/irc-logs / w3c / #html-wg / 2007-10-30 / end

Options:

  1. # Session Start: Tue Oct 30 00:00:00 2007
  2. # Session Ident: #html-wg
  3. # [00:02] <anne> It's funny how more "diffs" for a supposedly stable draft (W3C Recommendation) are marketed as if it's a great thing: http://lists.w3.org/Archives/Public/www-forms/2007Oct/0051.html
  4. # [00:04] <Hixie> well html5 is already well over a THOUSAND diffs!
  5. # [00:04] <Hixie> so it must be AWESOME!
  6. # [00:04] <gsnedders> not good enough.
  7. # [00:04] <gsnedders> we haven't reached 1337 yet.
  8. # [00:04] <anne> I think you have to count the amount of diffs since it reached Recommendation status
  9. # [00:04] <gsnedders> (nor 313373)
  10. # [00:05] <anne> The more the merrier at that point, it seems
  11. # [00:05] <Hixie> anne: ah well, that's not happening before 2022. so oh well.
  12. # [00:05] * gsnedders likes Hixie's certainty of his timetable
  13. # [00:05] <gsnedders> I expect we may well write the test cases quicker than you expect
  14. # [00:06] <Hixie> hey we've hit every milestone that's gone so far
  15. # [00:06] <Hixie> on the timetable i put forward last year
  16. # [00:06] <Dashiva> I dunno
  17. # [00:06] <Dashiva> You can never have enough testcases
  18. # [00:06] <Hixie> (http://lists.w3.org/Archives/Public/www-archive/2006Nov/0045.html)
  19. # [00:06] <anne> "every"? :)
  20. # [00:06] <Hixie> (though by "we" i mean "the whatwg", not the htmlwg, sadly)
  21. # [00:06] <Philip> It's much easier to write test cases for fun bits of the spec than for the rest of it
  22. # [00:06] <Hixie> anne: well ok, the one! :_P
  23. # [00:07] <gsnedders> Dashiva: I wrote several thousand during my spec review for a couple of algorithms
  24. # [00:07] <gsnedders> Philip: hey! I'm a loser with no life! What do you want me to do!?
  25. # [00:07] <Dashiva> I wonder how long until I regret checking the testcase box in the survey
  26. # [00:07] <anne> get one?
  27. # [00:07] <kingryan> gsnedders: I assume those are chardet tests? or sniffing tests? have you published them?
  28. # [00:07] <gsnedders> anne: yeah, that was the answer I was waiting for
  29. # [00:08] * Philip is only interested in the features that have pretty colours
  30. # [00:08] <gsnedders> kingryan: no, some of the common microsyntaxes
  31. # [00:08] <anne> gsnedders, I'm glad I live up to your expectations
  32. # [00:08] <gsnedders> kingryan: they were published, but my server had a RAID failure…
  33. # [00:09] <anne> maybe we should prioritize the rendering section then, Philip
  34. # [00:09] <Philip> gsnedders: Were the several thousand each written by hand?
  35. # [00:09] <gsnedders> Philip: yes, but they are very systematic and can be written very quickly
  36. # [00:09] <gsnedders> (i.e., there is one bit of test data with an output for each of a set of the algorithms)
  37. # [00:12] <Philip> gsnedders: Okay - that's much less like cheating than my 7809 tokeniser tests :-)
  38. # [00:13] <Philip> (Hmm, tokenisers don't have pretty colours, which destroys my theory)
  39. # [00:13] <Dashiva> Maybe we need someone to rate the different parts of the spec and assign testcase points of different amounts
  40. # [00:13] <anne> (you did generate graphs out of the tokenizer spec)
  41. # [00:14] <anne> one problem with any kind of testsuite organization at this point is that the specification isn't stable
  42. # [00:16] <gsnedders> kingryan: <http://hg.gsnedders.com/cgi-bin/hgwebdir.cgi/php-html-5-direct/file/d050bcea1a03/tests/numbersTest> now I've reuploaded them
  43. # [00:17] <Philip> Dashiva: It could be turned into an RPG where you get experience points for writing test cases, and whoever reaches Level 60 first wins a T-shirt
  44. # [00:17] <kingryan> gsnedders: cool. any explanation on what the data means? :)
  45. # [00:17] <gsnedders> anyhow, time for me to go to sleep so I can get up in time for school tomorrow (where I can try and get a life)
  46. # [00:17] <gsnedders> kingryan: see README is the same dir
  47. # [00:17] <Dashiva> Philip: "I wrote a million testcases, and all I got was this lousy candidate recommendation"
  48. # [00:18] <kingryan> ah, thanks gsnedders
  49. # [00:32] <Philip> Argh, there's too many bugs
  50. # [00:35] * Quits: tH (Rob@87.102.45.182) (Quit: ChatZilla 0.9.78.1-rdmsoft [XULRunner 1.8.0.9/2006120508])
  51. # [00:35] <Dashiva> Do you get that warm, happy feeling inside when you discover one and can make a testcase to display it?
  52. # [00:39] <Hixie> the problem is when you find more bugs while making the testcase
  53. # [00:39] <Hixie> then the warm happy feeling starts feeling more like standing on quicksand
  54. # [00:53] <Philip> I'm mostly just looking at my old canvas tests, so I'm not even discovering anything new :-(
  55. # [00:54] <Philip> (Found one Opera bug which I don't think I knew about before, though)
  56. # [00:55] * Joins: karl (karlcow@128.30.52.30)
  57. # [00:57] <Hixie> no comment: http://c2.com/~ward/ascent.jpg
  58. # [01:01] * Quits: ChrisWilson (cwilso@131.107.0.102) (Ping timeout)
  59. # [01:06] * Quits: gavin (gavin@99.227.30.12) (Ping timeout)
  60. # [01:07] * Quits: hasather (hasather@90.231.107.133) (Quit: leaving)
  61. # [01:11] * Joins: gavin (gavin@99.227.30.12)
  62. # [01:24] * Joins: ChrisWilson (cwilso@131.107.0.73)
  63. # [01:29] * Quits: kingryan (rking3@208.66.64.47) (Quit: kingryan)
  64. # [01:34] * Philip wonders why Opera 9.5 says http://philip.html5.org/tests/canvas/misc/exceptions.html fails the toDataURL test
  65. # [01:39] <Philip> It seems to think typeof(/foo/) != 'object', which is peculiar
  66. # [01:41] * Joins: sbuluf (ebeok@200.49.140.153)
  67. # [01:48] <Philip> http://philip.html5.org/tests/js/typeof-regexp.html makes no sense to me
  68. # [01:48] * Philip is getting the quicksand feeling
  69. # [02:06] * Quits: aroben_ (aroben@17.203.12.72) (Connection reset by peer)
  70. # [02:06] * Joins: aroben_ (aroben@17.203.12.72)
  71. # [02:46] * Quits: ChrisWilson (cwilso@131.107.0.73) (Ping timeout)
  72. # [02:56] * Joins: MikeSmith (MikeSmith@mcclure.w3.org)
  73. # [03:28] * Quits: gavin (gavin@99.227.30.12) (Ping timeout)
  74. # [03:33] * Joins: gavin (gavin@99.227.30.12)
  75. # [03:51] * Quits: dbaron (dbaron@63.245.220.241) (Quit: 8403864 bytes have been tenured, next gc will be global.)
  76. # [04:00] * Quits: aroben_ (aroben@17.203.12.72) (Quit: Leaving)
  77. # [04:21] * Joins: dbaron (dbaron@71.204.145.103)
  78. # [04:34] * Quits: mjs (mjs@17.255.106.186) (Quit: mjs)
  79. # [05:35] * Quits: gavin (gavin@99.227.30.12) (Ping timeout)
  80. # [05:40] * Joins: gavin (gavin@99.227.30.12)
  81. # [05:46] * Joins: mjs (mjs@64.81.48.145)
  82. # [06:19] * Quits: mjs (mjs@64.81.48.145) (Quit: mjs)
  83. # [06:33] * Joins: mjs (mjs@64.81.48.145)
  84. # [07:01] <sbuluf> karl , are you around? if so, some reactions on semantic web in slashdot might interest you: http://slashdot.org/article.pl?sid=07/10/29/1435254
  85. # [07:42] * Quits: gavin (gavin@99.227.30.12) (Ping timeout)
  86. # [07:47] * Joins: gavin (gavin@99.227.30.12)
  87. # [07:50] * Quits: dbaron (dbaron@71.204.145.103) (Quit: 8403864 bytes have been tenured, next gc will be global.)
  88. # [08:01] * Joins: billyjack (MikeSmith@mcclure.w3.org)
  89. # [08:02] * Quits: MikeSmith (MikeSmith@mcclure.w3.org) (Client exited)
  90. # [08:04] * billyjack is now known as MikeSmith
  91. # [08:29] * Quits: karl (karlcow@128.30.52.30) (Quit: Where dwelt Ymir, or wherein did he find sustenance?)
  92. # [08:41] * Quits: drry (drry@222.225.140.32) (Quit: drry)
  93. # [08:43] * Joins: drry (drry@222.225.140.32)
  94. # [09:32] <hsivonen> anne: some of the XForms 1.0 3rd ed diffs look rather substantive, too.
  95. # [09:33] * Joins: tH_ (Rob@87.102.45.182)
  96. # [09:33] * tH_ is now known as tH
  97. # [09:39] * Dashiva pokes Philip
  98. # [09:40] <Dashiva> You might interested to learn that typeof is an operator, not a function. And as such it has precedence that might be surprising :)
  99. # [10:07] <Philip> Dashiva: It has higher precedence than != so it shouldn't be surprising in this particular case :-)
  100. # [10:08] <Philip> (though I was thinking of it like a function, so I'll have to be more careful about that...)
  101. # [10:15] <anne> did you file bugs on the quicksand feeling?
  102. # [10:16] <Philip> Yes
  103. # [10:16] <Philip> (Well, only one bug, since that's all I found so far)
  104. # [10:23] * Quits: MikeSmith (MikeSmith@mcclure.w3.org) (Quit: Less talk, more pimp walk.)
  105. # [10:25] * Quits: Lachy (Lachy@213.236.208.22) (Quit: Leaving)
  106. # [10:25] * Joins: Lachy (Lachy@213.236.208.22)
  107. # [10:39] <anne> heh, replies from my fora@annevankesteren.nl days
  108. # [10:42] <anne> Dashiva, Firefox 3 actually has an implementation of access-control
  109. # [11:35] * Joins: myakura (myakura@122.17.98.20)
  110. # [11:45] * Quits: gavin (gavin@99.227.30.12) (Ping timeout)
  111. # [11:45] * Quits: sbuluf (ebeok@200.49.140.153) (Quit: sbuluf)
  112. # [11:50] * Joins: gavin (gavin@99.227.30.12)
  113. # [11:50] <anne> Philip, drawImage(SVGSVGElement, ...) requires the height/width arguments for the image size, it seems
  114. # [11:51] <anne> Philip, http://tc.labs.opera.com/html/canvas/svg/002.xml has a working example
  115. # [11:54] * Joins: ROBOd (robod@89.122.216.38)
  116. # [11:57] <Philip> anne: Ah, okay
  117. # [11:57] <Philip> It still gets scaled uglily :-(
  118. # [11:58] <anne> yeah, known
  119. # [11:59] <Philip> Is it intentional that width/height are required, or is that a bug?
  120. # [12:00] <anne> bug imo
  121. # [12:00] <anne> see 001.xml :)
  122. # [12:02] <Philip> http://philip.html5.org/demos/canvas/svg/scale.html - looks like the <object> itself gets clipped after it's been drawImaged
  123. # [12:05] <Philip> Also, it doesn't work at all if sx != 0 or sy != 0
  124. # [12:09] <anne> sx and sy are the offset arguments?
  125. # [12:10] <anne> oh, I see
  126. # [12:11] <Philip> ctx.drawImage(svg, 0, 0, 100, 100, 0, 0, 100, 100) works, ctx.drawImage(svg, 1, 0, 100, 100, 0, 0, 100, 100) says INDEX_SIZE_ERR
  127. # [12:12] <Philip> Oh, it's just throwing if sx+sw > 100
  128. # [12:13] <Philip> so you can only select the top 100x100 pixels of the source image
  129. # [12:13] <Philip> (for suitable definitions of 'pixel')
  130. # [12:13] <anne> even if the source image is larger?
  131. # [12:13] <anne> for a suitable definition of "larger"
  132. # [12:13] <Philip> Yes
  133. # [12:14] <anne> weird
  134. # [12:15] * Philip doesn't have time to try proper test cases for these things right now, unfortunately
  135. # [12:21] * Joins: olivier (ot@128.30.52.30)
  136. # [13:51] * Joins: aaronlev (chatzilla@66.31.86.217)
  137. # [13:52] * Quits: gavin (gavin@99.227.30.12) (Ping timeout)
  138. # [13:57] * Joins: gavin (gavin@99.227.30.12)
  139. # [13:59] * Joins: quark (hgfgfg@91.121.13.62)
  140. # [14:06] * Quits: quark (hgfgfg@91.121.13.62) (Quit: quark)
  141. # [15:19] * Quits: olivier (ot@128.30.52.30) (Quit: This computer has gone to sleep)
  142. # [15:20] * Joins: olivier (ot@128.30.52.30)
  143. # [15:20] * Quits: olivier (ot@128.30.52.30) (Client exited)
  144. # [15:21] * Joins: olivier (ot@128.30.52.30)
  145. # [15:29] * Quits: myakura (myakura@122.17.98.20) (Ping timeout)
  146. # [15:29] * Joins: myakura (myakura@122.17.98.20)
  147. # [15:33] <hsivonen> anne: I implemented the error message rendering features we talked about yesterday: http://html5.validator.nu/?doc=http%3A%2F%2Fhsivonen.iki.fi%2Ftest%2Fmoz%2Felaboration-demo.xhtml
  148. # [15:34] * Joins: billmason (billmason@69.30.57.156)
  149. # [15:34] <anne> nice
  150. # [15:42] <anne> public-html-comments has only 11 people subscribed ( http://cgi.w3.org/member-bin/MailingListQuery.pl?queryList=public-html-comments )
  151. # [15:44] <anne> then again, only 2 actual comments
  152. # [15:47] <hsivonen> hmm. I guess I should subscribe
  153. # [15:47] <anne> i just did, I sort of forgot about it
  154. # [15:49] <Lachy> hmm. I wonder why I'm not subscribed, I'm sure I subscried when it was first set up.
  155. # [15:49] <Lachy> I'll have to check when I get home.
  156. # [15:57] <anne> hsivonen, as feature request, it should probably also indicate that an element doesn't exist in a particular vocabulary at all; in case people spell <img> as <image> in XHTML for instance
  157. # [15:59] <hsivonen> anne: yeah, the demo document was supposed to elicit that message from the RNG engine. I have no idea why it doesn't.
  158. # [16:00] * Quits: gavin (gavin@99.227.30.12) (Ping timeout)
  159. # [16:01] <hsivonen> anne: hmm. perhaps the schema has a wild card somewhere, so it theory all names could exist somewhere...
  160. # [16:02] <hsivonen> unintended consequences...
  161. # [16:02] <anne> do you have a regression testing system in place already?
  162. # [16:03] <hsivonen> anne: only for parts of the schemas. not for the whole system
  163. # [16:03] <hsivonen> it is probably worth discussing in the unconference
  164. # [16:04] <hsivonen> I guess I could do full stack testing through HTTP and compare responses to known-good responses
  165. # [16:05] * Joins: gavin (gavin@99.227.30.12)
  166. # [16:05] <hsivonen> the problem is that a slightest improvement to error reporting would make large chunks of known-good data stale
  167. # [16:05] <anne> maybe each error should have some kind of type
  168. # [16:06] <anne> then you can write tests like "<!DOCTYPE html><title></title><body><bar></bar></body>", "unknown-element" or something like that
  169. # [16:06] <hsivonen> anne: depends on what constitutes "each"
  170. # [16:06] <hsivonen> anne: now the RNG errors have distinct exception classes
  171. # [16:07] <hsivonen> but there are only 10 of them
  172. # [16:08] <hsivonen> so recording only class identity would not capture the particular parameters (what element is not allowed)
  173. # [16:08] <anne> I'm wondering if you would share tests among validators or if you just need a single validator
  174. # [16:08] <hsivonen> well, that's a big issue as far as error identity goes
  175. # [16:08] <anne> if it's about a single validator you don't need too much abstraction, just a simple way to make a new test
  176. # [16:08] <hsivonen> because error identity is heavily implementation-dependent
  177. # [16:09] <anne> if it's about multiple validators you need to define some type of generic error classes...
  178. # [16:09] <hsivonen> that's the main failure of EARL
  179. # [16:09] <hsivonen> EARL assumes that each type of error has a URI
  180. # [16:10] <hsivonen> but the assumption does not work when the errors are grammar-based
  181. # [16:10] <hsivonen> anne: what about checking implmented in Schematron? to give errors common identity or distinct Java error classes, you'd need to drill a hole into the Schematron engine
  182. # [16:12] <hsivonen> every time there's some kind of abstraction layer in the path of error message generation, tracking the identity of errors becomes hard
  183. # [16:12] <hsivonen> and it would suck to limit possible implementation choices to match the notion of identity in a blessed test suite
  184. # [16:13] <anne> that was not my idea (you'd have subtypes), but given the complexity I'm not sure it's worth it to try for generic regression testsuites as we've done with the parser tests
  185. # [16:14] <hsivonen> anne: generic boolean suites would work
  186. # [16:14] <hsivonen> "these docs are valid", "these docs are invalid"
  187. # [16:15] <anne> true, those are pretty trivial
  188. # [16:23] <anne> maybe you can do something similar to the parser tests
  189. # [16:23] <anne> this document is invalid because it has 10 errors
  190. # [16:23] <anne> this document is valid because it has 0 errors
  191. # [16:23] <anne> and since we initially make them for your validator those error lines are based on whatever your validator generates
  192. # [16:24] <hsivonen> anne: counting errors is hard, too, because when you have a misplaced element, the stuff that happens in the subtree is likely implementation-defined
  193. # [16:25] <anne> maybe if you have 10 error lines an implementation needs to find 10 errors or more...
  194. # [16:25] <anne> hmm
  195. # [16:26] <hsivonen> precise error locations are implementation-specific, too
  196. # [16:26] <hsivonen> for example, Validator.nu tries to highlight the entire tag
  197. # [16:26] <hsivonen> the W3C Validator highlight the last character of the tag
  198. # [16:27] <hsivonen> a more sophisticated implementation could highlight an individual attribute inside a tag
  199. # [16:27] <hsivonen> it would help a bit, though, if a test suite had each tag fully on a line
  200. # [16:28] <hsivonen> and then only line numbers were compared
  201. # [16:32] <hsivonen> btw, counting errors is inconsistent on the HTML5 tree construction layer is practice
  202. # [16:33] <hsivonen> that is, the Validator.nu parser, html5lib and the spec sometimes disagree about the number of errors caused by a single thingy in the source
  203. # [16:45] <DanC> the traditional formal methods thing to do with syntax errors is to compute the first part of the document that can't match the grammar
  204. # [16:46] * Quits: myakura (myakura@122.17.98.20) (Quit: Leaving...)
  205. # [16:48] <hsivonen> DanC: that assumes that there is "the grammar"
  206. # [16:49] <hsivonen> DanC: If we were to make an implementation-independent validation test suite, it should probably focus on the approximate (tag length detail level) position of the first error in source order
  207. # [16:50] <DanC> well, it traditionally uses "the grammar", but it only requires "the language".
  208. # [16:51] <DanC> i.e. for a character sequence S, you find the longest initial sequence I of S such that there's some string in the language that starts with I.
  209. # [16:51] <DanC> if I=S, then S is in the language; else there's a syntax error.
  210. # [16:52] * Joins: Lachy_ (Lachlan@88.91.111.68)
  211. # [16:52] <DanC> but for cases like <a href="path#frag#oops"> it's much more helpful to have the whole tag as context, rather than saying "error at #oops"
  212. # [16:53] <DanC> (does html5 constrain the value o the href attribute to be a real URI reference? I don't even know.)
  213. # [16:53] <hsivonen> DanC: it does
  214. # [16:54] <hsivonen> DanC: but it doesn't say which IRI schemes validators must or must not have scheme-specific knowledge about
  215. # [16:54] <DanC> multiple #'s is an error regardless of scheme
  216. # [18:54] * Disconnected
  217. # [18:54] * Attempting to rejoin channel #html-wg
  218. # [18:54] * Rejoined channel #html-wg
  219. # [18:54] * Topic is 'next HTML WG telcon 25 Oct 2300Z http://www.w3.org/html/wg/ (more logs: http://krijnhoetmer.nl/irc-logs/ )'
  220. # [18:54] * Set by DanC on Mon Oct 22 15:50:08
  221. # [19:11] * Quits: laplink (link@193.157.66.6) (Quit: Leaving)
  222. # [19:15] * Quits: matt (matt@128.30.52.30) (Quit: matt)
  223. # [19:26] * DanC changes topic to 'next HTML WG telcon 1 Nov 16:00 UTC http://www.w3.org/html/wg/ (more logs: http://krijnhoetmer.nl/irc-logs/ )'
  224. # [19:31] * Joins: mjs (mjs@17.255.106.186)
  225. # [19:32] <ChrisWilson> *taptap* is this thing on?
  226. # [19:33] <Hixie> yup
  227. # [19:37] * Quits: mjs (mjs@17.255.106.186) (Quit: mjs)
  228. # [19:38] * Joins: mjs (mjs@17.255.106.186)
  229. # [20:00] <anne> ah, there will be a telcon...
  230. # [20:01] <anne> 16:00 UTC is 17:00 Amsterdam time these days, right?
  231. # [20:01] * anne can make that
  232. # [20:05] <anne> http://www.heise.de/english/newsticker/news/98175 (from #whatwg) is interesting
  233. # [20:06] <mjs> today?
  234. # [20:06] <anne> seems like it
  235. # [20:09] <anne> I suppose they follow the blog
  236. # [20:10] * Joins: matt (matt@128.30.52.30)
  237. # [20:14] * Quits: gavin (gavin@99.227.30.12) (Ping timeout)
  238. # [20:19] * Joins: gavin (gavin@99.227.30.12)
  239. # [21:00] <Hixie> anne: on Nov 26, 2006 2:38 AM you wrote from your opera address in an e-mail entitled "Text nodes and inter-element whitespace" the following:
  240. # [21:00] <Hixie> Why isn't the following a conforming block of computer output too:
  241. # [21:00] <Hixie> <pre>
  242. # [21:00] <Hixie> <!-- XXX ... -->
  243. # [21:00] <Hixie> <samp>12.12...</samp>
  244. # [21:00] <Hixie> <?test ...?>
  245. # [21:00] <Hixie> </pre>
  246. # [21:00] * Hixie does not understand that question and would like some help
  247. # [21:01] <Hixie> ohhhh
  248. # [21:01] <anne> I think it's about the definition of <samp> within <pre> not ignoring stuff
  249. # [21:01] <Hixie> i know what you're saying
  250. # [21:01] <Hixie> yeah
  251. # [21:01] <Hixie> ok
  252. # [21:01] * Hixie moves it from "semantics" to "semantics-phrasing"
  253. # [21:03] <Hixie> hey anyone still have the link to the reg page for the tpac?
  254. # [21:03] <anne> http://www.w3.org/2007/11/TPAC/
  255. # [21:04] <Hixie> aha thanks
  256. # [21:06] <Hixie> wow, htmlwg beat mwi for number of participants
  257. # [21:19] * Quits: timbl_ (timbl@128.30.6.228) (Quit: timbl_)
  258. # [21:20] * Joins: timbl_ (timbl@128.30.6.228)
  259. # [21:26] <gsnedders> mwi?
  260. # [21:27] <Hixie> mobile web initiative. they had like 60 people oen year.
  261. # [21:27] <Hixie> one
  262. # [21:27] <dbaron> I thought that was mmi.
  263. # [21:27] <Hixie> oh
  264. # [21:28] <Hixie> maybe
  265. # [21:28] <Hixie> hm
  266. # [21:28] <dbaron> which I think has been renamed to uwa
  267. # [21:29] <mjs> hello dbaron
  268. # [21:29] <dbaron> hi
  269. # [21:29] <anne> uwa is also the former device independent wg, at least
  270. # [21:30] <dbaron> oh
  271. # [21:30] <dbaron> mmi still exists
  272. # [21:32] * anne remembers a joint meeting between CDF and MMI
  273. # [21:32] <gsnedders> CDF?
  274. # [21:32] <gsnedders> :P
  275. # [21:32] <anne> http://www.w3.org/2004/CDF/
  276. # [21:47] * Quits: matt (matt@128.30.52.30) (Quit: matt)
  277. # [21:57] <DanC> speaking of CDF, anybody know the writer/journalist that wrote this?
  278. # [21:57] <DanC> Former OpenDocument advocates bolt for W3C standard
  279. # [21:57] <DanC> <http://www.news.com/8301-10784_3-9806369-7.html>
  280. # [21:57] <DanC> hmm... how many participants do we have? looking...
  281. # [22:07] <Hixie> wow, someone thinks cdf is going to get higher word fidelity than odf?
  282. # [22:08] <DanC> and somehow an OpenDocument Foundation got started by some folks that don't seem to like ODF.
  283. # [22:09] <Dashiva> If ODF is too closely linked to OpenOffice, what does that make OOXML?
  284. # [22:10] <DanC> there's a wikipedia article on CDF that looks like it's written by the same crowd
  285. # [22:11] <anne> wow, CDF was never releated to ODF at all
  286. # [22:11] <ChrisWilson> Dashiva: what's the relevance? The article is about
  287. # [22:11] <anne> it's just a "fancy" name for HTML+SVG+...
  288. # [22:11] <ChrisWilson> not tied to OpenOffice (or not)
  289. # [22:11] <ChrisWilson> (sorry about the double negative).
  290. # [22:12] <ChrisWilson> OOXML is certainly more interoperable with the "1/2-odd-billion" MS Office desktops out there.
  291. # [22:12] <ChrisWilson> Not that it doesn't have its own set of detractions. :)
  292. # [22:13] <ChrisWilson> anne: that's part of the point of the article, isn't it?
  293. # [22:13] <DanC> but interestingly, the article mentions W3C and CDF by name, but not HTML... nor even WICD, which is clearly a cooler name than CDF
  294. # [22:13] <anne> is it?
  295. # [22:13] <Dashiva> ChrisWilson: Sure, but OOXML isn't any less specific to its apps than ODF
  296. # [22:14] <anne> oh well, time to go
  297. # [22:14] * gsnedders waves
  298. # [22:14] <ChrisWilson> well, sort of. I read it as "ex-ODf advocates are jumping ship to CDF/web, because ODF is hard to interoperate with Office, while CDF is easier".
  299. # [22:15] <DanC> or at least "while we hope someday CDF will be easier."
  300. # [22:15] * DanC wouldn't hold my breath
  301. # [22:15] <ChrisWilson> Dashiva: I'm not personally arguing for OOXML as a solution to any generic problems.
  302. # [22:15] <ChrisWilson> heh. Well, Dan, inasmuch as CDF "=" HTML+SVG+CSS+Etc, it probably *IS* already easier to interoperate with Word docs, at least.
  303. # [22:16] <DanC> hmm... maybe so
  304. # [22:18] <Dashiva> ChrisWilson: Neither am I, just the random jumping between "interop is hard" and "format is too specific" seemed odd to me. Nothing to read too much into. :)
  305. # [22:19] <ChrisWilson> yeah, that's true.
  306. # [22:21] * Quits: gavin (gavin@99.227.30.12) (Ping timeout)
  307. # [22:24] <DanC> this is a common pattern... "implementations of spec X don't interoperate as well as I'd like, so I'll create spec Y. surely the grass will be greener."
  308. # [22:24] <Hixie> yeah
  309. # [22:24] <Hixie> i never understood that
  310. # [22:24] <DanC> I guess it works occasionally.
  311. # [22:24] <Hixie> "implementations of X have bugs, our technology Y will solve these problems!"
  312. # [22:25] <DanC> ah yes. that would perhaps work if Y included a time machine.
  313. # [22:25] <Hixie> and even then!
  314. # [22:25] <Hixie> i blogged about this once iirc
  315. # [22:26] <Hixie> http://ln.hixie.ch/?start=1108984991&count=1
  316. # [22:26] * Joins: gavin (gavin@99.227.30.12)
  317. # [22:27] <DanC> cnet again... coincidence? hmm.
  318. # [22:31] <DanC> speaking of http://ln.hixie.ch/ ... that's an odd branch of bistromathmatics there, hixie.
  319. # [22:31] <Hixie> hehe
  320. # [22:33] * Quits: ROBOd (robod@89.122.216.38) (Quit: http://www.robodesign.ro )
  321. # [22:33] <ChrisWilson> Actually, maybe it's proof that bistromathics exists, or your scheme would have worked.
  322. # [22:34] <Hixie> good point
  323. # [22:44] * Quits: kingryan (rking3@208.66.64.47) (Quit: kingryan)
  324. # [22:44] * Quits: timbl_ (timbl@128.30.6.228) (Quit: timbl_)
  325. # [22:58] * Quits: dbaron (dbaron@63.245.220.241) (Quit: 8403864 bytes have been tenured, next gc will be global.)
  326. # [23:04] * Quits: mjs (mjs@17.255.106.186) (Quit: mjs)
  327. # [23:07] * Joins: mjs (mjs@17.255.106.186)
  328. # [23:21] * Joins: laplink (link@62.92.248.70)
  329. # [23:26] * Quits: mjs (mjs@17.255.106.186) (Quit: mjs)
  330. # [23:29] * Quits: Sander (svl@86.87.68.167) (Quit: And back he spurred like a madman, shrieking a curse to the sky.)
  331. # [23:30] * Joins: mjs (mjs@17.255.106.186)
  332. # [23:30] <DanC> hmm... the data from this "what should we publish first" survey isn't very good. http://www.w3.org/2002/09/wbs/40318/wd7/results I suppose I could have put more thought into it when I re-opened it
  333. # [23:31] <DanC> "Which documents do you think are good for the group to publish soon but only after critical issues are addressed?" was really poor. I meant to ask "what stop-ship problems do you see?"
  334. # [23:34] <DanC> and it's hard to tell which people answered when
  335. # [23:39] <Hixie> the whatwg published a copy of the html5 spec on friday, btw, so the html5 spec is basically ready for a fpwd
  336. # [23:39] * Joins: timbl (timbl@146.115.66.146)
  337. # [23:40] * jgraham is +1 publish and be damned
  338. # [23:40] <jgraham> Either document.
  339. # [23:41] <Hixie> yeah i don't really understand why we haven't published yet
  340. # [23:41] <Hixie> we could have published in june as far as i can see
  341. # [23:42] <ChrisWilson> well, today's the first time you've frozen a copy of the spec for review.
  342. # [23:43] <Hixie> ?
  343. # [23:43] <Hixie> how do you mean?
  344. # [23:43] * Hixie doesn't see that he's done anything differently today vs any other day
  345. # [23:44] <Hixie> am i missing something?
  346. # [23:44] <ChrisWilson> sorry, not today - I meant Saturday.
  347. # [23:44] <ChrisWilson> (time warped with another threaD)
  348. # [23:44] <Hixie> still not sure what you mean
  349. # [23:45] <ChrisWilson> It's hard to do a thorough review when the laser ink is hardly dry on the paper before it's out of date.
  350. # [23:45] <Hixie> but what's frozen?
  351. # [23:45] <Hixie> i don't follow
  352. # [23:45] <Hixie> you mean the whatwg publishing a snapshot is a necessary step to the w3c publishing a snapshot?
  353. # [23:45] <Hixie> or...?
  354. # [23:45] <Hixie> *confused*
  355. # [23:46] <ChrisWilson> I believe we (HTML WG) as a group decided to take the WHATWG work as a starting point, and you are attempting to keep the HTML WG HTML5 and the WHATWG HTML5 up to date, yes?
  356. # [23:46] <Hixie> yes
  357. # [23:46] <Hixie> they're built from the same sources
  358. # [23:47] <ChrisWilson> For anyone to review that specification, and comment (or even, at a more basic level for the HTML WG, to compare that spec to our charter), sufficient time to review needs to be allowed, right?
  359. # [23:47] <ChrisWilson> As you said, "If anyone was hoping for a semi-stable version to start reviewing the draft, I would say that this is it. "
  360. # [23:48] <Hixie> i'm not aware of any time limit on reviews, if that's what you mean... but presumably that would be a w3c policy thing
  361. # [23:49] <jgraham> The rate at which the spec is changing is small compared to the size of the spec.
  362. # [23:49] <ChrisWilson> No, I don't mean a time limit. I mean HTML5 is a ream of paper, approximately. That's a lot of detailed review required. I would presume there would be a stable version to review.
  363. # [23:49] <ChrisWilson> hmm.
  364. # [23:49] <DanC> I haven't seen many (any?) problems where reviewers sent comments that were obsolete by the time they were considered, though I don't really know what the latency is.
  365. # [23:49] <jgraham> If you want to review it, it works pretty well to print a copy on any random day and review that
  366. # [23:49] <ChrisWilson> jgraham: That may well be true. I'm having trouble knowing where to review in order, though.
  367. # [23:50] <Hixie> ChrisWilson: if you mean that i've stopped making changes since publishing the snapshot friday, that's wrong -- there's been a dozen or more checkins since friday
  368. # [23:50] <ChrisWilson> sigh.
  369. # [23:50] <jgraham> ChrisWilson: I don't understand what you mean "in order"?
  370. # [23:50] <Hixie> i have 3600+ e-mails to reply to, i'm not gonna stop working on them :-)
  371. # [23:50] <Lachy_> I don't see the what the benefit is in having a snapshot to review vs. a version that is being updated in response to the reviewers feedback
  372. # [23:50] <Hixie> i mean, even if we froze all development until all feedback was in, by the time i got to the 3000th e-mail, the spec wouldn't be the one that it was when the 1st was read
  373. # [23:50] <ChrisWilson> jgraham: in order to not be tromping on whatever area Ian's currently working on.
  374. # [23:50] * DanC sees benefits to both
  375. # [23:51] <jgraham> ChrisWilson: Ah. Well I agree that's not quite transparent but you can always ask Ian, check the svn logs or looka at the mailing lust
  376. # [23:51] <jgraham> s/lust/list/
  377. # [23:52] <jgraham> e.g. recently the database stuff has been changing pretty rapidly
  378. # [23:52] <ChrisWilson> Okay, well, then perhaps you should tell me how I'm supposed to comment. I see big issues that need discussion in the HTML WG (canvas and some of the APIs not really being in our charter, for example), big issues masquerading as small issues ("versioning"), and small issues that may change rapidly (classid/guid removed off object).
  379. # [23:52] <Hixie> stuff changes in response to feedback
  380. # [23:53] <ChrisWilson> um, yes.
  381. # [23:53] <ChrisWilson> I would hope.
  382. # [23:53] <Hixie> so if i get feedback from person A while person B is reviewing the spec, the spec will change before B is done
  383. # [23:53] <Hixie> but equally, if I wait for B to be done, then respond to A, then the spec will have still changed by the time I get to B's feedback
  384. # [23:53] <Hixie> what's the difference?
  385. # [23:54] <Hixie> i'd like the feedback to be on the wiki with a wiki page per "problem", at least for anything involving design decisions
  386. # [23:54] <Hixie> stuff that's obviously just an error (e.g. contradictions) can just be sent as e-mail, they're not controversial
  387. # [23:55] * Joins: hober (ted@68.107.112.172)
  388. # [23:55] <Lachy_> ChrisWilson, I don't how the fact that the spec is changing is preventing you from commenting on it, particularly the relatively stable sections
  389. # [23:56] <ChrisWilson> I didn't say it was. I said it wasn't surprising that we hadn't published, given that reviewing a moving target doesn't imbue confidence in what we would publish.
  390. # [23:57] <ChrisWilson> (to sum up)
  391. # [23:57] <Hixie> i don't understand that
  392. # [23:57] <mjs> we should totally publish
  393. # [23:57] <Hixie> isn't any snapshot always going to be like that?
  394. # [23:57] <Hixie> the only way the spec could not change after publication is if the spec was perfect
  395. # [23:57] <Hixie> in which case we wouldn't be expecting feedback
  396. # [23:58] <Hixie> and it seems unlikely we'll get there ever, let alone before ever publishing a call for comments
  397. # [23:58] * DanC is updating http://esw.w3.org/topic/HTML/SpecReviews wondering if each section has been reviewed by some HTML WG members... wonders about preformatted text...
  398. # Session Close: Wed Oct 31 00:00:00 2007

The end :)