <zcorpan> <meta charset="utf-8"/> is easier to work with than a bom
  197. # [10:39] <Lachy> but that requires authors to have to look up the byte sequence for the UTF-8 BOM, which is harder than just typing <meta charset="UTF-8"/>
  198. # [10:39] <hsivonen> true. I just observed that there is an unobvious and brittle way to address the case already.
  199. # [10:41] <Philip`> The use case could also be addressed by outputting ASCII and using &#nnn;
  200. # [10:42] <zcorpan> Philip`: that might be safer to do anyway
  201. # [10:42] * Joins: virtuelv (n=virtuelv@pat-tdc.opera.com)
  202. # [10:43] <zcorpan> since for rubys' purpose the content might well end up being mislabeled anyway
  203. # [10:43] <Philip`> Inefficient for non-English-speakers, though
  204. # [10:43] <zcorpan> efficient enough for swedish, but yeah
  205. # [10:44] <annevk> PHP doesn't do BOM
  206. # [10:44] <annevk> it chokes on it (at least for version 4 it did, maybe it's fixed now)
  207. # [10:44] <hsivonen> with PHP, you are in control of headers, on the other hand
  208. # [10:45] <annevk> if you know how that works
  209. # [10:45] <Philip`> zcorpan: Still probably three times worse than UTF-8 for any funny Swedish characters, though
  210. # [10:45] <hsivonen> well, yeah, pretty much everything depends on knowing what to do
  211. # [10:45] <annevk> I expect that most people that work with PHP have limited knowledge of headers
  212. # [10:46] <annevk> s/headers/HTTP/
  213. # [10:46] <Philip`> Headers are things like <META>, aren't they?
  214. # [10:46] * Quits: pauld_ (n=pauld@92.40.28.229.sub.mbb.three.co.uk)
  215. # [10:47] <zcorpan> Philip`: yes but for a normal document that's still not so much
  216. # [10:47] <hsivonen> Swedish characters are two bytes in UTF-8 and 5 bytes as NCRs in the best case
  217. # [10:49] * Philip` can imagine implementations in which it's a pain to convert UTF-8 to NCRs, e.g. if all your strings are UTF-8 encoded in a database and you just get bytes back and print them straight out, and don't have an easy way to do substitutions on characters instead of bytes
  218. # [10:50] <zcorpan> Philip`: store it as NCRs in the database ;)
  219. # [10:52] <hsivonen> Hixie: fwiw, I just noticed that the tree builder stack already has mutable data on it beyond the element nodes
  220. # [10:52] * Quits: Lachy (n=Lachlan@85.196.122.246) ("This computer has gone to sleep")
  221. # [10:52] <hsivonen> Hixie: table elements have a taint bit
  222. # [10:52] <hsivonen> for dealing with foster parenting spaces and other text
  223. # [10:52] * Philip` is thinking of e.g. http://golem.ph.utexas.edu/~distler/code/instiki/svn/annotate/head%3A/lib/stringsupport.rb where is_utf8? indicates that all the string-processing is done on bytes
  224. # [10:53] * Joins: enki_ (n=enki@c-67-188-42-45.hsd1.ca.comcast.net)
  225. # [10:54] <Philip`> (Hmm, that code is going to think &#x26;#65536; is invalid utf8?)
  226. # [10:54] * Quits: enki (n=enki@c-67-188-42-45.hsd1.ca.comcast.net) (Read error: 110 (Connection timed out))
  227. # [10:54] <Philip`> s/65536/65535/
  228. # [10:54] <hsivonen> Hixie: was your foster parenting change editorial?
  229. # [10:56] <hsivonen> Hixie: I'm trying to work out if there was any change other than addressing the case where a script has moved the table into a DocumentFragment
  230. # [11:00] * Quits: pauld (n=pauld@92.40.186.226.sub.mbb.three.co.uk) (Connection timed out)
  231. # [11:05] <hsivonen> Philip`: did you have a good way of foster parenting text that doesn't cause single-char text nodes and that doesn't keep reallocating the buffer of a pre-existing text node?
  232. # [11:06] <hsivonen> in general, I don't want to fetch an existing text node back from the DOM and append to it
  233. # [11:06] <hsivonen> seems like asking for trouble in crazy script scenarios
  234. # [11:08] * Joins: ROBOd (n=robod@89.122.216.38)
  235. # [11:09] <hsivonen> should I try to buffer a run of foster-parented text but not coalesce it with pre-existing non-foster-parented text?
  236. # [11:10] <Philip`> hsivonen: I think the idea was that you would only have to extend a text node if it was the last node you inserted into the document, which (hopefully) means you can delay inserting it into the document and just build up the string in memory, and then when you next insert a non-text node you go back and insert that combined text node first
  237. # [11:11] <hsivonen> OK. so I need to be able to foster-parent upon flushing :-(
  238. # [11:11] <Philip`> hsivonen: "... if that node has a child immediately before where the character is to be inserted, and that child is a Text node, and that Text node was the last node that the parser inserted into the document, then the character must be appended to that Text node" - the "last node" bit is relevant
  239. # [11:11] <hsivonen> ok
  240. # [11:11] <hsivonen> I guess now is as good time as any to finally fix this
  241. # [11:11] <Philip`> I've never actually tried implementing that, so I don't know if it's feasible in practice
  242. # [11:12] * Joins: Lachy (n=Lachlan@pat-tdc.opera.com)
  243. # [11:12] <Philip`> but at some point in the past it seemed to me like it'd possibly work okay (and it would avoid the quadratic performance issues)
  244. # [11:12] <hsivonen> I don't like the idea of making flush check a flag, but I guess it's peanuts in the grand scheme of things
  245. # [11:12] * Joins: tndH (n=Rob@james-baillie-pc083-014.student-halls.leeds.ac.uk)
  246. # [11:13] * hsivonen feels good about writing code instead of email ratholing today
  247. # [11:13] <Philip`> You mean the flag can trigger anaphylaxis?
  248. # [11:14] <nessy> hsivonen: I work on the mac an my capture software is iShowU
  249. # [11:14] <nessy> it lets you set up circles like that
  250. # [11:14] <hsivonen> nessy: thanks. (I use Snapz Pro X)
  251. # [11:15] <nessy> but I would prefer if it didn't use circles on everything, e.g. click on scrollbar
  252. # [11:15] <Philip`> (I guess your Java implementation has the same problem that Python/JS/etc ones would, with <table>aaaaaaaaaa having O(n^2) performance if it was appending each character to the text node's immutable string?)
  253. # [11:15] <hsivonen> Philip`: right
  254. # [11:16] <hsivonen> Philip`: Gecko's strings are mutable, but IIRC, the DOM tries to not have dead weight buffers
  255. # [11:17] <Philip`> (Non-linear performance is pretty annoying, e.g. when someone accidentally hit an exponential case in a regexp in the html5lib sanitiser)
  256. # [11:18] <Philip`> (so I suppose html5lib should be careful to avoid these quadratic things too, because people rely on the library and don't want it to suffer from DOS attacks)
  257. # [11:18] <hsivonen> which reminds me that all my growing buffers don't have upper limits against malicious content
  258. # [11:18] <hsivonen> like <aaaaaaa... with a script serving an unbounded number of 'a's
  259. # [11:20] * Quits: roc (n=roc@121-72-173-196.dsl.telstraclear.net)
  260. # [11:20] <Philip`> The longest tag name I see is 594 characters
  261. # [11:21] <Philip`> <metaname"keywords"content="agorà,provincie,regioni,comuni,istituzioni,italia,commercio,sport,dante,libertà,liberale,liberalismo,liberismo,parlamento,camera,senato,capitalismo,free,informazioni,editoria,associazioni,formazione,forum,sicurezza,cultura,europa,comunicazione,comuni,circuito,stampa,manifestazioni,iscrizioni,rete,società,pli,elezioni,partiti,politica,politiche,politici,croce,einaudi,marx,berlusconi,d'alema,rutelli,ulivo,polo,laici,occupaz
  262. # [11:21] <hsivonen> Validator.nu protects itself by limiting the length of the network stream, but I need something else for Gecko
  263. # [11:21] <Philip`> (I guess you don't want to fatally abort when pages do silly things like that, but I'm not sure that anyone would object to truncation)
  264. # [11:22] <Philip`> hsivonen: Isn't Gecko equally vulnerable to malicious content like <script>a='a';while(1)a+=a</script> ?
  265. # [11:23] <hsivonen> Philip`: I haven't tried
  266. # [11:23] <Hixie> annevk: i'm sure i've acknowledged that one many times
  267. # [11:26] <Hixie> hsivonen: and the taint bit proved to be a huge pain in the ass :-)
  268. # [11:26] <Hixie> hsivonen: which revision?
  269. # [11:26] * Joins: pauld (n=pauld@host81-130-24-71.in-addr.btopenworld.com)
  270. # [11:26] <Hixie> re <meta charset>, lachy raises an interesting point
  271. # [11:27] <hsivonen> Hixie: the one that changed the paragraph starting with "The foster parent element is"
  272. # [11:27] * hsivonen tries to look up the rev
  273. # [11:27] <Hixie> since there can't be an xml decl, and since we're assuming the site can't set the http headers to give the encoding (otherwise we wouldn't need the meta charset), we have to assume that the document is either utf-8 or utf-16.
  274. # [11:28] <Hixie> in which case, the meta charset isn't needed since you can just use a bom
  275. # [11:28] <Hixie> which is already allowed
  276. # [11:28] <zcorpan> Hixie: bom vs <meta> was also discussed
  277. # [11:29] <hsivonen> Hixie: r2732
  278. # [11:29] <Lachy> Hixie, after having thought about it more, I would allow only <meta charset="UTF-8"/> in XHTML. Anything else should be an error
  279. # [11:30] <Philip`> BOMs are invisible metadata
  280. # [11:30] <hsivonen> Even after a second look, I can't figure out the non-editorial aspect of r2732 if there is one
  281. # [11:31] <Hixie> i guess i could be convinced that we could allow it for utf-8 only, since in that case it doesn't do anything in xml anyway and all it does in text/html is make the default match xml
  282. # [11:31] <Hixie> hsivonen: looking
  283. # [11:33] * jgraham thinks that anything that relys on BOMs is bad
  284. # [11:34] <virtuelv> oh, BOMs, those things you can never rely on tools outputting
  285. # [11:34] * hsivonen thinks that UTF-16 *without* a BOM is bad
  286. # [11:35] * Joins: annevk2 (n=opera@fnttkyo029008.tkyo.fnt.ftth2.ppp.infoweb.ne.jp)
  287. # [11:35] <Lachy> relying on a UTF-8 BOM is bad. Relying on UTF-16 BOM is good.
  288. # [11:35] <Hixie> UTF-16 is general is bad...
  289. # [11:35] <Hixie> :-)
  290. # [11:35] <jgraham> hsivonen: It may be bad to output it without a BOM but since people generally can't see the BOM, it is hard for them to tell if they got it right
  291. # [11:35] <Philip`> Watch out for unexploded BOMs
  292. # [11:36] <jgraham> So BOMs rely on a "tools will save us" kind of argument
  293. # [11:36] <Hixie> hsivonen: it's not an editorial change.
  294. # [11:36] <hsivonen> Hixie: so what's the change?
  295. # [11:37] <Hixie> hsivonen: the foster parent element is no longer defined in terms of its parent, only in terms of the stack.
  296. # [11:37] <Lachy> jgraham, I'm not aware of any UTF-16 capable editor that doesn't always output the BOM
  297. # [11:37] <Lachy> and it's non-trivial to strip it
  298. # [11:37] <Lachy> but that does make UTF-16 bad for use in server side includes
  299. # [11:38] <jgraham> I would be more convinced by a survey of billions of web pages showing that all UTF-16 ones always have BOMs :)
  300. # [11:38] <Hixie> hsivonen: the non-fragment case was changed from having two conditions -- if parent, if no parent -- to always doing what the original "no parent" case did
  301. # [11:38] <hsivonen> Hixie: the definition still first uses the parent of the table if there is one
  302. # [11:38] <Hixie> it does?
  303. # [11:39] <Hixie> how so?
  304. # [11:39] <Hixie> is there some other text somewhere that i missed?
  305. # [11:39] <hsivonen> Hixie: or then "the parent of foo in the stack of open element" is a really unobvious way of saying you don't actually care about the actual parent
  306. # [11:40] <Hixie> but the spec doesn't say that unless i'm misreading it
  307. # [11:40] <Hixie> the definition of "foster parent element" only uses the term "parent" in the term "foster parent element" itself
  308. # [11:40] <hsivonen> now this is embarassing. I'm reading my diff window the wrong way round
  309. # [11:40] <Hixie> heh
  310. # [11:41] <Hixie> it happens :-)
  311. # [11:42] <hsivonen> The simplification doesn't really simplify things much though, since you still need to check if the foster parent is the actual parent to decide if you insertBefore or appendChild
  312. # [11:42] <Hixie> yeah but you don't have to do that in the parser thread
  313. # [11:42] <zcorpan> hsivonen: your diff view should probably have fancy colors to avoid that mistake :)
  314. # [11:42] * virtuelv wonders just how many UTF-16 documents there actually are
  315. # [11:42] <Hixie> you can do that off on the main thread
  316. # [11:42] <Lachy> Hixie, would you like the meta charset discussion summarised into an email so you don't forget about it?
  317. # [11:42] <Hixie> Lachy: or a bug, yeah.
  318. # [11:42] * Quits: annevk (n=opera@fnttkyo029007.tkyo.fnt.ftth2.ppp.infoweb.ne.jp) (Read error: 110 (Connection timed out))
  319. # [11:42] <Lachy> ok, I'll bug it
  320. # [11:42] <Hixie> thanks
  321. # [11:43] <Hixie> i hope it's what sam meant
  322. # [11:43] <hsivonen> Hixie: thanks
  323. # [11:43] <Hixie> hsivonen: does that resolve that issue to your satisfaction?
  324. # [11:43] <Hixie> if it is what sam meant, i'm still surprised that i've never actually replied to that issue
  325. # [11:44] <Hixie> i'm sure he's raised it on the whatwg before
  326. # [11:44] <hsivonen> Hixie: either way works. the old way just would have required passing one extra pointer to the main thread
  327. # [11:44] <Philip`> Out of the five UTF-16 pages in my collection, all have a DOM
  328. # [11:45] <Hixie> the last few times he's mentioned it has always been on the public-html list in passing as something where he didn't get his way, i just assumed that meant that the issue had been resolved
  329. # [11:45] <Hixie> oh well
  330. # [11:45] <Philip`> (5 out of 130K)
  331. # [11:45] <Hixie> hsivonen: k
  332. # [11:45] <Philip`> Uh
  333. # [11:45] <Philip`> s/DOM/BOM/
  334. # [11:45] <zcorpan> Hixie: s/painter/pointer/ (in the load algoritm)
  335. # [11:45] <Hixie> Philip`: any of them have internal declarations claiming to be UTF-8? :-)
  336. # [11:45] * Philip` assumed Sam was referring to his namespaces-in-HTML proposal
  337. # [11:46] <hsivonen> Hixie: actually, they both require the same number of pointers to be passed
  338. # [11:46] * jgraham thinks that 5 might be in the realm of samll number statistics :)
  339. # [11:46] <Philip`> One says charset="unicode"
  340. # [11:46] <Hixie> Philip`: i've _definitely_ sent replies back on _that_ subject
  341. # [11:46] <ap> Lachy: did I read it correctly that no one is proposing to honor <meta charset> in XHTML?
  342. # [11:46] <Hixie> zcorpan: thanks
  343. # [11:46] <Philip`> http://www.hotellikarhu.com/ says utf-8
  344. # [11:46] <hsivonen> Hixie: so I'm OK with either, but if you'd prefer reverting to the old way, I'd prefer knowing now
  345. # [11:47] <Hixie> hsivonen: i'm fine with the text now. it's simpler.
  346. # [11:47] <annevk2> ap, yes
  347. # [11:47] <hsivonen> Hixie: ok. thanks
  348. # [11:47] <Hixie> hsivonen: and still doesn't throw content away.
  349. # [11:47] <ap> annevk2: good, thx
  350. # [11:47] <Hixie> hsivonen: (which your original proposal would have)
  351. # [11:47] <Hixie> ap: don't worry, that would be a layer violation, and we won't do that
  352. # [11:48] <Hixie> we only violate layers when legacy content forces us to
  353. # [11:48] <Philip`> Hmm, one of the five isn't utf-16, so I've got no idea why my grep thought it was
  354. # [11:48] <Philip`> http://www.prhoffman.com/site-map.htm says no charset at all
  355. # [11:49] <hsivonen> Hixie: validating Lachy's proposed change would violate layers, though, but the violation in the infrastructure is already needed for URL validation
  356. # [11:49] <Philip`> http://www.archimage.net/silvia2.html asserts via HTTP that it's iso-8859-1
  357. # [11:49] <Philip`> and via <meta> claims it's charset=unicode
  358. # [11:50] <Philip`> http://www.rimetea-torocko.com/ claims it's charset=iso-8859-1
  359. # [11:51] <Philip`> Oops, it's more than five, some just scrolled off my screen
  360. # [11:51] <Lachy> hsivonen, how would it violate the layers? All you would need to check is that it's "UTF-8" (case insensitive) and that the document's encoding is UTF-8 too.
  361. # [11:52] <hsivonen> Lachy: in proper layering, the application layer doesn't know the encoding of the XML document
  362. # [11:52] <Lachy> ok
  363. # [11:52] <Philip`> http://www.oregongolf.com/ - aha, there's one without a BOM
  364. # [11:53] <Philip`> http://www.southern-jet.com/ claims to be iso-8859-1
  365. # [11:56] <Philip`> Oh, so it's actually 30 pages that look like UTF-16
  366. # [11:56] <Philip`> which is a bit more than 5
  367. # [11:56] <jgraham> And at least one has no BOM
  368. # [11:57] <jgraham> Which is like 3 +/- a lot %
  369. # [11:58] * Philip` attempts to upload the list, before realising he's connected to the wrong Dreamhost server and so the files he uploads aren't actually accessible from the web
  370. # [12:00] <Philip`> http://philip.html5.org/data/pages-that-look-like-utf16.txt - that's better
  371. # [12:00] <Philip`> If anyone wants to bother looking through all those pages and working out which ones have BOMs, feel free :-)
  372. # [12:00] <Philip`> but at least there's definitely one without
  373. # [12:00] <Philip`> (These are the pages matching (?i)\x00[a-z]\x00[a-z]\x00, minus .jpg and .JPG and .doc and .wma files)
  374. # [12:14] <zcorpan> this seems useful http://blog.nihilogic.dk/2009/02/html5-canvas-cheat-sheet.html
  375. # [12:19] * Parts: annevk2 (n=opera@fnttkyo029008.tkyo.fnt.ftth2.ppp.infoweb.ne.jp)
  376. # [12:19] <Hixie> ok, dealt with all the video feedback
  377. # [12:20] <Hixie> (other than caption stuff)
  378. # [12:21] <zcorpan> Hixie: thanks :)
  379. # [12:21] <zcorpan> looks good on first read
  380. # [12:23] <Lachy> it's interesting that the table on this page uses summary="" itself, but it contains information that would have been useful to me if I didn't have to look at the source to see it. http://www.paciellogroup.com/blog/misc/summary.html
  381. # [12:25] <jgraham> That use of summary doesn't comply with WCAG2 AIUI
  382. # [12:26] <zcorpan> seems like that should be before or after the table labelled as "Conclusion:"
  383. # [12:26] <jgraham> Although I don't really understand WCAG2
  384. # [12:27] <jgraham> (specifically it doens't use the summary as indicated in http://www.w3.org/TR/WCAG20-TECHS/H73.html but I don't know if that is just supposed to be a suggestion or what)
  385. # [12:27] <zcorpan> jgraham: that's non-normative :)
  386. # [12:28] <jgraham> zcorpan: Does WCAG have any normative conditions?
  387. # [12:28] <Hixie> well in a shocking turn of events, i've actually run out of urgent feedback that isn't blocked on other stuff
  388. # [12:28] <jgraham> In any case violating non-normative suggestions about how to best use the attribute in an article designed to show that the attribute is used correctly is... interesting
  389. # [12:29] <Lachy> Hixie, take a holiday for a few weeks and relax :-)
  390. # [12:29] <zcorpan> jgraham: yes, http://www.w3.org/TR/2008/REC-WCAG20-20081211/#content-structure-separation-programmatic is normative
  391. # [12:30] <Hixie> i was thinking that i could do the editorial stuff sam was saying i should do
  392. # [12:30] <Lachy> Hixie, bug 6613 isn't blocked by anything
  393. # [12:30] <Hixie> i haven't looked at the bugs
  394. # [12:30] <Hixie> are any of them urgent?
  395. # [12:30] <Lachy> I don't know. That's the one I just filed.
  396. # [12:32] <zcorpan> Hixie: my issues! address my issues first!
  397. # [12:32] <zcorpan> :)
  398. # [12:32] <Hixie> if you have any that affect implementations, let me know
  399. # [12:32] <Hixie> if they're bugs, mark them P1 or critical (or both)
  400. # [12:34] * Quits: dave_levin (n=dave_lev@c-98-203-247-78.hsd1.wa.comcast.net)
  401. # [12:35] <zcorpan> http://www.w3.org/Bugs/Public/show_bug.cgi?id=6336 seems fixed in the spec already
  402. # [12:35] <Hixie> cool
  403. # [12:39] <zcorpan> http://www.w3.org/Bugs/Public/show_bug.cgi?id=6438 and http://www.w3.org/Bugs/Public/show_bug.cgi?id=6462 and http://www.w3.org/Bugs/Public/show_bug.cgi?id=6494 would probably make pf people happy
  404. # [12:40] <hsivonen> In my opinion, it doesn't make sense to stir things around bug 6494 before the group speccing ATAG decides what they want authoring tools to do
  405. # [12:41] <zcorpan> http://www.w3.org/Bugs/Public/show_bug.cgi?id=6353 seems fixed now
  406. # [12:42] <zcorpan> hsivonen: 6494 doesn't suggest any normative changes but i guess it might stir things anyway
  407. # [12:42] * Quits: pergj (n=pergj@home.kvaleberg.no) (Read error: 113 (No route to host))
  408. # [12:44] <Hixie> i guess we do want to disallow cyclic references in headers="", huh
  409. # [12:44] <Hixie> even though it doesn't cause any harm algorithm-wise
  410. # [12:44] <Hixie> hsivonen: you ok with checking for this and flagging it?
  411. # [12:46] <hsivonen> Hixie: yeah, eventually, but it'll probably wait on other higher-priority stuff
  412. # [12:46] <Hixie> k
  413. # [12:52] <Hixie> holy crap, a CFL light just blew on me
  414. # [12:53] <Hixie> i've only had it two years!
  415. # [12:53] <Hixie> what kind of crap is this!
  416. # [12:53] <zcorpan> Hixie: time for a new blog post?
  417. # [12:53] <Hixie> hah
  418. # [12:54] <Hixie> i don't care as much about CFL lights as about my LED bike lights :-)
  419. # [12:54] * Joins: xydyx (n=hdh@58.187.21.120)
  420. # [12:54] <hsivonen> Hixie: I just witnessed one go bad in under a month
  421. # [12:55] <Hixie> under a month! it was probably still under warranty!
  422. # [12:55] <Lachy> presumably you can only claim warranty if you kept the receipt. But who keeps those for everything?
  423. # [12:56] * Quits: enki_ (n=enki@c-67-188-42-45.hsd1.ca.comcast.net) (Read error: 110 (Connection timed out))
  424. # [12:56] <hsivonen> I might complain if I have another reason to visit the seller premises in the near future
  425. # [12:56] <zcorpan> Lachy: i do for things i might find broken and want to return
  426. # [12:57] <Hixie> hsivonen: you really think i shouldn't do 6494 yet?
  427. # [12:57] <hsivonen> Hixie: yes
  428. # [12:58] <Hixie> did steve just point-blank admit to trolling on public-html
  429. # [12:58] <Hixie> or am i seeing things
  430. # [12:58] <hsivonen> Hixie: I think we should wait for concrete ATAG 2.0 draft text and then make machine-checkable conformance broad enough to make conforming all outputs of ATAG 2.0-recommended authoring tool behavior
  431. # [12:59] <Hixie> you had me at "yes"
  432. # [13:00] <hsivonen> Hixie: either he is being sarcastic or he is admitting trolling
  433. # [13:02] * Quits: pauld (n=pauld@host81-130-24-71.in-addr.btopenworld.com)
  434. # [13:06] <zcorpan> or he implied "... according to WCAG 2 techniques with which i disagree"
  435. # [13:06] * Joins: pergj (n=pergj@home.kvaleberg.no)
  436. # [13:06] * Quits: webben (n=webben@dip5-fw.corp.ukl.yahoo.com) (Read error: 104 (Connection reset by peer))
  437. # [13:06] <zcorpan> assuming we're talking about the same thing here :)
  438. # [13:07] <hsivonen> zcorpan: wouldn't that be trolling?
  439. # [13:07] <Lachy> zcorpan, I think the trolling was where he said "I used the summary in this way on prupose, because i figured that people would look at it and comment on its incorrect use."
  440. # [13:08] <hsivonen> so who *does* agree with the WCAG advice here?
  441. # [13:08] * Joins: virtuelv_ (n=virtuelv@213.236.208.247)
  442. # [13:09] <hsivonen> Philip`: I'll try to make SAX Tree and XOM tree builders not suck perf-wise when there's a lot of stuff to foster parent
  443. # [13:09] <hsivonen> we'll see if I break something...
  444. # [13:09] <zcorpan> hsivonen: uh yeah i guess it would
  445. # [13:10] <Lachy> considering the actual conformance criteria in WCAG2 are so vague and incomprehensible without reading the Understanding and Techniques documents, choosing to ignore the advisory techniques you disagree with seems to leaves them open to a lot of interpretation
  446. # [13:10] <Lachy> s/seems to leaves/leaves/
  447. # [13:11] <hsivonen> Lachy: yeah, I think that's a pretty significant flaw. particularly in a document that clearly has been written as basis for litigation
  448. # [13:11] <Lachy> "There may also be cases where it may be a judgment call about what information should appear in text and what would need to be directly associated, and it may be most appropriate to duplicate some information (for instance, in an HTML table, providing the summary both in the paragraph before the table and in the summary attribute of the table itself). However, wherever possible it is necessary for the information to be programmatically determined rath
  449. # [13:11] <Lachy> er than providing a text description before encountering the table."
  450. # [13:11] <Lachy> -- http://www.w3.org/TR/UNDERSTANDING-WCAG20/content-structure-separation-programmatic.html
  451. # [13:12] <Lachy> So WCAG2 recommends dupicating information just to achieve programmatic association?!
  452. # [13:12] <hsivonen> I love the use of "wherever possible" in the light of comments on the use of "when possible" in HTML Design Principles
  453. # [13:12] <Lachy> LOL
  454. # [13:13] <Lachy> well, this is an informative document. WCAG 1 had "when possible" in normative criteria
  455. # [13:14] * zcorpan doesn't think it makes sense to have normative criteria for "guidelines" in the first place
  456. # [13:15] <zcorpan> or the title should have been WCAR
  457. # [13:16] <hsivonen> it seems to me WCAG family of documents wants to have its cake and eat it too when it comes to normativity
  458. # [13:17] <zcorpan> hmm autobuffer="" seems nice, i had only thought about using heuristics
  459. # [13:18] * Philip` wonders if http://www.whatwg.org/issues/top has turned out to be of significant value or not
  460. # [13:18] <Hixie> not enough people voted on issues for it to be valuable
  461. # [13:18] * hsivonen wonders if it's bad for SAX Tree to have a tree node per character for foster-parented text
  462. # [13:19] * zcorpan wonders what a tree node is
  463. # [13:19] * Philip` wonders what a character is
  464. # [13:19] <hsivonen> if I fix the O(n^2) search behavior
  465. # [13:19] <hsivonen> Philip`: UTF-16 code unit
  466. # [13:20] <hsivonen> zcorpan: a node is an object with 7 pointers and two ints
  467. # [13:21] <Philip`> (<table>aaaaaa is presumably easier to handle than <table>aaa</td>aaa because in the former case you could just pass a string of consecutive characters into the tree builder and treat them as a single unit)
  468. # [13:21] <hsivonen> seems like no big deal in usual cases and bad in the kind of cases Philip` finds on the Web
  469. # [13:22] <Philip`> The case where I had problems was because a bug in my code put a megabyte of null bytes after a <table> - I don't think I encountered any real web content with the same problem
  470. # [13:22] <hsivonen> ok. then I don't optimize further than eliminating the O(n^2) behavior
  471. # [13:23] * Quits: virtuelv (n=virtuelv@pat-tdc.opera.com) (Read error: 110 (Connection timed out))
  472. # [13:24] <Philip`> It seems a bit bad that I could cause you to use ~576 bytes of memory for every byte I send over the network
  473. # [13:25] <Lachy> Hixie, how about also adding a note to clarify that <meta charset> doesn't actually have any effect in XHTML even if set to another value
  474. # [13:25] <hsivonen> Philip`: yeah. the alternative is adding the complexity of early coalescing...
  475. # [13:26] <Philip`> Hixie: "There must not be more than one element with a charset attribute per document." sounds a bit misinterpretable, because you can have <script charset/><script charset/> in a document, so maybe it should be clarified to say "meta element"
  476. # [13:26] <hsivonen> which would have the benefit of saving per code-unit method invocation overhead with all tree builders
  477. # [13:27] <hsivonen> foster parenting is such an annoyance
  478. # [13:27] <Lachy> basically, something to preempt the bad advice about needing to use it to specify the encoding of XHTML, which we've already seen plenty of for XHTML1
  479. # [13:28] * jgraham is now thoroughly confused about when it is OK to disagree with various parts of WCAG
  480. # [13:28] <jgraham> Does it make a difference if one is a self-proclaimed accessbility expert or not?
  481. # [13:29] <Philip`> hsivonen: Would early coalescing still result in <table>a</td>a</td>a... using ~576 bytes of memory for every 5 bytes I send over the network?
  482. # [13:29] <Philip`> (where '5 bytes' means 'approximately 1 bit, if I use compression')
  483. # [13:29] <Hixie> fixed the charset rules
  484. # [13:29] * Joins: enki (n=enki@c-67-188-42-45.hsd1.ca.comcast.net)
  485. # [13:30] <Hixie> thanks for the feedback
  486. # [13:30] <hsivonen> Philip`: what's your pointer size if you get 576 instead of 120 or 92?
  487. # [13:31] <hsivonen> Philip`: but yes
  488. # [13:31] <hsivonen> Philip`: note that you can already make SAX Tree use a lot of memory by sending <b>b</b>a<b>b</b>...
  489. # [13:31] <Philip`> "the resource must be encoded using the UTF-8 character encoding" - what does that mean? Could I write a file containing <meta charset="utf-8"/> and encode the characters as UTF-8, and then serve it as application/xhtml+xml;charset=iso-8859-1? I still encoded the resource as UTF-8
  490. # [13:31] <Philip`> hsivonen: Um...
  491. # [13:31] <Philip`> hsivonen: 64 bytes
  492. # [13:32] <hsivonen> Philip`: my pointer size is 64 *bits* ;-)
  493. # [13:32] <Philip`> hsivonen: I realise that now :-)
  494. # [13:32] <Hixie> Philip`: that would be a violation of some other rule
  495. # [13:33] <Philip`> hsivonen: so it'd be like 7*64 + 2*32 bits, which is like 64 bytes, plus a little bit of overhead for the JVM, I guess
  496. # [13:33] <jgraham> hsivonen: When we have 512bit cpus then Philip` will look prophetic
  497. # [13:33] <Philip`> Hixie: Even if the document is perfectly valid iso-8859-1?
  498. # [13:34] <Hixie> Philip`: if it's actually US-ASCII, and thus both UTF-8 and ISO-8859-1, I don't really care if there's a mismatch.
  499. # [13:34] <hsivonen> Philip`: on of the requirements for SAX Tree is being able to reconstuct SAX Locator data upon playback
  500. # [13:34] <hsivonen> oops it's 7 pointers and 4 ints
  501. # [13:35] <Philip`> Hixie: What if it's not US-ASCII, but still is both UTF-8 and ISO-8859-1?
  502. # [13:35] <hsivonen> plus JVM overhead, which I assume to be at least two pointers and one int (considering that there's an array involved)
  503. # [13:35] <Philip`> (with different meaning in each)
  504. # [13:36] * hsivonen can't count anymore
  505. # [13:36] <Hixie> Philip`: well then there is a bug, because assuming you only have one document, one of the encodings is wrong.
  506. # [13:36] <hsivonen> anyway, a handful of pointers and ints :-)
  507. # [13:36] * Hixie wonders if rubys wanted status updates only on the "Open issues with open and pending review action items" or on everything on the agenda
  508. # [13:36] <Hixie> the latter being way the hell too many things to comment on
  509. # [13:37] * Joins: webben (n=webben@nat/yahoo/x-b3b7963feac12586)
  510. # [13:39] <Philip`> hsivonen: Fair enough :-)
  511. # [13:41] <hsivonen> considering how stable the parsing algorithm is, it changes surprisingly often.
  512. # [13:42] <zcorpan> Hixie: why no legacy encoding declaration for xhtml?
  513. # [13:42] <Hixie> hm?
  514. # [13:42] <Hixie> hsivonen: yeah :-(
  515. # [13:43] <Hixie> hsivonen: it'll be interesting to see how much the spec changes once we're in CR
  516. # [13:43] <Philip`> hsivonen: (If you're fixing non-linear behaviour, will you add some limit on <i><b><i><b><i><b></i>x</i>x</i>x too?)
  517. # [13:43] <hsivonen> Philip`: not today, no
  518. # [13:43] <zcorpan> Hixie: you only allowed <meta charset="utf-8"/> in xhtml, not <meta http-equiv="content-type" content="text/html; charset=utf-8"/>
  519. # [13:43] <hsivonen> Philip`: I'm only fixing insertBefore in SAX Tree and XOM not to search the children of the foster parent repeatedly
  520. # [13:44] <Philip`> hsivonen: Okay
  521. # [13:44] <Philip`> zcorpan: Encouraging people to say their application/xhtml+xml files are text/html seems like a confusing idea
  522. # [13:45] <Philip`> and given how many restrictions there already are when writing polyglot documents, having to use <meta charset> doesn't seem like a big deal
  523. # [13:47] <Hixie> zcorpan: yes? we only need one way to do this, as far as i can tell.
  524. # [13:47] <zcorpan> Hixie: ok, fair enough. it wasn't clear to me if it was intentional or not
  525. # [13:47] <Hixie> yeah
  526. # [13:48] <zcorpan> Hixie: is it intended that the requirements in #character-encoding-declaration apply to xml too?
  527. # [13:48] <Hixie> no
  528. # [13:49] <Hixie> though i suppose it should
  529. # [13:49] * Hixie goes to change that
  530. # [13:49] <Hixie> again
  531. # [13:53] * Quits: virtuelv_ (n=virtuelv@213.236.208.247) (Read error: 110 (Connection timed out))
  532. # [13:58] <Hixie> http://getluky.net/2009/01/08/a-warning-about-the-real-cost-of-microformats/ seems to apply equally well to rdfa
  533. # [14:02] <Philip`> When I sort of tried using hCard once, the main difficulty was that its required structure didn't match the markup structure I wanted to use, so I ended up adding stuff like <span style=display:none>,</span> to get the right output for both hCard and visual display
  534. # [14:04] <Philip`> and if I was doing something more complex, like having details for multiple people mixed together in the same page, it seems it'd probably be impossible to get right
  535. # [14:05] <Philip`> RDFa seems much better in that regard, because you're just specifying a load of triples and you can put them wherever you want, rather than having to nest your markup precisely how someone thought your data should be nested
  536. # [14:05] <Lachy> Philip`, that's the same problem I encountered when I used hCard once
  537. # [14:06] <Lachy> hmm, I recall trying to use hcard on my own site's contact page once, but it appears I must have given up since it's not even there now
  538. # [14:06] <jgraham> Are microformats considered a success?
  539. # [14:07] <Philip`> (and RDFa adds @content for the cases where the structured data doesn't quite appear in the markup at all)
  540. # [14:07] <Lachy> jgraham, they're not a total failure, since they've been put to good use in some places that I've heard about. It's difficult to say whether they're a success though
  541. # [14:08] <Philip`> so if I have a page which says "<h1>Mr & Mrs Smith</h1> ...their postal address...", there isn't anything in the page which says "Mr Smith" so I can't associate the address with him if I'm using microformats, as far as I'm aware
  542. # [14:08] <Lachy> (one good use I heard about was in a university intranet, where contact pages were marked up with hcard and a service was provided to extract the information to a vcard
  543. # [14:09] <jgraham> Lachy: Any good public examples? I tried using Operator once which is the kind of magic GUI generation thing that all the RDFa people are so excited about and it was just as bad as you might imagine an autogenerated GUI would be
  544. # [14:09] <Philip`> (I guess RDFa might have the problem that it can't associate the same address details with two separate people?)
  545. # [14:10] <hsivonen> jgraham: they don't seem like a total success in the sense that mainstream browsing setups doesn't support saving hCard and hCalendar entries into Address Book and iCal
  546. # [14:10] <Philip`> (but at least it works somewhat better than microformats in that case)
  547. # [14:13] <hsivonen> whee! I broke foster parenting
  548. # [14:14] * zcorpan wonders whether hsivonen is happy to check that <meta charset="utf-8"/> is within 512 bytes and doesn't use entities in xml
  549. # [14:14] <Philip`> Think of the children!
  550. # [14:14] <hsivonen> zcorpan: I'm not
  551. # [14:14] <Philip`> You can't just go around breaking their critical support structures :-(
  552. # [14:14] <Lachy> jgraham, I think this is the one I mentioned above. It seems it is public after all. http://www.griffith.edu.au/cgi-bin/phone_search.pl?format=browse_detail&gotFrame=y&rec_num=1854
  553. # [14:15] * Quits: nessy (n=nessy@203-158-43-109.dyn.iinet.net.au) ("This computer has gone to sleep")
  554. # [14:15] <Hixie> ok nn
  555. # [14:16] <zcorpan> nn
  556. # [14:17] <jgraham> Lachy: That is only really interesting if you have a client that doesn't need tht VCard though (since otherwise you may as well ignore the microformat and generate the vCard on the server)
  557. # [14:17] <jgraham> nn
  558. # [14:17] <jgraham> ^That was simed at Hixie
  559. # [14:19] <hsivonen> something weird happens with the new foster parenting rule with <a><table><a>
  560. # [14:20] <hsivonen> although the new rule isn't supposed to change anything in non-scripted scenarios
  561. # [14:22] * Joins: enki_ (n=enki@c-67-188-42-45.hsd1.ca.comcast.net)
  562. # [14:22] <Lachy> jgraham, yeah, except I think the service itself uses the hcard markup to generate the vcard. But I could be wrong, and perhaps it just generates it directly from the database.
  563. # [14:26] <hsivonen> aargh. the foster parenting breaks compat
  564. # [14:27] * Quits: enki (n=enki@c-67-188-42-45.hsd1.ca.comcast.net) (Read error: 110 (Connection timed out))
  565. # [14:27] <hsivonen> I haven't yet figured out why
  566. # [14:31] <hsivonen> ok. figured out why
  567. # [14:32] <hsivonen> the second <a> removes the previous <a> from stack before foster parenting runs
  568. # [14:32] * Joins: rubys (n=rubys@cpe-075-182-092-038.nc.res.rr.com)
  569. # [14:35] <zcorpan> Lachy: there's a requirement in the "encoding declaration state" text that prevents having both <meta charset> and <meta content...>
  570. # [14:36] <zcorpan> although i agree it could be clearer
  571. # [14:36] <zcorpan> e.g. by being a bullet-point in the list of requirements for encoding declarations
  572. # [14:42] <Lachy> zcorpan, feel free to comment on the bug about that
  573. # [14:44] * Quits: mpt (n=mpt@canonical/launchpad/mpt) (Remote closed the connection)
  574. # [14:46] * Joins: annevk2 (n=opera@p028adc.tokynt01.ap.so-net.ne.jp)
  575. # [14:47] * Joins: MikeSmith (n=MikeSmit@p028adc.tokynt01.ap.so-net.ne.jp)
  576. # [14:47] * Joins: billyjackass (n=MikeSmit@p028adc.tokynt01.ap.so-net.ne.jp)
  577. # [14:47] * Quits: billyjackass (n=MikeSmit@p028adc.tokynt01.ap.so-net.ne.jp) (Read error: 104 (Connection reset by peer))
  578. # [14:54] * Quits: enki_ (n=enki@c-67-188-42-45.hsd1.ca.comcast.net) (Read error: 110 (Connection timed out))
  579. # [14:57] * Joins: aroben (n=aroben@unaffiliated/aroben)
  580. # [15:01] <annevk2> Hixie, I'd argue <meta charset="utf8"/> and friends should be ok as well...
  581. # [15:01] <annevk2> or is there a requirement that you use the canonical version of the character encoding somewhere?
  582. # [15:06] <Lachy> annevk2, utf8 isn't a valid alias of UTF-8
  583. # [15:06] <Lachy> it's not listed here http://www.iana.org/assignments/character-sets
  584. # [15:07] <annevk2> it matches per Unicode though
  585. # [15:07] <annevk2> but fair enough I suppose
  586. # [15:08] <rubys> I'm very much OK with Ian crafting a very narrow exception that permits just this one use case
  587. # [15:09] * Joins: zdobersek (n=zan@cpe-92-37-67-89.dynamic.amis.net)
  588. # [15:11] <Philip`> "If you are Sam Ruby then you MAY include the byte sequence <meta charset="UTF-8" /> in XML documents"
  589. # [15:11] <Lachy> hey, was the requirement that <meta charset> had to be the first element in the <head> dropped?
  590. # [15:11] * Joins: mpt (n=mpt@canonical/launchpad/mpt)
  591. # [15:11] <Lachy> I'm updating the FAQ, and that's still mentioned in there.
  592. # [15:11] <rubys> um, that's a wee bit too narrow. :-)
  593. # [15:11] <hsivonen> Lachy: yes, the first child req sucked
  594. # [15:12] <Lachy> ok, I'll change it in the FAQ to make it a recommendation for authors only.
  595. # [15:13] <annevk2> rubys, but I'm not :p
  596. # [15:14] <rubys> annevk2: go for it! But seriously, the goal isn't to get meta charset support for XHTML5 documents; if that were the goal, I'd argue for ISO-8859-7 and any other ASCII-based but non ISO-8859-1 based charset.
  597. # [15:15] <annevk2> I should probably check the exact requirements on <meta charset> in general
  598. # [15:15] <annevk2> I'm fine with polyglot being restricted to UTF-8
  599. # [15:17] <rubys> For maximal interop, all XML should be either UTF-8 or UTF-16.
  600. # [15:18] <annevk2> ah, it seems it is consistent with <meta charset> requirements
  601. # [15:18] <hsivonen> did Hixie impose the 512 byte req on XHTML?
  602. # [15:18] <annevk2> though it would be slightly better imo if it just said that the encoding was restricted to utf-8
  603. # [15:18] <annevk2> hsivonen, that's the intent
  604. # [15:18] <hsivonen> not cool
  605. # [15:19] <annevk2> because in that case the other requirements would follow naturally
  606. # [15:19] <hsivonen> after all, XHTML5 validation doesn't otherwise involve checking that the doc is a polyglot doc
  607. # [15:20] <rubys> Do you flag, as an error, a meta charset after 512 in text/html?
  608. # [15:21] <hsivonen> rubys: I think I do. If I don't, it's a regression.
  609. # [15:22] <rubys> my question was error vs warning.
  610. # [15:22] <hsivonen> rubys: however, I'm not at all interested in implementing the same check into an XML parser I didn't write
  611. # [15:22] <rubys> the feed validator has a select few checks it does on the byte stream before passing the whole mess off to the xml parser
  612. # [15:23] <hsivonen> rubys: still works, and it's an error
  613. # [15:25] <Lachy> hsivonen, just give a warning to users when its encountered in XHTML saying the 512 byte restriction was not checked
  614. # [15:26] <hsivonen> Lachy: giving a warning means I can't implement the check as a RELAX NG datatype
  615. # [15:26] <hsivonen> (even after hacking Jing to expose the encoding to datatypes)
  616. # [15:26] <Lachy> I don't think the 512 byte restriction needs to apply to XHTML, since it's just a talisman in that case
  617. # [15:26] <annevk2> (was just going to ask about that)
  618. # [15:27] <hsivonen> eww. we now have a "frameset-ok" flag? :-(
  619. # [15:28] <Philip`> If you want to write a valid polyglot document, you'll already have to validate it as both HTML and XHTML to detect all the errors, and so it'll warn you about the 512-byte thing when you validate as HTML, and it doesn't need an extra warning in XHTML
  620. # [15:28] <rubys> lachy: a warning was my thoughts too
  621. # [15:28] <rubys> but I don't sense that hsivonen's reluctance has anything to do with warning vs error, more with having to do the check at all
  622. # [15:29] <Lachy> FAQ updated http://wiki.whatwg.org/wiki/FAQ#How_do_I_specify_the_character_encoding.3F
  623. # [15:30] <Philip`> I think the reluctance is because it's ugly to implement a correct check, and Lachy was suggesting a vague non-correct warning instead (which is easier to implement because it doesn't violate layering)
  624. # [15:30] <Philip`> s/non-correct/non-precise/
  625. # [15:30] <Philip`> s/vague//
  626. # [15:30] <Philip`> s/.*//
  627. # [15:30] <rubys> lol
  628. # [15:34] <hsivonen> Hixie: dude! you added a whitespaceness-sensitive action to the most common insertion mode!
  629. # [15:35] <Philip`> hsivonen: You don't have to check the whitespaceness except in the rare cases when frameset-ok == ok, though
  630. # [15:36] <hsivonen> Philip`: still, not nice
  631. # [15:38] <Philip`> Split the mode into in-body-and-frameset-ok and in-body-and-frameset-not-ok modes, and switch between them the appropriate time, and then the latter mode won't have to do anything fancy at all :-)
  632. # [15:38] * Philip` guesses it's not really that easy in practice
  633. # [15:41] * Philip` discovers that "mae east" and "mae west" give very different sets of results on Google
  634. # [15:41] * Joins: virtuelv (n=virtuelv@193.214.121.161)
  635. # [15:42] <hsivonen> what's the logic in what start tags set frameset-ok to "not ok"?
  636. # [15:43] <hsivonen> Philip`: treating frameset-ok as an insertion mode looks like a sensible idea on the face of it
  637. # [15:43] <hsivonen> Hixie: is there a reason why frameset-ok is not an insertion mode?
  638. # [15:45] <jgraham> Today we will be playing a game called confuse the f**k out of james about how, when, or if, people advocating @summary think it should be used
  639. # [15:45] <rubys> oooh, can I plan?
  640. # [15:45] <jgraham> This is not a hard game because james is easilly confused
  641. # [15:45] <rubys> s/plan/play/... more funnier that way
  642. # [15:46] <jgraham> rubys: Sure, let me just reset my confusedness bit
  643. # [15:47] * annevk2 sets it to one
  644. # [15:47] <Philip`> jgraham: It's not necessary to know how or when @summary should be used; it suffices to prove the existence of cases where it should be used, and thus it must be allowed, regardless of what anyone's actually going to do with it
  645. # [15:47] * Quits: shepazu (n=schepers@cpe-76-168-214-103.socal.res.rr.com)
  646. # [15:48] * Quits: MikeSmith (n=MikeSmit@p028adc.tokynt01.ap.so-net.ne.jp) ("Tomorrow to fresh woods, and pastures new.")
  647. # [15:48] * Parts: annevk2 (n=opera@p028adc.tokynt01.ap.so-net.ne.jp)
  648. # [15:50] <Lachy> hsivonen, I think there's a bug in your parser. http://validator.nu/?doc=http%3A%2F%2Ftinyurl.com%2Fbqpwl9&showsource=yes
  649. # [15:50] <Lachy> have you not added support for <datalist> yet?
  650. # [15:53] <hsivonen> Lachy: for now, I don't understand why that happens
  651. # [15:54] <Lachy> would you like me to file a bug about it?
  652. # [15:54] <hsivonen> Lachy: that would help, yes.
  653. # [15:55] <hsivonen> ??? how could the stack have an <option> element on it in any case?
  654. # [15:55] * hsivonen reads changes to </body> 'in body'
  655. # [15:56] <Dashiva> If </datalist> doesn't close </option>...
  656. # [15:59] <hsivonen> Lachy: I think Dashiva has the answer
  657. # [16:00] <Lachy> but <datalist> should close the <option>, shouldn't it? Just like <select> does
  658. # [16:02] <Lachy> but there's also the other error about list="" not pointing to a datalist or select element, which doens't make sesne.
  659. # [16:05] <Philip`> Hmm, the BBC iPlayer appears to provide subtitles using http://www.w3.org/TR/ttaf1-dfxp/
  660. # [16:05] <hsivonen> I think it's weird that <var> doesn't behave like the other italicizing elements when it comes to AAA
  661. # [16:06] <Lachy> hsivonen, I got an error from your bugzilla: "undef error - Insecure dependency in exec while running with -T switch at /usr/share/perl5/Mail/Mailer/sendmail.pm line 22. "
  662. # [16:06] <hsivonen> Philip`: do they have a dedicated app that consumes DFXP?
  663. # [16:06] <Lachy> that showed up after I tried to submit a new bug report
  664. # [16:06] <hsivonen> Lachy: yeah, that's annoying. Sorry about that. I have no idea how to fix.
  665. # [16:07] <Lachy> should I just try again, or just email you the bug report instead?
  666. # [16:07] <hsivonen> Lachy: no, the bug was filed ok. Thanks
  667. # [16:07] <Philip`> http://philip.html5.org/misc/subtitles-1.xml
  668. # [16:07] <Lachy> ok
  669. # [16:07] * Quits: xydyx (n=hdh@58.187.21.120) (Read error: 104 (Connection reset by peer))
  670. # [16:07] <Philip`> (Was that transcribed live? It looks pretty rubbish quality...)
  671. # [16:08] <Philip`> (which is strange since it's a prerecorded show)
  672. # [16:09] <Philip`> http://philip.html5.org/misc/subtitles-2.xml looks much better quality
  673. # [16:11] <Philip`> hsivonen: The streaming Flash player doesn't seem to support subtitles, as far as I've seen; I don't know about the offline downloader application, or about the versions on other devices
  674. # [16:12] <Philip`> Oh
  675. # [16:13] <Philip`> Apparently the Flash version does, and I just haven't clicked on that button
  676. # [16:13] <Philip`> http://www.bbc.co.uk/blogs/bbcinternet/2008/12/iplayer_subtitles_increase_our.html
  677. # [16:13] <hsivonen> Philip`: do you know if BBC ever uses features that go beyond rendering plain string at given times?
  678. # [16:14] <Philip`> hsivonen: The subtitles-2.xml uses colours to distinguish speakers
  679. # [16:14] * danbri doesn't know
  680. # [16:15] <danbri> at joost we had quite a fancy api for subtitles / annotations, ...
  681. # [16:15] <danbri> ...but not much actual data
  682. # [16:16] * Joins: xydyx (n=hdh@58.187.21.120)
  683. # [16:17] * Quits: virtuelv (n=virtuelv@193.214.121.161) ("Leaving")
  684. # [16:18] <Philip`> (subtitles-2 uses multiple colours in a single line to distinguish speakers, too)
  685. # [16:19] <jgraham> Do they outline in a different colour to make sure the words stand out against the background?
  686. # [16:20] <hsivonen> interesting. I'm surprised that someone actually uses the more advanced features at all
  687. # [16:20] * Joins: dglazkov (n=dglazkov@c-98-207-88-44.hsd1.ca.comcast.net)
  688. # [16:20] <Philip`> jgraham: http://farm4.static.flickr.com/3128/3131207136_59b2f4dccf_o.jpg looks outlined
  689. # [16:22] <Philip`> Oh, the colours don't work at all in the Flash player
  690. # [16:22] <Philip`> It's all just white with black outline
  691. # [16:22] <hsivonen> http://stackoverflow.com/questions/134727/whats-the-key-difference-between-html4-and-html5/135585#135585
  692. # [16:22] <hsivonen> "HTML 5 invites you give add a lot of semantic value to your code."
  693. # [16:23] * Philip` looks at half a dozen other subtitle files, but none appear to use anything more complex than colour and <br/>
  694. # [16:24] * Philip` supposes the offline player might not ignore colours
  695. # [16:24] * Quits: zalan (n=kvirc@catv-89-133-232-199.catv.broadband.hu) ("KVIrc 3.4.0 Virgo http://www.kvirc.net/")
  696. # [16:25] <jgraham> hsivonen: It is well known that HTML n+1 has that property for all n because semantics are good and so newer releases must add more of them. Or something.
  697. # [16:27] <Philip`> jgraham: Your explanation is incomplete, as it does not include the fact that XHTML 1.0 has more semantic value than HTML 4
  698. # [16:28] <jgraham> Philip`: Adding an X to the front puts you on a triple semantics score
  699. # [16:28] * hsivonen mumbles about releases with .2 minor number
  700. # [16:29] <hsivonen> I'm puzzled by the the modeness of 'in foreign' and the flagness of frameset-ok
  701. # [16:30] <hsivonen> seems backwards to me :-)
  702. # [16:36] * Joins: weinig (n=weinig@adsl-99-140-60-47.dsl.scrm01.sbcglobal.net)
  703. # [16:39] * Quits: ap (n=ap@194.154.88.46)
  704. # [16:39] * Quits: weinig (n=weinig@adsl-99-140-60-47.dsl.scrm01.sbcglobal.net) (Client Quit)
  705. # [16:42] * Joins: zdobersek1 (n=zan@cpe-92-37-77-244.dynamic.amis.net)
  706. # [16:44] * Joins: eric_carlson (n=ericc@nat/apple/x-15e5bd6e6ec4b907)
  707. # [16:53] * Joins: weinig (n=weinig@99.140.60.47)
  708. # [16:57] * Joins: billmason (n=bmason@ip102.unival.com)
  709. # [16:57] * Parts: billmason (n=bmason@ip102.unival.com)
  710. # [16:58] * Quits: dglazkov (n=dglazkov@c-98-207-88-44.hsd1.ca.comcast.net)
  711. # [16:58] * Quits: zdobersek (n=zan@cpe-92-37-67-89.dynamic.amis.net) (Read error: 110 (Connection timed out))
  712. # [17:00] * Quits: maikmerten (n=merten@ls5dhcp196.cs.uni-dortmund.de) (Remote closed the connection)
  713. # [17:01] <hsivonen> so the new thread-friendly AAA sometimes creates a node that isn't needed. seems harmless enough if the test cases are any indication
  714. # [17:01] <gsnedders> jgraham: There yet?
  715. # [17:02] <jgraham> gsnedders: Yes
  716. # [17:02] <gsnedders> jgraham: See my changes to Anolis?
  717. # [17:02] <jgraham> gsnedders: No
  718. # [17:02] <gsnedders> jgraham: Go, look.
  719. # [17:03] * Quits: weinig (n=weinig@99.140.60.47)
  720. # [17:04] <jgraham> gsnedders: Just pulled them
  721. # [17:04] * Quits: Maurice (n=ano@a80-101-46-164.adsl.xs4all.nl) ("Disconnected...")
  722. # [17:06] <jgraham> gsnedders: From the logs it seems good. I'll need to change the frontend to match the new option names
  723. # [17:06] <jgraham> gsnedders: I don't understand your question about html5lib 0.12
  724. # [17:06] <jgraham> the ns stuff
  725. # [17:07] <gsnedders> jgraham: Does the SVG/MathML branch put HTML elements into the HTML namespace?
  726. # [17:07] <jgraham> Optionally yes
  727. # [17:07] <gsnedders> k
  728. # [17:07] <jgraham> I am prepared to be persuaded about the default value of that option
  729. # [17:07] <gsnedders> I'd say yes, per spec.
  730. # [17:07] <jgraham> (my preference is "no")
  731. # [17:08] <jgraham> gsnedders: So, per Hixie's latest tinking there is no spec for libraries except insofar as they are used in a UA. Therefore I can do what I like as long as there is at least an option to make the behaviour conforming :)_
  732. # [17:08] <gsnedders> jgraham: I'm not entirely sure I agree with that reading of the spec
  733. # [17:09] <gsnedders> I know it's what Hixie meant by the edit, but I'm not entirely convinced from the text.
  734. # [17:09] <hsivonen> I'm not convinced that letting browsers and libs diverge is a good thing
  735. # [17:09] <jgraham> gsnedders: Which edit (I was talking about an IRC conversation)
  736. # [17:09] <jgraham> ?
  737. # [17:10] <gsnedders> The spec should, for parsing, define what the output tree should be like. What is doing the parsing should be irrelevant.
  738. # [17:10] <gsnedders> jgraham: I can't remember :P
  739. # [17:12] <Philip`> gsnedders: Libraries don't necessarily output trees; it only makes sense to require a tree in (scripted) browsers, where there has to be a DOM tree that results from parsing
  740. # [17:13] <gsnedders> Philip`: What else could the library do? It needs to be a tree — it doesn't work streaming without fatal errors.
  741. # [17:13] <Philip`> If you don't have a whole browser, HTML5 doesn't define any way to determine the output of parsing, so there's no restriction on libraries
  742. # [17:13] <jgraham> gsnedders: You could stream a mutation event
  743. # [17:14] <Philip`> You could construct an internal tree and then stream it out as a series of events
  744. # [17:14] * Joins: dglazkov (n=dglazkov@nat/google/x-60a79afb46a4f9d2)
  745. # [17:14] <Philip`> You could output a stream of RDF triples
  746. # [17:15] * gsnedders collapses off his chair laughing at the suggestion
  747. # [17:15] <Philip`> You could strip all the tags and just output plain text
  748. # [17:15] <Philip`> There's lots of things a parser library could (and should be allowed to) do
  749. # [17:19] <hsivonen> the triple thing could work. you could use http://n.whatwg.org/2009/02/23/previous-sibling as a predicate that gives order to children that have the same http://n.whatwg.org/2009/02/23/parent
  750. # [17:20] <Philip`> Presumably you'd have to generate a GUID-based URI for every element in the document
  751. # [17:22] * Quits: mpt (n=mpt@canonical/launchpad/mpt) (Read error: 113 (No route to host))
  752. # [17:22] * Joins: mpt (n=mpt@canonical/launchpad/mpt)
  753. # [17:26] * Joins: svl (n=me@69.Red-88-2-205.staticIP.rima-tde.net)
  754. # [17:30] * Joins: dave_levin (n=dave_lev@c-98-203-247-78.hsd1.wa.comcast.net)
  755. # [17:32] * gsnedders stretches
  756. # [17:32] <gsnedders> Time to review Web sockets, me thinks
  757. # [17:38] * Joins: shepazu (n=schepers@m1f0436d0.tmodns.net)
  758. # [17:41] * Joins: zdobersek (n=zan@cpe-92-37-69-188.dynamic.amis.net)
  759. # [17:44] * Quits: pesla\work (n=retep@procurios.xs4all.nl) ("( www.nnscript.com :: NoNameScript 4.21 :: www.esnation.com )")
  760. # [17:56] * Quits: zdobersek1 (n=zan@cpe-92-37-77-244.dynamic.amis.net) (Read error: 110 (Connection timed out))
  761. # [17:59] * Joins: sayrer (n=chatzill@user-160ve4s.cable.mindspring.com)
  762. # [18:03] * Quits: sayrer (n=chatzill@user-160ve4s.cable.mindspring.com) (Read error: 104 (Connection reset by peer))
  763. # [18:10] * Joins: erlehmann (n=erlehman@86.59.25.121)
  764. # [18:11] * Quits: shepazu (n=schepers@m1f0436d0.tmodns.net) (Read error: 60 (Operation timed out))
  765. # [18:11] * Joins: grimboy (n=grimboy@78-86-152-156.zone2.bethere.co.uk)
  766. # [18:13] * Joins: weinig (n=weinig@adsl-99-140-60-47.dsl.scrm01.sbcglobal.net)
  767. # [18:13] * Joins: shepazu (n=schepers@m1f0436d0.tmodns.net)
  768. # [18:16] * Joins: Maurice (i=copyman@5ED548D4.cable.ziggo.nl)
  769. # [18:21] * Quits: jwalden (n=waldo@c-76-102-160-171.hsd1.ca.comcast.net) ("inbound")
  770. # [18:24] * Quits: Lachy (n=Lachlan@pat-tdc.opera.com) ("This computer has gone to sleep")
  771. # [18:38] * Quits: shepazu (n=schepers@m1f0436d0.tmodns.net) (Read error: 110 (Connection timed out))
  772. # [18:38] * Joins: shepazu (n=schepers@m180436d0.tmodns.net)
  773. # [18:41] * Joins: mlpug (n=mlpug@a88-115-168-225.elisa-laajakaista.fi)
  774. # [18:48] * Joins: Lachy (n=Lachlan@85.196.122.246)
  775. # [18:48] * Quits: svl (n=me@69.Red-88-2-205.staticIP.rima-tde.net) (Read error: 60 (Operation timed out))
  776. # [18:54] * Parts: erlehmann (n=erlehman@86.59.25.121)
  777. # [19:00] * Joins: svl (n=me@69.Red-88-2-205.staticIP.rima-tde.net)
  778. # [19:08] * Quits: shepazu (n=schepers@m180436d0.tmodns.net) (Read error: 60 (Operation timed out))
  779. # [19:09] * Joins: shepazu (n=schepers@m180436d0.tmodns.net)
  780. # [19:15] * Joins: jwalden (n=waldo@corp-246.mountainview.mozilla.com)
  781. # [19:26] * Joins: pauld_ (n=pauld@92.40.211.133.sub.mbb.three.co.uk)
  782. # [19:28] * Quits: svl (n=me@69.Red-88-2-205.staticIP.rima-tde.net) ("And back he spurred like a madman, shrieking a curse to the sky.")
  783. # [19:31] <karlcow> I wonder if this explains the social behavior of many geeks in the silicon valley - http://www.boston.com/bostonglobe/ideas/articles/2008/03/30/a_singles_map_of_the_united_states_of_america/
  784. # [19:38] * Joins: jrharshath (i=d2d43dfb@gateway/web/ajax/mibbit.com/x-10f5d8cc35afa9fa)
  785. # [19:39] <jrharshath> I was going through a few threads of the whatwg's mailing list archives...
  786. # [19:39] <jrharshath> can anyone tell me what the TCPConnection() thing is?
  787. # [19:40] <Lachy> it was a DOM api for creating TCP connections. I think it was largely replaced by WebSockets
  788. # [19:41] <jrharshath> TCP connections to ... ?
  789. # [19:41] <jrharshath> to the server where the web app originated from, i'd guess?
  790. # [19:45] <Philip`> jrharshath: http://www.whatwg.org/specs/web-apps/2007-10-26/multipage/section-network.html#tcpconnection
  791. # [19:47] * Quits: dave_levin (n=dave_lev@c-98-203-247-78.hsd1.wa.comcast.net)
  792. # [19:49] <gsnedders> jrharshath: Anywhere, IIRC
  793. # [19:50] * gsnedders is wrong
  794. # [19:50] <gsnedders> (again)
  795. # [19:52] * Joins: roc (n=roc@121-72-173-196.dsl.telstraclear.net)
  796. # [19:56] <jrharshath> gsnedders: pardon? i am new to IRC, so i don't follow the jargon..
  797. # [20:00] * Quits: Amorphous (i=jan@unaffiliated/amorphous) (Read error: 110 (Connection timed out))
  798. # [20:02] * Joins: Amorphous (i=jan@unaffiliated/amorphous)
  799. # [20:03] * Joins: scherkus1 (n=scherkus@72.14.227.1)
  800. # [20:03] * Quits: pauld_ (n=pauld@92.40.211.133.sub.mbb.three.co.uk) (Read error: 110 (Connection timed out))
  801. # [20:06] * Quits: scherkus1 (n=scherkus@72.14.227.1) (Client Quit)
  802. # [20:12] * Joins: dimich (n=dimich@72.14.227.1)
  803. # [20:13] * Joins: dave_levin (n=dave_lev@72.14.227.1)
  804. # [20:20] * Joins: tantek (n=tantek@c-67-161-5-143.hsd1.ca.comcast.net)
  805. # [20:34] * Quits: jrharshath (i=d2d43dfb@gateway/web/ajax/mibbit.com/x-10f5d8cc35afa9fa) ("http://www.mibbit.com ajax IRC Client")
  806. # [20:34] * Quits: roc (n=roc@121-72-173-196.dsl.telstraclear.net)
  807. # [20:34] * Joins: jrharshath (i=d2d43dfb@gateway/web/ajax/mibbit.com/x-4c6ef3fceb92f2e2)
  808. # [20:41] * Joins: pauld (n=pauld@host86-134-106-105.range86-134.btcentralplus.com)
  809. # [20:43] * Quits: dolske (n=dolske@firefox/developer/dolske)
  810. # [20:56] * Quits: shepazu (n=schepers@m180436d0.tmodns.net) (Read error: 110 (Connection timed out))
  811. # [20:58] * Joins: kingryan (n=kingryan@adsl-99-27-42-97.dsl.pltn13.sbcglobal.net)
  812. # [21:02] * Joins: KevinMarks (n=KevinMar@72.158.245.34)
  813. # [21:03] <jrharshath> On http://www.whatwg.org/specs/web-apps/2007-10-26/multipage/section-network.html#tcpconnection, I read: "This interface does not allow for raw access to the underlying network. For example, this interface could not be used to implement an IRC client without proxying messages through a custom server."
  814. # [21:03] <jrharshath> I also read that "[the network connections interface will ] enable Web applications to communicate with each other in local area networks..."
  815. # [21:04] <jrharshath> what do the two statements imply?
  816. # [21:04] <Philip`> jrharshath: You can connect to arbitrary servers, but only using a special protocol which is incompatible with standard protocols
  817. # [21:05] <jrharshath> so its not like writing a C program and opening a TCP Socket?
  818. # [21:05] <Philip`> No
  819. # [21:05] <Philip`> because that would be very insecure
  820. # [21:06] <Philip`> e.g. it would let you speak HTTP to any server, and circumvent all the same-origin restrictions in current browsers
  821. # [21:06] <jrharshath> hmm.. I agree.
  822. # [21:06] <jrharshath> can you give me more reasons please?
  823. # [21:07] * Joins: dolske (n=dolske@corp-246.mountainview.mozilla.com)
  824. # [21:08] <Philip`> Isn't that enough reason? :-)
  825. # [21:11] * Quits: pauld (n=pauld@host86-134-106-105.range86-134.btcentralplus.com)
  826. # [21:11] <jrharshath> and, what would the special protocol be?
  827. # [21:12] <Philip`> It would be whatever is described in the spec, I presume
  828. # [21:12] * Quits: jwalden (n=waldo@corp-246.mountainview.mozilla.com) ("->K")
  829. # [21:13] <Philip`> in section 6.3.7.1
  830. # [21:13] <jrharshath> okay, got to it.
  831. # [21:13] <jrharshath> ty :)
  832. # [21:14] * Joins: pauld (n=pauld@host86-134-106-105.range86-134.btcentralplus.com)
  833. # [21:14] * Quits: pauld (n=pauld@host86-134-106-105.range86-134.btcentralplus.com) (Client Quit)
  834. # [21:14] * Joins: ap (n=ap@194.154.88.46)
  835. # [21:17] * Joins: roc (n=roc@202.0.36.64)
  836. # [21:23] * Joins: jwalden (n=waldo@corp-246.mountainview.mozilla.com)
  837. # [21:29] * Joins: smedero (n=smedero@mdp-nat251.mdp.com)
  838. # [21:31] * Quits: rubys (n=rubys@cpe-075-182-092-038.nc.res.rr.com) (Remote closed the connection)
  839. # [21:32] * Quits: grimboy (n=grimboy@78-86-152-156.zone2.bethere.co.uk) (Remote closed the connection)
  840. # [21:33] * Joins: grimboy (n=grimboy@78-86-152-156.zone2.bethere.co.uk)
  841. # [21:40] * Joins: enki (n=enki@c-67-188-42-45.hsd1.ca.comcast.net)
  842. # [21:40] * Quits: mlpug (n=mlpug@a88-115-168-225.elisa-laajakaista.fi) (Remote closed the connection)
  843. # [21:43] * Parts: jrharshath (i=d2d43dfb@gateway/web/ajax/mibbit.com/x-4c6ef3fceb92f2e2)
  844. # [21:52] * Joins: yecril71 (n=giecrilj@piekna-gts.2a.pl)
  845. # [21:53] * Joins: epeus (n=KevinMar@72.14.228.1)
  846. # [22:03] * Quits: tantek (n=tantek@c-67-161-5-143.hsd1.ca.comcast.net)
  847. # [22:04] * Quits: ROBOd (n=robod@89.122.216.38) ("http://www.robodesign.ro")
  848. # [22:13] * Joins: tantek (n=tantek@c-67-161-5-143.hsd1.ca.comcast.net)
  849. # [22:17] * Joins: pauld (n=pauld@host86-134-106-105.range86-134.btcentralplus.com)
  850. # [22:25] * Joins: mpt_ (n=mpt@canonical/launchpad/mpt)
  851. # [22:28] * Quits: tantek (n=tantek@c-67-161-5-143.hsd1.ca.comcast.net)
  852. # [22:32] * Joins: tantek (n=tantek@c-67-161-5-143.hsd1.ca.comcast.net)
  853. # [22:38] * Joins: hdh (n=hdh@58.187.17.182)
  854. # [22:39] * Joins: doublec (n=chris@202.0.36.64)
  855. # [22:52] * Quits: tantek (n=tantek@c-67-161-5-143.hsd1.ca.comcast.net)
  856. # [22:56] * Quits: dolske (n=dolske@firefox/developer/dolske)
  857. # [22:58] * Quits: pauld (n=pauld@host86-134-106-105.range86-134.btcentralplus.com)
  858. # [22:58] * Quits: jwalden (n=waldo@corp-246.mountainview.mozilla.com) ("->S")
  859. # [23:02] * Joins: tantek (n=tantek@c-67-161-5-143.hsd1.ca.comcast.net)
  860. # [23:02] * Joins: dolske (n=dolske@corp-246.mountainview.mozilla.com)
  861. # [23:04] * Quits: xydyx (n=hdh@58.187.21.120) (Read error: 113 (No route to host))
  862. # [23:04] * Joins: dbaron (n=dbaron@corp-246.mountainview.mozilla.com)
  863. # [23:08] * Joins: virtuelv (n=virtuelv@mic92-9-82-236-193-123.fbx.proxad.net)
  864. # [23:11] * Joins: jwalden (n=waldo@corp-246.mountainview.mozilla.com)
  865. # [23:14] * Quits: enki (n=enki@c-67-188-42-45.hsd1.ca.comcast.net) (Read error: 60 (Operation timed out))
  866. # [23:22] * Quits: Maurice (i=copyman@5ED548D4.cable.ziggo.nl) ("Disconnected...")
  867. # [23:25] <Hixie> hsivonen: i could have the spec say that if there's a <meta charset=""> in an XML doc, the doc must also validate as a text/html doc :-)
  868. # [23:25] <Hixie> hsivonen: i can remove the 512 requirement if you don't think it's useful
  869. # [23:26] * Joins: olliej (n=oliver@nat/apple/x-f675d39448e4b12f)
  870. # [23:28] * Quits: ap (n=ap@194.154.88.46)
  871. # [23:32] * Joins: tantek_ (n=tantek@c-67-161-5-143.hsd1.ca.comcast.net)
  872. # [23:38] * Quits: yecril71 (n=giecrilj@piekna-gts.2a.pl)
  873. # [23:39] * Quits: eric_carlson (n=ericc@nat/apple/x-15e5bd6e6ec4b907)
  874. # [23:42] * Quits: heycam (n=cam@210-84-48-144.dyn.iinet.net.au) ("bye")
  875. # [23:44] * Joins: dolske_ (n=dolske@corp-246.mountainview.mozilla.com)
  876. # [23:49] <roc> Hixie: thanks for all those video edits. they look good so far
  877. # [23:51] * Quits: tantek (n=tantek@c-67-161-5-143.hsd1.ca.comcast.net) (Read error: 110 (Connection timed out))
  878. # [23:56] * Joins: ojan (n=ojan@nat/google/x-8e7ea02dd539dad6)
  879. # Session Close: Tue Feb 24 00:00:00 2009

The end :)