Options:
- # Session Start: Sun Jul 22 00:00:00 2007
- # Session Ident: #html-wg
- # [00:00] * Quits: briansuda (briansuda@85.220.95.76) (Quit: briansuda)
- # [00:00] <anne> I suppose it's the acknowledgments list that mentions IE explicitly
- # [00:00] <anne> and Safari
- # [00:01] <hsivonen> ah
- # [00:02] <anne> from #xhtml: " good way to get out of CR -- implement parts of XHTML2 in any browser with assistance of script"
- # [00:02] <hsivonen> aargh.
- # [00:03] <anne> they seem to be more concerned with getting the spec done than with giving developers something useful
- # [00:04] * Parts: hasather (hasather@81.235.209.174)
- # [00:04] <Philip`> Isn't it useful to implement parts of XHTML2 with scripts, like how WF2 and <canvas> can be done?
- # [00:04] <anne> sure, but not to exit CR
- # [00:04] <anne> Philip`, did you check in your 10000 tests?
- # [00:05] <Philip`> I didn't, since they're mostly pointlessly repetitive and they take ages to run
- # [00:05] <anne> oh ok
- # [00:06] <Philip`> (assuming you mean the tokeniser tests)
- # [00:06] <anne> yeah
- # [00:06] <anne> quite nice that you couldn't find much bugs btw :)
- # [00:06] <hsivonen> I wonder if native XHTML2 impls become impossible to deploy without Breaking The Web once a handful of sites starts using kewl XHTML2 plus scripts that turn it into HTML
- # [00:08] * anne isn't sure what to make of the thread on <code> and <pre>
- # [00:08] * anne thought it was pretty clear that <pre> indicates that the whitespace it encloses is significant
- # [00:08] <Philip`> Why would they turn it into HTML rather than into XHTML1?
- # [00:09] <Philip`> The latter should work in at least two browsers, and then it'd do stuff like proper XML parsing, and there wouldn't be so many ways in which a native XHTML2 implementation would severely break sites
- # [00:11] <hsivonen> Philip`: the problem exists even if transforming to XHTML1
- # [00:12] <hsivonen> hmm. view source and English reading comprehension: the instant semantic UA
- # [00:14] <Philip`> That's not very good for accessibility - people browsing the web on mobile phones don't (probably?) have a view source button
- # [00:16] <hsivonen> Philip`: the source becomes more accessible if you "quote your attributes" and "close your tags"
- # [00:19] * Philip` should update his HTML-stats code to see how many people quote attributes and which tags they do/don't close
- # [00:20] <Philip`> (I used to collect that data, but then I stopped because it was boring and I really didn't care, so actually maybe I should bother doing it again)
- # [00:21] <hsivonen> Philip`: wouldn't you need to instrument a tree builder for the latter? or do you mean />?
- # [00:22] <Philip`> I just mean counting e.g. the number of "li" start tags and the number of "li" end tags, and subtracting
- # [00:23] <hsivonen> ok
- # [00:23] <Philip`> which would be a rough approximation of how many people use optional end tags
- # [00:26] <Philip`> (I always used to think <p> was a paragraph separator, so I'd never use a </p>, but I didn't know </td> was optional so I always included that)
- # [00:26] <Philip`> (so maybe it'd be vaguely interesting to see how other people use these things)
- # [00:27] <hsivonen> there's some historical revisionism around the containerness of <p>
- # [00:37] <tH> i'm finding more end tags than start tags for a lot of elements, which is... interesting.
- # [00:38] <anne> did you study the pages where it occured?
- # [00:39] <tH> not yet, i have a feeling it is just one or two really broken pages messing up the stats
- # [00:40] <anne> pretty cool that so many people are doing studies now
- # [00:45] <Philip`> Oh, I've still got the end tag stats from the ~2000 pages I looked at ages ago
- # [00:47] <Philip`> For the end tag : start tag ratio, 'ziffhtml' wins with a score of infinity
- # [00:47] <Philip`> and there's six more </applet> than there is <applet>
- # [00:48] <Philip`> 55K <p> vs 40K </p>
- # [00:48] <Philip`> 159K <td> vs 158K </td>
- # [00:50] <Philip`> 467 <embed> vs 314 </embed>
- # [00:51] <Philip`> 59K <li> vs 53K </li>
- # [00:51] <Philip`> (I hope I've not got some horrible error in my hacked-together analysis code)
- # [00:51] <Philip`> (but I guess those numbers sound reasonable)
- # [00:52] <tH> those look like they're similar to mine
- # [00:53] <tH> so maybe a quarter of the world is deliberately leaving out </p> :P
- # [00:54] <Philip`> I only see one missed </html>
- # [00:55] <Philip`> People waste so much bandwidth on optional tags :-(
- # [00:57] <tH> http://bugs.rdmsoft.com/html/stats/2007-07-18/endtags
- # [00:57] <tH> plenty of missing </dt> and </dd> compared to how much they're used
- # [00:57] <tH> (insert usual disclaimer about my stats probably being all wrong)
- # [00:58] <Philip`> Could you show diff as a percentage of start?
- # [01:01] <Philip`> Hmm, your numbers of start tags seem significantly different to what I have at http://canvex.lazyilluminati.com/misc/stats/analyse.cgi/index
- # [01:02] <Philip`> Oops, is that my old one?
- # [01:03] <Philip`> Oh, yes
- # [01:03] <Philip`> http://canvex.lazyilluminati.com/survey/2007-07-17/analyse.cgi/index seems closer
- # [01:05] <Philip`> The /stats/ one is from the Alexa top 500, which is presumably why it's significantly biased away from <td> and <br>
- # [01:06] <tH> my pages come from web searches for random words, that might bias it towards newer pages with less tables, but that's only a guess
- # [01:06] <tH> I added percentages to the endtags page
- # [01:07] <Philip`> That's an awful lot of </nobr>
- # [01:12] <Philip`> http://bugs.rdmsoft.com/html/stats/2007-07-18/class - that's also rather a lot of donMenuText - can you see what pages that comes from?
- # [01:12] <tH> i have a page with 9301 </nobr> and 0 <nobr>
- # [01:13] <tH> it also has 27922 <font>
- # [01:13] <Philip`> Ooh, great fun
- # [01:14] <tH> all the donMenuText are from one page
- # [01:15] <Philip`> It's good to know the target audience which HTML5 has to be designed for
- # [01:15] <Philip`> or at least designed to cope with
- # [01:15] * tH waits for firefox to unhang loading the page
- # [01:15] <Philip`> or at least designed to be abused by, or something along those lines
- # [01:18] <tH> http://www.active.com/donate/sdkomen06 - I suppose it's nice that it's a list of charity donations making the page code 3MB :)
- # [01:20] <tH> the other one, with loads of </nobr> and <font>, is a spammed-up forum page, I didn't let it finish loading
- # [01:21] <Philip`> Nice, they use <SPACER> too
- # [01:42] * Joins: Lachy (chatzilla@203.214.140.60)
- # [01:55] * Quits: tH (Rob@87.102.14.144) (Quit: ChatZilla 0.9.78.1-rdmsoft [XULRunner 1.8.0.9/2006120508])
- # [02:14] * Quits: Sander (svl@86.87.68.167) (Quit: And back he spurred like a madman, shrieking a curse to the sky.)
- # [03:37] * Quits: beowulf (carisenda@91.84.50.132) (Ping timeout)
- # [03:40] * Joins: beowulf (carisenda@91.84.50.132)
- # [04:13] * Quits: zcorpan (zcorpan@84.216.41.90) (Ping timeout)
- # [04:22] * Joins: mjs (mjs@67.41.157.35)
- # [04:51] * Joins: MikeSmith (MikeSmith@mcclure.w3.org)
- # [05:17] * Quits: MikeSmith (MikeSmith@mcclure.w3.org) (Ping timeout)
- # [05:29] * Quits: mjs (mjs@67.41.157.35) (Ping timeout)
- # [05:35] * Joins: mjs (mjs@67.41.195.89)
- # [06:06] * Quits: mjs (mjs@67.41.195.89) (Quit: mjs)
- # [06:24] * Quits: Lachy (chatzilla@203.214.140.60) (Quit: ChatZilla 0.9.78.1 [Firefox 2.0.0.4/2007051502])
- # [07:53] * Quits: Preston (chatzilla@70.181.68.187) (Quit: ChatZilla 0.9.78.1 [Firefox 2.0.0.5/2007071812])
- # [08:28] * Quits: gavin (gavin@74.103.208.221) (Ping timeout)
- # [08:33] * Joins: gavin (gavin@74.103.208.221)
- # [09:53] * Joins: mjs (mjs@67.41.201.40)
- # [10:01] * Quits: mjs (mjs@67.41.201.40) (Quit: mjs)
- # [10:01] * Joins: mjs (mjs@67.41.201.40)
- # [10:06] * Joins: MikeSmith (MikeSmith@mcclure.w3.org)
- # [10:10] * Quits: mjs (mjs@67.41.201.40) (Ping timeout)
- # [10:18] * Joins: mjs (mjs@67.41.149.11)
- # [10:30] * Joins: ROBOd (robod@86.34.246.154)
- # [10:35] * Quits: gavin (gavin@74.103.208.221) (Ping timeout)
- # [10:40] * Joins: gavin (gavin@74.103.208.221)
- # [10:45] * Quits: MikeSmith (MikeSmith@mcclure.w3.org) (Ping timeout)
- # [10:51] * Joins: MikeSmith (MikeSmith@mcclure.w3.org)
- # [10:54] * Joins: Sander (svl@86.87.68.167)
- # [11:01] * Quits: mjs (mjs@67.41.149.11) (Ping timeout)
- # [11:09] * Joins: mjs (mjs@67.41.147.136)
- # [11:16] * Quits: mjs (mjs@67.41.147.136) (Ping timeout)
- # [11:24] * Joins: mjs (mjs@67.40.154.212)
- # [11:38] * Quits: mjs (mjs@67.40.154.212) (Ping timeout)
- # [12:20] * Joins: mjs (mjs@70.56.38.192)
- # [12:30] * Quits: mjs (mjs@70.56.38.192) (Quit: mjs)
- # [12:31] * Joins: zcorpan (zcorpan@84.216.41.90)
- # [12:32] <zcorpan> anne: updated dom2string.js. and the data/ seems to be outdated
- # [12:41] <anne> I believe I updated both already
- # [12:43] <zcorpan> ah. indeed
- # [12:50] * Quits: MikeSmith (MikeSmith@mcclure.w3.org) (Quit: Less talk, more pimp walk.)
- # [13:02] <anne> http://www.w3.org/html/wg/html5/ should prolly say "Editors' draft" ...
- # [13:19] * Quits: sbuluf (dso@200.49.140.186) (Ping timeout)
- # [13:35] * Joins: tH_ (Rob@87.102.14.144)
- # [13:35] * tH_ is now known as tH
- # [14:02] * Quits: gavin (gavin@74.103.208.221) (Ping timeout)
- # [14:07] * Joins: gavin (gavin@74.103.208.221)
- # [15:23] * Joins: briansuda (briansuda@85.220.95.76)
- # [15:59] * Joins: MikeSmith (MikeSmith@mcclure.w3.org)
- # [15:59] * Quits: schepers (schepers@128.30.52.30) (Client exited)
- # [16:00] * Joins: schepers (schepers@128.30.52.30)
- # [16:09] * Quits: gavin (gavin@74.103.208.221) (Ping timeout)
- # [16:14] * Joins: gavin (gavin@74.103.208.221)
- # [16:25] * Joins: tH_ (Rob@87.102.85.210)
- # [16:25] * Quits: tH (Rob@87.102.14.144) (Connection reset by peer)
- # [16:26] * tH_ is now known as tH
- # [16:40] * Quits: schepers (schepers@128.30.52.30) (Quit: Trillian (http://www.ceruleanstudios.com)
- # [16:56] * Joins: myakura (myakura@58.88.37.26)
- # [17:10] * Quits: tH (Rob@87.102.85.210) (Ping timeout)
- # [17:38] * Joins: tH (Rob@87.102.85.210)
- # [18:16] * Quits: ROBOd (robod@86.34.246.154) (Quit: http://www.robodesign.ro )
- # [18:16] * Quits: gavin (gavin@74.103.208.221) (Ping timeout)
- # [18:20] * Joins: ROBOd (robod@86.34.246.154)
- # [18:21] * Joins: gavin (gavin@74.103.208.221)
- # [18:27] * Quits: myakura (myakura@58.88.37.26) (Quit: Leaving...)
- # [18:43] * Quits: briansuda (briansuda@85.220.95.76) (Quit: briansuda)
- # [19:47] * Joins: dbaron (dbaron@71.198.189.81)
- # [20:23] * Quits: gavin (gavin@74.103.208.221) (Ping timeout)
- # [20:28] * Joins: gavin (gavin@74.103.208.221)
- # [21:59] * Quits: MikeSmith (MikeSmith@mcclure.w3.org) (Ping timeout)
- # [22:01] * Quits: tH (Rob@87.102.85.210) (Ping timeout)
- # [22:09] * Joins: tH (Rob@87.102.85.210)
- # [22:12] <hsivonen> http://validator.nu/ http://html5.validator.nu/ and http://parsetree.validator.nu/ are up
- # [22:19] <Philip`> hsivonen: Is it recommended to not use http://validator.nu/html5/?doc=... (and to use html5.v.n/?doc= instead)?
- # [22:20] <hsivonen> Philip`: yes (but I added a redirect for you :-)
- # [22:21] <Philip`> I probably shouldn't link to unadvertised interfaces :-)
- # [22:22] <Philip`> I've fixed the things at e.g. http://canvex.lazyilluminati.com/survey/2007-07-17/analyse.cgi/tag/v%253Astroke now
- # [22:22] <hsivonen> parsetree.v.n is not an end user site. it is for WG members to use so they can get another opinion besides html5lib with a different set of bugs
- # [22:22] <hsivonen> Philip`: thanks. I guess I could now remove the redirect. :-)
- # [22:24] <Philip`> (The validator links on my page aren't incredibly useful, but at least I found it handy when looking at sites with duplicate attributes and wanting more informative error reports, which is why I've left them in)
- # [22:26] <Philip`> Would it be possible to pass bits of HTML directly to parsetree.validator.nu, instead of having to upload it somewhere first?
- # [22:27] <hsivonen> Philip`: sometime in the future, yes. not at the moment
- # [22:27] <hsivonen> Philip`: this would require refactoring the Content-Type handling code out of the entity resolver and making the servlet build an InputSource from the POST data using that same code
- # [22:28] <hsivonen> not a big deal. just not done yet
- # [22:31] * Quits: gavin (gavin@74.103.208.221) (Ping timeout)
- # [22:33] <hsivonen> parsetree.v.n uses the new parser in a tree building mode
- # [22:33] <hsivonen> the validation facets use it in a streaming mode
- # [22:34] * Joins: mjs (mjs@67.41.147.72)
- # [22:34] * Quits: ROBOd (robod@86.34.246.154) (Quit: http://www.robodesign.ro )
- # [22:35] * Quits: dbaron (dbaron@71.198.189.81) (Quit: 8403864 bytes have been tenured, next gc will be global.)
- # [22:36] <hsivonen> judging from http://html5.validator.nu/?doc=http%3A%2F%2Fwww%2Ebluebirds%2Ecom%2Ept , Jing/oNVDL works fine even if SAX reports non-NCName names to it
- # [22:36] * Joins: gavin (gavin@74.103.208.221)
- # [22:39] <zcorpan> Error: Element v:shapetype from namespace http://www.w3.org/1999/xhtml not allowed in this context. Line 95, column 99 in resource http://www.bluebirds.com.pt/
- # [22:39] <zcorpan> Error: Attribute id not allowed at this point; ignored. Line 95, column 99 in resource http://www.bluebirds.com.pt/
- # [22:39] <zcorpan> the id attribute is always allowed on elements in the html namespace
- # [22:40] <zcorpan> "The following attributes are common to and may be specified on all HTML elements (even those not defined in this specification):"
- # [22:40] <hsivonen> zcorpan: after the first error, anything goes. :-)
- # [22:40] <Philip`> Why does http://html5.validator.nu/?doc=http%3A%2F%2Fyahoo.com complain about "Attribute href not allowed at this point" on what appears to be just an <a href="...">?
- # [22:40] <hsivonen> zcorpan: yeah, but living up to the letter of the spec when one error is triggered would have a great cost of not being able to use certain off-the-shelf tools
- # [22:42] <zcorpan> ok
- # [22:42] <hsivonen> Philip`: I have no idea. Chances are you've found a bug.
- # [22:43] <Philip`> http://www.microsoft.com/en/us/default.aspx is a single line of HTML - I guess they're not fans of letting people use 'view source'
- # [22:45] <hsivonen> Philip`: http://validator.nu/?doc=http%3A%2F%2Fyahoo.com&schema=http%3A%2F%2Fhsivonen.iki.fi%2Fcheckers%2Fdebug%2F+http%3A%2F%2Fsyntax.whattf.org%2Frelaxng%2Fhtml5full.rnc+http%3A%2F%2Fsyntax.whattf.org%2Frelaxng%2Fassertions.sch+http%3A%2F%2Fhsivonen.iki.fi%2Fcheckers%2Fall%2F&parser=html5
- # [22:45] <hsivonen> Philip`: the parser seems to be reporting sane things
- # [22:47] <Philip`> When it says "Element p ... requires significant inline content but did not have any.", is it easy to make the line number point to the insignificantly-contented element's start tag rather than to the (sometimes implied) end tag?
- # [22:49] <hsivonen> Philip`: harder but relatively doable
- # [22:49] <hsivonen> (need to copy the start Locator on stack and add an error method that takes a Locator. that would be it)
- # [22:50] <hsivonen> but I have to pack a bag and get some sleep now
- # [22:50] <Philip`> Is it intentional that <pre>if (1 < 2) ...</pre> is a parse error in HTML5, but valid in HTML4?
- # [22:51] <zcorpan> perhaps < should get the same treatment as & ?
- # [22:52] <Philip`> "<p>If the pointer points to a 0x3C (ASCII '<') byte" has the same situation
- # [22:52] <Philip`> because the < is interpreted as a tag opener
- # [22:52] <hsivonen> will either of you email the list about this?
- # [22:53] <hsivonen> (except pandering to poor authorship, not-the-pro-way replies...)
- # [22:53] <Philip`> Semicolon in "<td>EURO SIGN ('€')" has the same situation too
- # [22:53] <Philip`> The HTML5 spec isn't very good at being valid HTML5 :-)
- # [22:56] <Philip`> (The < and ; and <p><p> are the only issues I can see, though)
- # [22:57] <zcorpan> <p><p>?
- # [22:58] <Philip`> If someone understands the situation and how/whether to fix these things, it'd probably be best if they could post something about it :-)
- # [22:58] <Philip`> zcorpan: More like "<p> <p>"
- # [22:58] <Philip`> which is used in the HTML5 spec
- # [22:58] <Philip`> (but isn't valid HTML5)
- # [22:59] <Philip`> so presumably the spec should be changed so that it's valid
- # [22:59] <zcorpan> where is that used?
- # [22:59] <Philip`> (although there are two totally different ways in which "the spec should be changed" can be interpreted...)
- # [23:00] <hsivonen> Philip`: yeah, instead of making the significant inline checker be smarter about Locator, I'd rather see Hixie zap the entire significant inline concept
- # [23:00] <Philip`> The end tags of the empty <p>s are on lines 13734, 13746, 13761, 13773
- # [23:00] <Philip`> of http://www.whatwg.org/specs/web-apps/current-work/
- # [23:02] <Philip`> or they're in http://html5.validator.nu/?doc=http://www.whatwg.org/specs/web-apps/current-work/multipage/section-video.html
- # [23:02] <zcorpan> ok
- # [23:06] <hsivonen> Philip`: my initial suspicion is that the href error is due to the <font> parent that is unsupported
- # [23:07] <hsivonen> looks like hacking Jing to silence attribute errors on forbidden elements is something I need to do at some point
- # [23:07] * Quits: mjs (mjs@67.41.147.72) (Ping timeout)
- # [23:11] <hsivonen> btw, XHTML is no longer fatal:
- # [23:11] <hsivonen> http://validator.nu/?doc=http%3A%2F%2Fzeldman.com%2F
- # [23:11] <hsivonen> (and it appears that I have forgotten to suppress xmlns)
- # [23:15] * Joins: mjs (mjs@67.41.138.134)
- # [23:26] * Quits: mjs (mjs@67.41.138.134) (Ping timeout)
- # [23:34] * Joins: mjs (mjs@67.41.194.94)
- # [23:34] <Lionheart> BTW, anne, just got around to reading your "Why video?" blog post. I thought your argument for <video> over <object> was cogent.
- # [23:37] * Quits: xover (xover@193.157.66.5) (Ping timeout)
- # [23:57] * Quits: mjs (mjs@67.41.194.94) (Ping timeout)
- # Session Close: Mon Jul 23 00:00:00 2007
The end :)