/irc-logs / freenode / #whatwg / 2009-01-03 / end

Options:

  1. # Session Start: Sat Jan 03 00:00:00 2009
  2. # Session Ident: #whatwg
  3. # [00:10] * Joins: nessy (n=nessy@124-168-169-109.dyn.iinet.net.au)
  4. # [00:26] <heycam> Hixie, which goal would allowing document.write() from <svg:script> violate?
  5. # [00:31] * Quits: dbaron (n=dbaron@c-69-140-1-234.hsd1.va.comcast.net) ("8403864 bytes have been tenured, next gc will be global.")
  6. # [00:45] * Quits: Maurice (i=copyman@5ED548D4.cable.ziggo.nl) ("Disconnected...")
  7. # [01:04] <Hixie> heycam: the goal of keeping svg fragments in text/html separable, so that you can drop them into standalone xml documents and still have them work
  8. # [01:04] * Quits: tantek (n=tantek@adsl-63-195-114-133.dsl.snfc21.pacbell.net)
  9. # [01:06] * Joins: tantek (n=tantek@adsl-63-195-114-133.dsl.snfc21.pacbell.net)
  10. # [01:10] <heycam> Hixie, what other HTML APIs wouldn't work when you're working in a standalone xml document?
  11. # [01:17] <heycam> i mean, you can easily come up with some script in an SVG-embedded-in-HTML that would fail to work when you extract it out
  12. # [01:17] * heycam afk
  13. # [01:18] <roc> that's sort of a generic problem with that goal
  14. # [01:19] <roc> that goal isn't even realized in XHTML
  15. # [01:19] <roc> because of namespaces, stylesheets and other things that prevent cutting and pasting from an XML document without data loss
  16. # [01:20] * Philip` discovers that his ISP is seemingly blocking an entirely harmless page (a copy of a robots.txt) file from web.archive.org, via a system that claims to be filtering "indecent images of children" (and was blocking a Wikipedia page some weeks ago)
  17. # [01:20] <Philip`> although actually they're not blocking the page, they're rewriting links in the HTML in the page pointing to that page
  18. # [01:21] <Philip`> Oh, actually they seem to be blocking pretty much all of web.archive.org
  19. # [01:36] * Joins: dimich (n=dimich@c-98-203-230-54.hsd1.wa.comcast.net)
  20. # [01:41] <Hixie> heycam: there aren't many APIs that would flat out not work (i.e. where there is no way to use the API in a way that would work)
  21. # [01:50] * Joins: weinig (n=weinig@cpe-66-65-132-93.nyc.res.rr.com)
  22. # [01:54] * Joins: MikeSmith (n=MikeSmit@EM114-48-136-19.pool.e-mobile.ne.jp)
  23. # [02:20] * Quits: danbri (n=danbri@unaffiliated/danbri) (Read error: 104 (Connection reset by peer))
  24. # [02:20] * Joins: danbri (n=danbri@77.78.195.22)
  25. # [02:20] * Quits: dimich (n=dimich@c-98-203-230-54.hsd1.wa.comcast.net)
  26. # [02:29] * Quits: roc (n=roc@222-152-171-232.jetstream.xtra.co.nz)
  27. # [02:35] * Joins: shepazu (n=schepers@c022112.adsl.customers.cinergycom.net)
  28. # [02:41] * Quits: Morphous (i=jan@unaffiliated/amorphous) (Read error: 104 (Connection reset by peer))
  29. # [02:42] * Quits: shepazu (n=schepers@c022112.adsl.customers.cinergycom.net)
  30. # [02:44] * Quits: danbri (n=danbri@unaffiliated/danbri) (Read error: 60 (Operation timed out))
  31. # [02:45] * Quits: weinig (n=weinig@cpe-66-65-132-93.nyc.res.rr.com)
  32. # [02:47] * Quits: svl (n=me@ip565744a7.direct-adsl.nl) ("And back he spurred like a madman, shrieking a curse to the sky.")
  33. # [02:49] * Joins: danbri (n=danbri@77.78.195.22)
  34. # [03:00] <takkaria> hmm
  35. # [03:01] <takkaria> maybe I should write an SVG fragment extractor for hubbub
  36. # [03:01] * Joins: Morphous (i=jan@unaffiliated/amorphous)
  37. # [03:04] <Philip`> What is an SVG fragment extractor?
  38. # [03:04] <Philip`> Isn't it just getElementsByTagName('svg') plus an XML serialiser? :-)
  39. # [03:08] * Quits: VasilievVV (i=vvv@wikimedia/VasilievVV) (kornbluth.freenode.net irc.freenode.net)
  40. # [03:08] * Quits: Hixie (i=ianh@trivini.no) (kornbluth.freenode.net irc.freenode.net)
  41. # [03:08] * Quits: syp (n=syp@lasigpc9.epfl.ch) (kornbluth.freenode.net irc.freenode.net)
  42. # [03:10] * Joins: Hixie (i=ianh@trivini.no)
  43. # [03:10] * Joins: syp (n=syp@lasigpc9.epfl.ch)
  44. # [03:16] * Quits: nessy (n=nessy@124-168-169-109.dyn.iinet.net.au) ("This computer has gone to sleep")
  45. # [03:27] * Joins: nessy (n=nessy@124-168-169-109.dyn.iinet.net.au)
  46. # [03:32] <Philip`> When it comes to getting significant long-term performance gains from web browsers running on multi-core CPUs, the parser seems like it's on the wrong side of Amdahl's law
  47. # [03:34] <Philip`> It's still got to loop through every byte in the document in order, and splitting one component into a separate thread can't do more than double the performance regardless of how many processors you have
  48. # [03:34] * Quits: nessy (n=nessy@124-168-169-109.dyn.iinet.net.au) ("This computer has gone to sleep")
  49. # [03:34] * Philip` would like to learn why he's wrong, though :-)
  50. # [03:53] <MikeSmith> the Refresh header has never been defined in any RFC or other standard (other than the HTML5 draft), right?
  51. # [03:57] <MikeSmith> http://www.informit.com/articles/article.aspx?p=1193856
  52. # [03:58] <MikeSmith> Knuth interview where he's generally critical/skeptical of value of parallelism and multithreading, but says "From the opposite point of view, I do grant that web browsing probably will get better with multicores. I’ve been talking about my technical work, however, not recreation."
  53. # [04:07] * Quits: danbri (n=danbri@unaffiliated/danbri) (Read error: 110 (Connection timed out))
  54. # [04:09] * Parts: erlehmann (n=erlehman@86.59.25.121)
  55. # [04:09] * Joins: erlehmann (n=erlehman@86.59.25.121)
  56. # [04:14] * Joins: weinig (n=weinig@cpe-66-65-132-93.nyc.res.rr.com)
  57. # [04:15] * Parts: dave_levin_ (n=dave_lev@72.14.224.1)
  58. # [04:19] * Joins: olliej_ (n=oliver@nat/apple/x-c0dd0b83ad02c767)
  59. # [04:19] * Quits: olliej (n=oliver@nat/apple/x-3a68f4fcc5ea69a4) (Read error: 104 (Connection reset by peer))
  60. # [04:28] <Hixie> MikeSmith: not to my knowledge
  61. # [04:28] * Quits: tantek (n=tantek@adsl-63-195-114-133.dsl.snfc21.pacbell.net)
  62. # [04:29] * Joins: roc (n=roc@121-72-177-3.dsl.telstraclear.net)
  63. # [04:33] <MikeSmith> Hixie: thanks
  64. # [04:49] * Quits: weinig (n=weinig@cpe-66-65-132-93.nyc.res.rr.com)
  65. # [04:51] * Joins: olliej (n=oliver@nat/apple/x-288f6ae72080602f)
  66. # [04:51] * Quits: olliej_ (n=oliver@nat/apple/x-c0dd0b83ad02c767) (Read error: 104 (Connection reset by peer))
  67. # [05:00] * Joins: weinig (n=weinig@cpe-66-65-132-93.nyc.res.rr.com)
  68. # [05:14] * Parts: erlehmann (n=erlehman@86.59.25.121)
  69. # [05:24] * Quits: weinig (n=weinig@cpe-66-65-132-93.nyc.res.rr.com)
  70. # [05:25] * Joins: othermaciej (n=mjs@c-69-181-43-20.hsd1.ca.comcast.net)
  71. # [05:27] * Quits: othermaciej (n=mjs@c-69-181-43-20.hsd1.ca.comcast.net) (Read error: 104 (Connection reset by peer))
  72. # [05:27] * Joins: othermaciej (n=mjs@c-69-181-43-20.hsd1.ca.comcast.net)
  73. # [05:33] * Joins: othermaciej_ (n=mjs@c-69-181-43-20.hsd1.ca.comcast.net)
  74. # [05:33] * Quits: othermaciej (n=mjs@c-69-181-43-20.hsd1.ca.comcast.net) (Read error: 131 (Connection reset by peer))
  75. # [05:39] * Quits: webben (n=webben@91.85.205.99) ("Lost terminal")
  76. # [06:29] * othermaciej_ is now known as othermaciej
  77. # [07:32] * Joins: hdh (n=hdh@58.187.19.148)
  78. # [07:49] * Joins: billyjackass (n=MikeSmit@EM114-48-173-76.pool.e-mobile.ne.jp)
  79. # [08:03] * Quits: MikeSmith (n=MikeSmit@EM114-48-136-19.pool.e-mobile.ne.jp) (Read error: 110 (Connection timed out))
  80. # [09:07] <hsivonen> is it just me or isn't incompatible ABIs on one operating system between different compilers a result of C++ being a Technical Specification?
  81. # [09:08] <hsivonen> and if you want to implement a comptetitive underdog C++ compiler, you need to reverse engineer the ABI of the incumbent compiler
  82. # [09:24] <Hixie> i don't understand why anyone would think a "Technical Specification" is a good idea, given the scale of evidence against the idea
  83. # [09:25] <Hixie> (that is, a spec that doesn't define precise conformance criteria for all conformance classes and doesn't leave things undefined)
  84. # [09:26] <Hixie> (at least in terms of processing content appropriate to the specification, of course; e.g. a network-level spec shouldn't talk about UI, etc)
  85. # [09:44] <roc> hsivonen: that is a good point
  86. # [10:40] * Joins: ROBOd (n=robod@89.122.216.38)
  87. # [10:53] * Joins: danbri (n=danbri@77.78.195.22)
  88. # [11:15] * Joins: Maurice (i=copyman@5ED548D4.cable.ziggo.nl)
  89. # [11:30] * Quits: hallvors (n=hallvord@softbank221089079197.bbtec.net) (Read error: 104 (Connection reset by peer))
  90. # [11:42] * Joins: nessy (n=nessy@124-168-169-109.dyn.iinet.net.au)
  91. # [11:43] * Quits: olliej (n=oliver@nat/apple/x-288f6ae72080602f)
  92. # [11:44] <Hixie> i wonder why some IE installs send a UA-CPU HTTP header with requests
  93. # [11:46] * Quits: roc (n=roc@121-72-177-3.dsl.telstraclear.net)
  94. # [11:49] <hsivonen> Hixie: to allow ActiveX install sites offer the right binary for 32 vs 64 bit?
  95. # [11:58] <Hixie> oh, maybe
  96. # [11:58] <Hixie> what a lot of bandwidth to waste just for that
  97. # [11:59] <Hixie> especially since it's already in the User-Agent header
  98. # [12:12] * Joins: kinetik_ (n=kinetik@121.98.132.55)
  99. # [12:14] * Joins: VeXocide_ (i=vexocide@snail.stack.nl)
  100. # [12:14] * Quits: jgraham (n=jgraham@web22.webfaction.com) (kornbluth.freenode.net irc.freenode.net)
  101. # [12:14] * Quits: VeXocide (i=vexocide@snail.stack.nl) (kornbluth.freenode.net irc.freenode.net)
  102. # [12:14] * Quits: kinetik (n=kinetik@121.98.132.55) (kornbluth.freenode.net irc.freenode.net)
  103. # [12:14] * Quits: takkaria (n=takkaria@isparp.co.uk) (kornbluth.freenode.net irc.freenode.net)
  104. # [12:15] * kinetik_ is now known as kinetik
  105. # [12:16] * Joins: takkaria (n=takkaria@isparp.co.uk)
  106. # [12:16] * Joins: jgraham (n=jgraham@web22.webfaction.com)
  107. # [12:31] * Quits: nessy (n=nessy@124-168-169-109.dyn.iinet.net.au) ("This computer has gone to sleep")
  108. # [12:43] <Philip`> http://blogs.msdn.com/iemobile/archive/2006/08/03/Detecting_IE_Mobile.aspx mentions UA-CPU, and I assume it's not for ActiveX if it's on a mobile platform
  109. # [12:46] <Philip`> http://www.ppczone.net/forum/archive/index.php/t-268.html - fun
  110. # [12:47] * Joins: olliej_ (n=oliver@c-67-164-125-23.hsd1.ca.comcast.net)
  111. # [13:00] * Joins: webben (n=webben@91.85.205.112)
  112. # [13:25] * Joins: maikmerten (n=maikmert@Lae60.l.pppool.de)
  113. # [13:27] <hsivonen> Philip`: a use case for Unicode in HTTP headers!
  114. # [13:46] * VeXocide_ is now known as VeXocide
  115. # [13:47] <jgraham> gsnedders: I assume by "doctypes" the original twitterer meant "doctypes I have used in my web publishing" which suggests he probably was including XHTML 1.x (but probably served as text/html)
  116. # [13:49] <jgraham> Hmm
  117. # [13:49] <jgraham> Didn't meant to do that
  118. # [13:49] <jgraham> Sorry, dody connection + scrollback = bad
  119. # [13:49] <jgraham> *dodgy
  120. # [13:50] <jgraham> I meant to say http://james.html5.org/tables/table_inspector.html now with (probably very buggy) support for the new HTML 5 algorithm
  121. # [13:51] <jgraham> (I may not be very responsive to bug reports over the weekend)
  122. # [14:10] * Quits: billyjackass (n=MikeSmit@EM114-48-173-76.pool.e-mobile.ne.jp) (Read error: 110 (Connection timed out))
  123. # [14:15] * Joins: MikeSmith (n=MikeSmit@EM114-48-17-231.pool.e-mobile.ne.jp)
  124. # [14:20] * Joins: BenMillard (i=cerbera@cpc1-flee1-0-0-cust285.glfd.cable.ntl.com)
  125. # [14:21] <BenMillard> jgraham, header text is being shown in cells when you click them, which makes the layout jump around while checking relationships.
  126. # [14:22] <BenMillard> jgraham, was much better when clicking only did highlighting, imho
  127. # [14:23] <BenMillard> jgraham, this table works fine in Smart Headers mode but gives an error page in HTML5 December: http://james.html5.org/tables/table_inspector.py?input_type=type_uri&uri=http%3A%2F%2Fprojectcerbera.com%2Fgta%2F2%2Ftutorials%2Fmultiplayer-strategy&source=&algorithm=html5c
  128. # [14:29] <BenMillard> Philip`, I also see: <meta name="MobileOptimized" content="width">
  129. # [14:29] <BenMillard> Philip`, it's justified as: "It's really useful if you try to build pages or sites that are targeted towards mobile end users."
  130. # [14:30] <BenMillard> Philip`, so if I've used semantic markup, logical source order, CSS layout and send a media="handheld" stylesheet with a linearised and compact layout which hides all but the most important things, I haven't "targeted towards mobile end users"? :|
  131. # [14:33] <webben> BenMillard: Where's this?
  132. # [14:33] <BenMillard> webben, http://blogs.msdn.com/iemobile/archive/2006/08/03/Detecting_IE_Mobile.aspx
  133. # [14:33] <BenMillard> (at the end)
  134. # [14:34] <webben> that article is lulzsome.
  135. # [14:35] <BenMillard> maybe device independence doesn't work? :)
  136. # [14:35] <Philip`> BenMillard: If you've expended all that effort, but not actually tested in mobile devices and modified your page so that it works as well as possible in them (using whatever hacks are necessary if there's no better solution), you haven't done a good job of targeting real end users
  137. # [14:35] <BenMillard> Philip`, what I describe above works perfectly in all but the stupidest of mobiles, in my experience.
  138. # [14:36] <BenMillard> like, the ones which apply media="screen" and try to do side-by-side page layout on narrow screens with no zoom :P
  139. # [14:36] <BenMillard> Openwave on my several-year-old, low-end Sharp handset handles it fine
  140. # [14:37] <BenMillard> Pocket IE tended to mess it up
  141. # [14:37] <Philip`> BenMillard: If end users use the stupidest of mobiles, that doesn't excuse you from using the necessary hacks to target them :-)
  142. # [14:37] <webben> mind you I guess it's older than the recent popularity of sophisticated media queries
  143. # [14:38] <BenMillard> webben, "popularity" seems rather an overstatement...
  144. # [14:38] <webben> meh s/popularity/appearance/
  145. # [14:38] <BenMillard> Philip`, they can probably install Opera Mini if they want to browse the actual web
  146. # [14:39] <Philip`> (But if most people use decent browsers nowadays, it seems to makes sense to target them in the cleanest possible way)
  147. # [14:39] <webben> BenMillard: Yep. If most of the web is unusable with their browser, are they likely to constitute much of an audience.
  148. # [14:41] <BenMillard> Philip`, I've only seen a couple of devices try doing full-page layout on tiny screens without zoom. I've seen several which ignore positioning CSS but apply colours, some font styling and things like text-align. (Including 2 devices older than 5 years.)
  149. # [14:41] <Philip`> BenMillard: I would imagine most people have never heard of Opera Mini, so they would never realise they could install it
  150. # [14:42] <BenMillard> Philip`, yeah, so as webben says they probably don't do mobile browsing since their experience is "it doesn't work"
  151. # [14:42] <Philip`> BenMillard: (20 million Opera Mini users vs >3.5 billion moDbile phone users is not a very good ratio)
  152. # [14:42] <Philip`> Um
  153. # [14:43] <Philip`> Thank you SSH for inserting random garbage into my message
  154. # [14:43] <webben> Philip`: Yeah, but Opera Mini's hardly the only bearable mobile browser.
  155. # [14:43] <webben> Philip`: There's also the issue of depth of use.
  156. # [14:44] <BenMillard> Philip`, my impression is that stupid mobile browsers are a minority. A device I tested in 2004 ignored all but a few text styling properties, which is very conservative but sensible: http://projectcerbera.com/misc/mobile-support
  157. # [14:44] <webben> I know I use the web more for having a bearable browser on my phone than for having one that makes sites unusable.
  158. # [14:44] <BenMillard> (And that device was several years old, even in 2004)
  159. # [14:44] <Philip`> I suppose I must count as two Opera Mini users, but I've visited about three pages with it in the past couple of years
  160. # [14:44] <BenMillard> (it had a monochrome display)
  161. # [14:44] <webben> Philip`: I use it all the time.
  162. # [14:44] <BenMillard> oh wait, it could do white, black and blue
  163. # [14:45] <Philip`> I would be more likely to use a web browser on my phone if it didn't seem to cost huge amounts of money :-)
  164. # [14:45] <webben> for wikipedia/news/googling/twitter-following (no free SMS in UK) ... but I haz a fixed contract for "unlimited data"
  165. # [14:46] <BenMillard> Philip`, there was a big fanfare around flat-rate Internet access throughout the UK mobile networks last yea
  166. # [14:46] <BenMillard> r
  167. # [14:48] <BenMillard> so my point is: building websites following the now long-established best practices for standards-based web design is enough for authors to reasonably expect mobile devices to work on their website
  168. # [14:48] <BenMillard> without adding <meta> tags, UA sniffing and suchlike
  169. # [14:49] <webben> I guess my point is: if you're going to spend $ optimizing, might as well spend $ on people who spend a lot of time on the phone browser - and those folks will tend to have decent browsers and favorable modern data contracts.
  170. # [14:49] <BenMillard> (by best practiceses I specifically don't mean MWBP, btw)
  171. # [14:49] <Philip`> All I know is that the person who pays the phone bill commented that there was ~1 UKP of WAP usage, and I'd only used it for about five minutes
  172. # [14:49] <Philip`> Also: I need to sort out Unicode support in my IRC client - the layout totally breaks if I type a £ :-(
  173. # [15:06] * Joins: philippe_ (n=chatzill@93.1.56.122)
  174. # [15:08] * hsivonen notes that @font-face mostly sucks in browsers that use the native TrueType rasterization of Windows
  175. # [15:08] <hsivonen> on XP
  176. # [15:09] <hsivonen> due to lack of anti-aliasing and due to lack of Windows-oriented hinting in free-as-in-freedom fonts
  177. # [15:10] * Quits: danbri (n=danbri@unaffiliated/danbri)
  178. # [15:12] * hsivonen wonders what proportion of XP users has enabled ClearType
  179. # [15:13] <BenMillard> hsivonen, in my experience it's only a few expert users (like me)
  180. # [15:13] <BenMillard> hsivonen, the school I work for, my local library and the college I attended didn't have it turned on
  181. # [15:14] <webben> didn't IE7 enable cleartype by default?
  182. # [15:14] <hsivonen> BenMillard: seems like bad news for @font-face in Opera and Firefox on XP
  183. # [15:14] <BenMillard> webben, yes but only inside IE7, not the whole OS
  184. # [15:14] * Quits: olliej_ (n=oliver@c-67-164-125-23.hsd1.ca.comcast.net)
  185. # [15:15] * Joins: danbri (n=danbri@77.78.195.22)
  186. # [15:15] <BenMillard> I thought XPsp3 would switch it on as a new default but it seems not
  187. # [15:16] <webben> oh i see
  188. # [15:16] <BenMillard> webben, so if you install IE7 on a non-ClearType XP you get smooth fonts in IE7 but not in any other browsers you use (unless they do smoothing in their own viewport by their own means)
  189. # [15:16] <webben> yep
  190. # [15:17] <webben> like Safari on Win does, right?
  191. # [15:17] <hsivonen> I can't even figure out how to turn ClearType on without resorting to googling
  192. # [15:18] <BenMillard> hsivonen, right-click desktop, Properties item, Appearence tab, Effects button, "Use the following method..." checkbox, then set "ClearType" in the dropdown, then OK out of all the windows
  193. # [15:18] <BenMillard> simple :P
  194. # [15:18] <hsivonen> ah. I was in the right dialog but I was looking for a checkbox where I should have looked for a two-item popup
  195. # [15:19] <hsivonen> I opened the dialog about 3 times without seeing it, because I was looking for [ ] Use ClearType
  196. # [15:19] <BenMillard> hsivonen, there are...interesting UI decisions like that throughout Windows
  197. # [15:20] <BenMillard> hisivonen, although Standard font smoothing is slightly different to none at all
  198. # [15:21] <hsivonen> @font-face examples are still ugly in Opera. do I need to reboot or something?
  199. # [15:21] <BenMillard> hsivonen, seems to switch immediately for me
  200. # [15:21] <hsivonen> oh well. I guess the solution is to upgrade to Linux or OS X
  201. # [15:21] <BenMillard> oh, actually
  202. # [15:21] <BenMillard> unticking it seems to work without reboot
  203. # [15:21] <BenMillard> ticking and selecting Standard works
  204. # [15:22] <BenMillard> ah, ticking and selecting ClearType worked the 2nd time
  205. # [15:22] <BenMillard> (I used the Apply button in Display Properties after each change, and used Opera's view of this channel to check if the change took effect)
  206. # [15:24] <BenMillard> hsivonen, which examples are you using?
  207. # [15:24] <hsivonen> the ones under howcome's princexml.com directory
  208. # [15:25] <hsivonen> http://www.princexml.com/howcome/2008/webfonts/
  209. # [15:26] <BenMillard> hsivonen, ugliness confirmed here (Opera 9.63)
  210. # [15:28] <BenMillard> heh, Ray Larabie made the Pricedown font used heavily in GTA3...he also designed cars and suchlike for the early GTA's
  211. # [15:28] * Quits: Philip` (n=philip@xvm-11-39.ghst.net) ("leaving")
  212. # [15:29] * Joins: philip_ (n=philip@xvm-11-39.ghst.net)
  213. # [15:29] * philip_ is now known as Philip`
  214. # [15:30] <hsivonen> now it worked
  215. # [15:31] <hsivonen> still ugly compared to OS X, though
  216. # [15:35] <hsivonen> This whole cleartype thing seems like snakeoil compared to FreeType, Quartz and whatever Adobe's renderer is called
  217. # [15:42] <hsivonen> hmm. is ClearType enabled regardless of OS setting in Firefox 3.1?
  218. # [15:43] <hsivonen> perhaps Opera 10 should enable ClearType regardless of OS setting at least for @font-face fonts
  219. # [15:45] <Philip`> Vista's new fonts were designed specifically to work with ClearType, which apparently makes it better
  220. # [15:45] <hsivonen> hmm. no, it's not on all the time in Firefox. I just forget the microsoftish "Apply" button
  221. # [15:46] <BenMillard> hsivonen, there are differences in font rendering between browser engines, even on the same OS with the same settings: http://www.hicksdesign.co.uk/journal/cairo-beats-safari
  222. # [15:47] <BenMillard> (I dare say desktop applications differ in similar ways, too)
  223. # [15:47] <BenMillard> s/desktop/desktop publishing/
  224. # [15:48] <hsivonen> fortunately, Microsoft says it has patents on ClearType, which hopefully keeps it away from Linux and OS X
  225. # [15:50] <takkaria> yeah, libfreetype can use cleartype but works around said patents by working out its own hintings, I believe
  226. # [15:50] <hsivonen> takkaria: I think that's about Apple's TT hinting patents
  227. # [15:51] <takkaria> oh, fair enough
  228. # [15:51] <Philip`> http://freetype.sourceforge.net/patents.html ?
  229. # [15:51] <takkaria> one day I should probably start knowing things before talking about them :)
  230. # [15:57] * Parts: BenMillard (i=cerbera@cpc1-flee1-0-0-cust285.glfd.cable.ntl.com)
  231. # [16:08] * Joins: weinig (n=weinig@cpe-66-65-132-93.nyc.res.rr.com)
  232. # [16:08] * Quits: weinig (n=weinig@cpe-66-65-132-93.nyc.res.rr.com) (Client Quit)
  233. # [16:10] * Quits: philippe_ (n=chatzill@93.1.56.122) (Read error: 104 (Connection reset by peer))
  234. # [16:15] <hsivonen> so if natural language processing tools aren't available for $LANGUAGE, people whose native language is $LANGUAGE are better off using RDFa than using English???
  235. # [16:16] * Joins: annevk (n=annevk@cm-84.215.140.104.getinternet.no)
  236. # [16:27] <webben> hsivonen: Wouldn't the argument more be: it's easier for authors to use RDFa, then to write NLP tools for those languages?
  237. # [16:28] <webben> (I don't think it's a good argument, but anyhow...)
  238. # [16:29] <webben> to be fair, it's something that comes up in microformats discussions
  239. # [16:29] <webben> people suggested trying to ditch the machine-data datatimes in favour of human-friendly formats
  240. # [16:29] <webben> but the idea always got shot down because of the complexity of parsing all the different locales
  241. # [16:30] * Joins: kangax (n=kangax@pool-72-70-206-155.sctnpa.east.verizon.net)
  242. # [16:30] <hsivonen> webben: isn't it easier to use English than RDFa?
  243. # [16:31] <webben> hsivonen: I wouldn't say it's easier for me to learn (say) Chinese than use RDFa, so not necessarily.
  244. # [16:31] <hsivonen> perhaps it's offensive to suggest using English, but one might consider having to use RDF even more offensive
  245. # [16:32] <hsivonen> webben: Chinese and English are different
  246. # [16:32] <Dashiva> Dates aren't really a fair example, since machine-readable dates are also not-entire-stupid-human-readable :)
  247. # [16:32] <webben> hsivonen: And also, that's asking end-users to learn a different language just to read your content.
  248. # [16:33] <webben> hsivonen: Chinese and English are different because Chinese is objectively harder to learn or because English is more internationally spoken?
  249. # [16:33] <hsivonen> webben: both
  250. # [16:33] <webben> I think it would be easier for me to use RDFa than write my content in Spanish and ask my users to learn Spanish.
  251. # [16:34] <webben> (well, and learn it myself)
  252. # [16:34] <webben> heck, I'd probably find it easier to use RDFa than publish content in French - and I learned that at schoo.
  253. # [16:35] <hsivonen> webben: at least English, Spanish and French are human-authorable
  254. # [16:35] <webben> How is RDFa not human-authorable (as opposed to not easy to author)?
  255. # [16:36] <webben> I think the rules for writing correctly and idiomatically in a foreign language are much harder to learn than any technical language I've seen.
  256. # [16:38] <hsivonen> webben: everything is human authorable on some level
  257. # [16:39] <webben> human-author-friendly you meant?
  258. # [16:39] <hsivonen> for example, English is more authorable than Finnish, because there are better text input methods for English
  259. # [16:40] <webben> you can type RDFa without worrying about accented characters, so I'm not sure on that reckoning it's less author-friendly
  260. # [16:41] <webben> but I think the interface difficulties are not comparable with the cognitive load
  261. # [16:41] <hsivonen> webben: you can dictate English. dictation solutions for RDFa aren't available AFAIK
  262. # [16:41] <hsivonen> or predictive text entry with limited keys
  263. # [16:42] <webben> hsivonen: well, you can presumably dictate RDFa the same as RSI sufferers dictate other code?
  264. # [16:43] <hsivonen> webben: the alternative wasn't using other kind of code but putting a few natural-language words into Google/Yahoo!/Live search
  265. # [16:44] <webben> hsivonen: Hmm. I don't think the idea is for authors to construct RDF queries (well, not directly) and submit them to search engines.
  266. # [16:45] <hsivonen> webben: then the range of UI limits things even though RDF is touted as being able to express concepts not prescribed by an authority
  267. # [16:45] <webben> hsivonen: I'm not really sure what you mean.
  268. # [16:46] <webben> for example, a user of $LANGUAGE could enter a keyword in $LANGUAGE into Yahoo! search, and data could be extracted for display in the results page with a SearchMonkey
  269. # [16:46] <webben> without Yahoo! having to implement NLP tools for $LANGUAGE.
  270. # [16:46] <webben> (that much, at least, is possible now)
  271. # [16:47] <hsivonen> it seems to me that it's more efficient to implement superficial NLP for $LANGUAGE at a few companies than to do RDF at each publication point
  272. # [16:48] * Joins: svl (n=me@ip565744a7.direct-adsl.nl)
  273. # [16:48] <hsivonen> about UI concepts I mean this: Microformats solve n use cases with a small value of n.
  274. # [16:48] <webben> hsivonen: That's why I think the argument is weak, though to be fair it's hard to know without knowing $ cost of sufficient NLP.
  275. # [16:49] <hsivonen> even if RDF syntax could express solutions for more use cases, the practical number of cases addressed will be limited by UI in both cases--not by syntax
  276. # [16:49] <webben> hsivonen: I'm confused why it couldn't be limited by both
  277. # [16:50] <webben> i.e. let's say there's a set of use cases which are not limited by UI and a set which are.
  278. # [16:51] <webben> it's not necessary that those sets can be mapped to the sets limited by microformat limitations and those not.
  279. # [16:51] <webben> that is, there could be a set limited by microformat limitations but not UI limitation.
  280. # [16:53] <webben> just as there could be a set limited by UI limitations not microformats limitations
  281. # [17:12] * Quits: karlcow (n=karl@modemcable211.1-80-70.mc.videotron.ca) ("O public road, I say back I am not afraid to leave you, yet I love you, you express me better than I can express myself.")
  282. # [17:20] * Joins: erlehmann (n=erlehman@86.59.25.121)
  283. # [17:33] <takkaria> I'm still not entirely clear on what RDFa does that you can't do with plain HTML
  284. # [17:33] <takkaria> except for "be RDF"
  285. # [17:34] <Dashiva> Curies? :)
  286. # [17:34] * Joins: aaronlev (n=chatzill@g228081247.adsl.alicedsl.de)
  287. # [17:35] <takkaria> mm, curry
  288. # [17:35] <jcranmer> all I know about RDF is what I found inside mozilla
  289. # [17:35] <jcranmer> where I think it ought to be slaughtered and left for dead
  290. # [17:37] <annevk> hmm, RDF-EASE is something I thought of years ago but abandoned it because I thought the markup would simply be <div> everywhere
  291. # [17:38] <takkaria> I dunno, I guess I just don't see the use in marking up data as being about a given URL
  292. # [17:38] <takkaria> data about an URL should be in the resource itself, really
  293. # [17:39] <Dashiva> Is there a solution to what to do about malicious or just plain incorrect RDF statements?
  294. # [17:40] <takkaria> and what's this thing foaf and hcards do, assigning names and dates of births to urls? tis the wrong way round
  295. # [17:41] <takkaria> actually, hcard doesn't do it
  296. # [17:41] <takkaria> foaf does though
  297. # [17:43] <Philip`> takkaria: The URI is the identifier of a representation of a human being, so assigning names and dates of birth to a URI is just as good as assigning them to any other identifier
  298. # [17:46] <takkaria> the ontology is still upside-down, though; people have URIs, and people have names and dates of birth
  299. # [17:47] <Dashiva> It's indirection :)
  300. # [17:48] <takkaria> I guess I just don't like saying that takkaria.org represents me, I'd rather say I owned takkaria.org
  301. # [17:49] <danbri> if i understand you, it's XFN that most directly equates people with their pages
  302. # [17:49] <danbri> the XFN idiom is to write in http://danbri.org/ that i'm xfn:friend with http://takkaria.example.org/
  303. # [17:49] <danbri> of course it is understood that the pages are serving as proxy identifiers, but the indirection isn't made as explicit as RDF (eg. FOAF) tends to encourage
  304. # [17:50] <danbri> in FOAF we say there's a Person, and they might have a 'weblog', 'homepage', or 'openid' property, whose value is a page URI
  305. # [17:50] <Philip`> takkaria: It's not upside-down, it's a circle - people have names and names identify people, just like how people have URIs and URIs identify people, and you describe the properties of a person in relation to an identifier of that person (e.g. "John Smith was born on 1 Jan 1980" or "http://johnsmith.example.com was born on 1980-01-01" or whatever)
  306. # [17:50] <danbri> and they might have jabberID, name, mbox etc too, of course
  307. # [17:50] <danbri> re Dashiva: Is there a solution to what to do about malicious or just plain incorrect RDF statements?
  308. # [17:51] <danbri> ...nope
  309. # [17:51] <Philip`> (and then you might also say "http://johnsmith.example.com owns the web page http://johnsmith.example.com")
  310. # [17:51] <takkaria> danbri: fair enough, I'm just going off wikipedia examples and what they look like
  311. # [17:51] <danbri> Can I politely say 'of course not'; the situation is same as with 'what to do about malicious or plain incorrect English/Spanish/... statements
  312. # [17:52] <danbri> there are techniques of course, eg. favouring statemnts that come from sources you have evidence for trusting (which will depend on the problem you're currently concerned with, the risks of getting it wrong, etc)
  313. # [17:52] <takkaria> Philip`: I see what you're saying, but names only identify people and sites don't only identify people
  314. # [17:52] <takkaria> I just see using site URIs as representing people as a bit of a semantic hack, I guess
  315. # [17:53] <danbri> there are techniques for encouraging rdf statements to tend towards not going stale, eg. define 'date of birth' rather than 'age'; however that is in tension with a reasonable desire to represent the things people write in web pages
  316. # [17:53] <danbri> so i'd write 'i'm 36' more than 'i was born in 1972', even if the latter is never going to go out of date, and the former will be false in < 365 days
  317. # [17:53] <danbri> with db-backed sites, the page containing 'i'm 36' is even more prevalent
  318. # [17:54] <danbri> yeah it's a bit of a hack takkaria, but one that lets us do data joins a lot easier --- less guesswork etc
  319. # [17:54] <Philip`> takkaria: It's not a site URI, it's a page URI (which just happens to have no page name after the domain name) :-)
  320. # [17:54] <takkaria> heh
  321. # [17:55] <takkaria> true, but doesn't really affect the argument ^_^
  322. # [17:56] <takkaria> danbri: I guess-- I suppose I don't see the value of doing it particularly over the coolness factor
  323. # [17:57] * Quits: aaronlev (n=chatzill@g228081247.adsl.alicedsl.de) (Read error: 110 (Connection timed out))
  324. # [17:58] <Dashiva> danbri: Humans have builtin nonsense detectors. RDF parsers don't.
  325. # [17:59] <annevk> what's the use of marking up your age?
  326. # [18:00] <Philip`> Dashiva: The output of RDF parsers is eventually read by humans, who still have builtin nonsense detectors - I guess the issue is that it introduces a potentially large separation between the output (which is all the user can judge for nonsensicality) and the input (which is what the user ought to judge for nonsensicality)
  327. # [18:00] <Dashiva> And probably disassociates the source from the data as well...
  328. # [18:01] <takkaria> annevk: address book applications extracting data and showing it?
  329. # [18:04] <danbri> dashiva, one thing we took care to include in SPARQL (the rdf query language) is a construct for querying the source of the data (identified by uri)
  330. # [18:04] <danbri> so every claim in the rdf store is tagged with an optional URI, indicating where you found it (or who said it --- the admin scheme is up to you)
  331. # [18:05] <danbri> a reasonable compare/contrast is SQL, consider reasking the question 'what to do about malicious or incorrect SQL statements'?
  332. # [18:05] <Dashiva> Isn't aggregation one of the big points about RDF? Wouldn't assigning sources to an aggregated statement get messy?
  333. # [18:05] <danbri> firstly - people don't dump raw SQL data in the Web; it isn't designed for that kind of interop/merging
  334. # [18:06] <danbri> secondly - RDF+SPARQL have some tricks (like the graph keyword alluded to just now) for keeping track of data imported from untrusted or semi-trusted sources
  335. # [18:07] <takkaria> I'm not sure SQL is analagous to RDF, really? not on the web, anyway. SQL might be a solution for storing data, but it's got nothing to do with decentralised aggregation and merging
  336. # [18:08] * Joins: danbri_ (n=danbri@77.78.195.22)
  337. # [18:08] * Quits: danbri (n=danbri@unaffiliated/danbri) (Read error: 54 (Connection reset by peer))
  338. # [18:10] <annevk> takkaria, people still use those?
  339. # [18:12] <takkaria> annevk: not convinced they do, but that's the only use I can think of
  340. # [18:13] <takkaria> really people are just better off using a vcard, though, I'd have thought
  341. # [18:13] <annevk> it's not like people are actually going to tag all that stuff when they write a blog or forum post
  342. # [18:13] <annevk> and don't tell me it will happen automatically, because then the data is just redundant
  343. # [18:14] <Philip`> Lots of sites ask your age, so they can refuse to show you content if you're under a certain age (to comply with COPPA laws or common sense or whatever) or so they can congratulate you on your birthday, and it'd be nicer if they could automatically determine that information when you've simply logged in and told them your homepage
  344. # [18:14] <takkaria> the other thing is of course that most peopel do
  345. # [18:15] <webben> danbri_ and takkaria: in the SQL-on-the-web line of things, http://developer.yahoo.com/yql/ be interesting if you haven't seen it.
  346. # [18:15] <takkaria> don't have homepages
  347. # [18:15] <annevk> I haven't seen studies, but I'd imagine lots of people have a profile page of some sorts
  348. # [18:15] <annevk> Philip`, wouldn't they use some kind of OpenID data provider for that info?
  349. # [18:16] <takkaria> true. I imagine you're far better off implementing support for grabbing info from Facebook profiles than using any kind of markup extraction though
  350. # [18:16] * Philip` doesn't have a homepage or a global profile page
  351. # [18:16] <annevk> philip.html5.org
  352. # [18:17] <Philip`> That's not a homepage, it's just a page with links to some stuff that I made :-)
  353. # [18:17] <Philip`> I have other sites with links to other stuff that I made
  354. # [18:17] <annevk> I would count all of them
  355. # [18:17] <Philip`> and none of them are a main homepage
  356. # [18:17] <annevk> doesn't have to be pretty
  357. # [18:18] <Philip`> philip.html5.org doesn't even give my name, and it didn't even give an email address until fairly recently
  358. # [18:18] <annevk> but now it does
  359. # [18:19] <Philip`> It still doesn't say who I am :-)
  360. # [18:20] <annevk> why is that needed?
  361. # [18:24] * Parts: annevk (n=annevk@cm-84.215.140.104.getinternet.no)
  362. # [18:26] <danbri_> YQL - interesting, I knew it was on the way but missed the announcement. Google have a perhaps-similar subset used in the Google Viz APIs... wonder how they compare...
  363. # [18:27] <danbri_> re philip.html5.org ... if people use it like a homepage, eg. hyperlink to it when they mention you, ... maybe that's some large % of what being a homepage really is...?
  364. # [18:43] <Philip`> People rarely link to it - they just refer to me by name :-)
  365. # [19:01] * Quits: kangax (n=kangax@pool-72-70-206-155.sctnpa.east.verizon.net)
  366. # [19:02] * Joins: weinig (n=weinig@cpe-66-65-132-93.nyc.res.rr.com)
  367. # [19:06] * Quits: MikeSmith (n=MikeSmit@EM114-48-17-231.pool.e-mobile.ne.jp) ("sex break")
  368. # [19:11] * Quits: danbri_ (n=danbri@unaffiliated/danbri)
  369. # [19:12] * Parts: erlehmann (n=erlehman@86.59.25.121)
  370. # [19:19] * Joins: kangax (n=kangax@pool-72-70-206-155.sctnpa.east.verizon.net)
  371. # [19:35] * Quits: weinig (n=weinig@cpe-66-65-132-93.nyc.res.rr.com)
  372. # [20:03] * Quits: kangax (n=kangax@pool-72-70-206-155.sctnpa.east.verizon.net)
  373. # [20:29] * gsnedders has not got into Clare College for comp.sci.
  374. # [20:30] * gsnedders is also less sure than ever if he actually wants to do phys.
  375. # [20:30] <gsnedders> So, basically, I have no idea what I'm doing later this year :P
  376. # [20:30] <gsnedders> (I have been put into the pool, but I really don't want to at a college out from the centre, who are the likely ones to have any space for pool applicants)
  377. # [20:31] <gsnedders> Meh.
  378. # [20:32] <Philip`> Oh :-(
  379. # [20:32] * gsnedders obviously sucks at comp.sci.
  380. # [20:33] <Philip`> (Ones near the centre might still have places, because they sometimes just don't have enough candidates they would want to accept)
  381. # [20:33] <gsnedders> Philip`: They do however tend to have more applicants
  382. # [20:33] <gsnedders> (Ignoring Churchhill)
  383. # [20:34] <gsnedders> More waiting. fun.
  384. # [20:51] * Quits: webben (n=webben@91.85.205.112) (Read error: 60 (Operation timed out))
  385. # [20:59] * Quits: eric_carlson (n=ericc@adsl-67-112-12-110.dsl.anhm01.pacbell.net) (Read error: 145 (Connection timed out))
  386. # [21:01] * Joins: eric_carlson (n=ericc@adsl-67-112-12-110.dsl.anhm01.pacbell.net)
  387. # [21:06] * Joins: erlehmann (n=erlehman@86.59.25.121)
  388. # [21:07] * Quits: maikmerten (n=maikmert@Lae60.l.pppool.de) (Read error: 110 (Connection timed out))
  389. # [21:07] * Joins: maikmerten (n=maikmert@L8f29.l.pppool.de)
  390. # [21:10] * Joins: aaronlev (n=chatzill@g228081247.adsl.alicedsl.de)
  391. # [21:15] * Joins: othermaciej_ (n=mjs@c-69-181-43-20.hsd1.ca.comcast.net)
  392. # [21:15] * Quits: othermaciej (n=mjs@c-69-181-43-20.hsd1.ca.comcast.net) (Read error: 54 (Connection reset by peer))
  393. # [21:49] * Joins: othermaciej (n=mjs@c-69-181-43-20.hsd1.ca.comcast.net)
  394. # [21:49] * Quits: othermaciej_ (n=mjs@c-69-181-43-20.hsd1.ca.comcast.net) (Read error: 104 (Connection reset by peer))
  395. # [22:06] * Joins: danbri (n=danbri@92.36.152.188)
  396. # [22:09] * Quits: hdh (n=hdh@58.187.19.148) (Remote closed the connection)
  397. # [22:21] * Joins: kangax (n=kangax@ool-182f8118.dyn.optonline.net)
  398. # [22:32] * Joins: danbri_ (n=danbri@92.36.152.188)
  399. # [22:35] * Quits: danbri (n=danbri@unaffiliated/danbri) (Read error: 110 (Connection timed out))
  400. # [22:35] * Joins: weinig (n=weinig@cpe-66-65-132-93.nyc.res.rr.com)
  401. # [22:37] * Joins: danbri (n=danbri@92.36.143.24)
  402. # [22:38] * Quits: danbri_ (n=danbri@92.36.152.188) (Read error: 145 (Connection timed out))
  403. # [22:39] * Quits: danbri (n=danbri@unaffiliated/danbri) (Read error: 104 (Connection reset by peer))
  404. # [22:40] * Joins: danbri (n=danbri@92.36.143.24)
  405. # [22:43] * Quits: weinig (n=weinig@cpe-66-65-132-93.nyc.res.rr.com)
  406. # [22:45] * Quits: ROBOd (n=robod@89.122.216.38) ("http://www.robodesign.ro")
  407. # [22:46] * Joins: danbri_ (n=danbri@92.36.143.24)
  408. # [22:48] * Joins: danbri__ (n=danbri@92.36.143.24)
  409. # [23:03] * Quits: danbri (n=danbri@unaffiliated/danbri) (Read error: 110 (Connection timed out))
  410. # [23:05] * Quits: maikmerten (n=maikmert@L8f29.l.pppool.de) (Remote closed the connection)
  411. # [23:07] * Joins: nessy (n=nessy@124-168-169-109.dyn.iinet.net.au)
  412. # [23:08] * Quits: danbri_ (n=danbri@unaffiliated/danbri) (Connection timed out)
  413. # [23:20] * Joins: dbaron (n=dbaron@c-69-140-1-234.hsd1.va.comcast.net)
  414. # [23:24] * Quits: danbri__ (n=danbri@92.36.143.24) (Read error: 54 (Connection reset by peer))
  415. # [23:26] * Quits: othermaciej (n=mjs@c-69-181-43-20.hsd1.ca.comcast.net)
  416. # [23:36] * Joins: webben (n=webben@91.85.205.112)
  417. # [23:43] * Quits: aaronlev (n=chatzill@g228081247.adsl.alicedsl.de) (Read error: 110 (Connection timed out))
  418. # [23:58] * Joins: danbri (n=danbri@92.36.152.188)
  419. # Session Close: Sun Jan 04 00:00:00 2009

The end :)