| 06:53 | <hsivonen> | jgraham: btw, it is now official that my next work items after getting the conformance checker source code better out there are proper Java implementations of the character encoding sniffing algorithm and the tokenizer |
| 06:54 | <hsivonen> | I intend to write and independent implementation to spec instead of porting html5lib |
| 06:54 | <hsivonen> | should be good both for getting the Javaness right and for getting more review of the spec |
| 06:55 | <hsivonen> | s/write and/write an/ |
| 06:55 | <MikeSmith> | hsivonen - how much work/time you estimate that will take? |
| 06:59 | <hsivonen> | MikeSmith: based on talking with jgraham and annevk and based on comparing # of lines of code against previous code, the expectation is that the first unpolished and untested version takes a week (after I'm done sharing the existing source code properly with build scripts) and testing and productization-quality polishing and testing is going to take a couple of weeks more. |
| 07:01 | <hsivonen> | basically, I expect the "get something running" take less time than all the productization polishing after it |
| 07:01 | <hsivonen> | although it isn't clear yet how well the those parts can be polished before the tree-builders exist |
| 07:02 | <hsivonen> | anyway, the goal is to produce a reusable library--not just to write the minimum that would allow the conformance checker to run |
| 07:18 | <MikeSmith> | hsivonen - are there others picking up your code so far? (that you're aware of) |
| 07:19 | <hsivonen> | MikeSmith: Jirka Kosek and Petr Nálevka will probably pick up the table integrity checker real soon now |
| 07:19 | <hsivonen> | MikeSmith: they are also interested in the parser |
| 07:19 | <hsivonen> | I'd advice against using my current parser, though |
| 07:20 | <hsivonen> | but once I'm done with writing a productized HTML5 parsing library with 4 tree builders, I'm all for others using it :-) |
| 07:20 | <MikeSmith> | yeah, I know you planning to update your parser to matech the spec |
| 09:42 | <hsivonen> | now that Vlad is suggesting dashing, it would be interesting to hear from the canvas implementors at Opera and Apple if they'd be ok with adding dashes |
| 09:42 | <Hixie> | adding this to canvas when canvas is so poorly implemented at this stage would be stupid |
| 09:42 | <Hixie> | every browser is swimming in bugs and unimplemented features already |
| 09:43 | <Hixie> | adding more stuff will just make it even less likely that we'll reach the sort of quality point we need for exiting CR |
| 09:48 | <virtuelv> | Hixie: agreed, canvas is suffering from poor interoperability |
| 09:53 | <hsivonen> | which is interesting since in theory, canvas should be the area of the spec where you could get the most interop per unit of effort (because the back end imaging model is well understood and old) |
| 09:58 | <Hixie> | it's the area of the spec with the most interop, i think |
| 09:58 | <Hixie> | but it would only get worse if we added more features at this stage, imho |
| 09:58 | <Hixie> | i mean, if the vendors think we should, i'll bow to pressure, but it just seems stupid to me |
| 10:02 | <virtuelv> | hsivonen: Some of the interop problems in canvas are a result of different interpretations of spec text, I believe |
| 10:03 | <met_> | so we need some spec how to read the spec? 8-) |
| 10:04 | <virtuelv> | met_: you mean you want formal grammar for a spec language |
| 10:04 | <virtuelv> | good luck |
| 10:04 | <annevk> | the spec text is improving |
| 10:04 | <virtuelv> | indeed it is |
| 10:05 | <hsivonen> | met_: the spec already says that it should be read forwards, backwards and in pieces |
| 10:05 | <met_> | hsivonen: nice 8-) |
| 10:16 | <ROBOd> | good morning to all |
| 10:16 | <ROBOd> | Hixie: ping? |
| 10:23 | <Hixie> | hey |
| 10:27 | <ROBOd> | hey Hixie, i was surprised to read your replies to my very old emails |
| 10:27 | <Hixie> | yeah, i'm going through all the feedback we received |
| 10:28 | <ROBOd> | very good, thanks for the replies |
| 10:41 | <Hixie> | ROBOd: btw in case i have other e-mails from you, i apologise in advance if i send the replies to the wrong address. i don't look at who wrote the e-mail when i reply to them, i just hit "reply all", write the reply, fix the spec if needed, and send the mail. |
| 10:42 | <ROBOd> | Hixie: absolutely no problem |
| 10:42 | <ROBOd> | i understand that |
| 10:43 | <ROBOd> | i just mentioned the email address change, in case others reply |
| 10:43 | <ROBOd> | i can no longer reply from the old email account - it's not subscribed to the ML |
| 10:44 | <hsivonen> | offtopic for the svn users: does svn integration actually work these days in eclipse? |
| 10:44 | annevk | uses SVN but not Eclipse :) |
| 10:45 | <met_> | http://subclipse.tigris.org/ looks so |
| 10:45 | <hsivonen> | *working* Eclipse integration is a must-have for me. I'm considering whether I should put the new html parser code in the schema CVS repo or start a new svn repo |
| 10:46 | <hsivonen> | met_: have you tried subclipse lately? |
| 10:46 | <met_> | no |
| 10:46 | <hsivonen> | my experience of subclipse is from two years ago and back then it was unacceptably b0rked |
| 10:46 | <met_> | looks its from same people like tortoiseSVN which i am using |
| 10:46 | <Hixie> | right, bed time |
| 10:46 | <Hixie> | nn |
| 10:47 | <ROBOd> | good night Hixie :) |
| 10:47 | <hsivonen> | Hixie: g'night |
| 10:47 | met_ | have lunchtime in 5 minutes |
| 10:48 | <ROBOd> | bon appètit met_ |
| 10:49 | <met_> | thanks |
| 11:22 | <Philip`> | The canvas implementations all seem to be fine at just drawing stuff - they do curves and lines and rotated bitmaps and gradients and everything - and it's only in the mapping between the JS API and the drawing bits that has lots of problems |
| 11:24 | <mikeday> | don't they draw stuff in response to JavaScript API calls? |
| 11:25 | <Philip`> | They do, but usually they don't draw the correct stuff |
| 11:25 | <mikeday> | different implementations behave slightly differently? |
| 11:26 | <Philip`> | They often behave totally differently :-) |
| 11:26 | <mikeday> | oh :) |
| 11:26 | <mikeday> | since SQL is now part of the HTML5 spec, why not base the canvas on the PDF/PostScript rendering model? |
| 11:27 | <annevk> | ... |
| 11:27 | <Philip`> | There are simple cases like "moveTo(0, 0); translate(10, 10); lineTo(20, 20); stroke()" that are handled very differently |
| 11:27 | <mikeday> | really? how differently? |
| 11:28 | <Philip`> | Some apply the transformation as points are being added to the path, and some apply it just before the stroke so it affects all the points equally |
| 11:28 | <mikeday> | oh, right |
| 11:29 | <mikeday> | problem with imperative interface I guess |
| 11:29 | <mikeday> | needs a well defined mapping to an SVG DOM, perhaps :) |
| 11:29 | <Philip`> | http://canvex.lazyilluminati.com/tests/tests/results.html shows lots of problems and I haven't even got around to testing radial gradients or patterns or paths, which are particularly bad |
| 11:29 | <Philip`> | The behaviour is well-defined already - it just doesn't match two thirds of the implementations :-) |
| 11:30 | <mikeday> | not really a specification problem, then :) |
| 11:31 | Philip` | should write more tests and submit bug reports |
| 11:32 | <Philip`> | (and maybe write patches if they're not complicated and if everyone else wants to ignore them and add new features instead) |
| 11:33 | hsivonen | wonders what happens in PostScript if you manipulate the CTM between adding path segments |
| 11:33 | <mikeday> | good question |
| 11:34 | <mikeday> | I thought it was illegal |
| 11:34 | <Philip`> | What happens when you do something illegal? |
| 11:35 | <mikeday> | ask ghostscript :) |
| 11:35 | <zcorpan_> | a SWAT team comes to get you |
| 11:36 | <mikeday> | hmm, at least in PDF, once you start constructing a path you can't do anything else until you paint it |
| 11:36 | <hsivonen> | zcorpan_: the Web lacks the SWAT team |
| 11:36 | <mikeday> | if you *do* do anything else the results are not defined by the specification |
| 11:37 | <hsivonen> | zcorpan_: thesis comment noted. thanks. however, if at all feasible, I'm treating the dated files as frozen |
| 11:38 | <Philip`> | Aha, undefined behaviour sounds an excellent strategy to follow ;-) |
| 11:38 | <mikeday> | "Note: A content stream whose operations violate these rules for describing graphics objects can produce unpredictable behavior, even though it may display and print correctly." |
| 11:39 | <mikeday> | that's PDF, but I think it inherited the restriction from PostScript. |
| 11:40 | <hsivonen> | iirc, adobe reader silently accepts some non-conforming files that Ghostscript accepts but complains about |
| 11:40 | <hsivonen> | iirc, there were some semi-prominent Macromedia or Corel legacy products outputting the non-conforming stuff |
| 11:42 | <hsivonen> | but fewer people author PDF by hand, so the problem is not as bad as with HTML/CSS |
| 11:42 | <mikeday> | when it comes to PDF, acrobat compatibility is more important than following the spec |
| 11:42 | <hsivonen> | mikeday: are there cases where you have to violate the spec to be compatible with acrobat? |
| 11:43 | <hsivonen> | on the producer side, that is |
| 11:43 | <mikeday> | not that I recall, |
| 11:43 | <mikeday> | but there are some cases where PDF files that follow the spec will not be accepted |
| 11:44 | <mikeday> | funny how similar the situation is to browsers, actually |
| 11:44 | <mikeday> | there are just fewer user agents in PDF world |
| 11:45 | <hsivonen> | actually, there's also the top gang of four on the PDF side |
| 11:45 | <hsivonen> | (Adobe, Apple, xpdf-based, Ghostscript) |
| 11:47 | <mikeday> | yes |
| 11:47 | <mikeday> | I guess you can map Acrobat -> IE |
| 11:48 | <mikeday> | in that if it doesn't work in Acrobat, you've lost 90% of your userbase |
| 11:49 | <mikeday> | must go |
| 11:50 | <annevk> | Philip`, for that particular case I believe Opera matches the SVG model... |
| 11:50 | <annevk> | Philip`, and I think other vendors agreed it was more logical but they haven't updated their impl so far... |
| 11:53 | <Philip`> | annevk: Okay - I think the spec text matches Opera but it was waiting for implementor feedback when I last heard |
| 11:54 | <annevk> | yeah, Firefox and Safari still need to try to fix it... |
| 11:58 | <hsivonen> | annevk: http://tc.labs.opera.com/robots.txt isn't particularly friendly to people who try to locate the test suite |
| 11:58 | <hsivonen> | also, it suggest you don't want me to hit it with recursive wget |
| 11:58 | <hsivonen> | annevk: what's the right way to obtain a local copy of the test suite? |
| 12:06 | <annevk> | I suppose I could lift some restrictions |
| 12:07 | <annevk> | done |
| 12:08 | <hsivonen> | annevk: thanks |
| 12:26 | <annevk> | hsivonen, that's how text/plain loading is defined in HTML5... |
| 12:26 | <annevk> | using <plaintext> |
| 12:30 | <Philip`> | Does anyone happen to know how Safari probably does canvas shadows? It looks like it multiplies the shadow colour by the alpha value at each pixel, then does some blur (Gaussian with cutoff or something?) and composites stuff - maybe it's just the same as the PS / PDF definition, and hopefully that definition actually defines it... |
| 12:30 | <annevk> | heh |
| 12:31 | <Philip`> | Oh, PDF doesn't have shadows |
| 12:32 | <Philip`> | (and PostScript doesn't have shadows either) |
| 12:35 | <annevk> | maybe someone can make a <canvas> Acid test |
| 12:35 | <Philip`> | (and http://developer.apple.com/documentation/GraphicsImaging/Conceptual/drawingwithquartz2d/dq_shadows/chapter_8_section_1.html is uselessly vague) |
| 12:36 | <annevk> | http://developer.apple.com/documentation/GraphicsImaging/Conceptual/drawingwithquartz2d/dq_shadows/chapter_8_section_2.html#//apple_ref/doc/uid/TP30001066-CH208-DontLinkElementID_7 seems to have details |
| 12:36 | <annevk> | (linked from that page) |
| 12:37 | <Philip`> | It only has about one detail (that the default alpha is 1/3, which doesn't matter for canvas because you always override the default alpha) :-( |
| 12:37 | <Philip`> | Hmm, about colour spaces, does everyone implement canvas with sRGB? And should the spec state that? |
| 12:38 | <annevk> | doesn't CSS say something about that? |
| 12:38 | <annevk> | it should be the same as what CSS uses, anyway |
| 12:38 | <Philip`> | CSS3 Color says it's all in sRGB |
| 12:39 | <Philip`> | though I think canvas still needs to specify the colour space, so that e.g. linear interpolation along gradients makes sense |
| 12:39 | <hsivonen> | annevk: yes, but if you generate <plaintext> on the server, the sniffing does not kick in |
| 12:41 | <hsivonen> | Philip`: the sane way to spec it is that the canvas color space should be the same color space as what the UA uses for CSS colors in the absence of an explicit profile |
| 12:42 | <hsivonen> | Philip`: CSS says sRGB, but in reality browsers use the color space of the OS |
| 12:44 | <zcorpan_> | hsivonen: if you say text/plain; charset=utf-8, the sniffing doesn't kick in either (or it shouldn't) |
| 12:45 | <zcorpan_> | iirc |
| 12:45 | Philip` | wonders if trying to draw gamma-corrected PNGs onto a canvas is likely to involve far too much pain and is best avoided |
| 12:45 | <hsivonen> | zcorpan_: even in real-word IE? |
| 12:45 | <hsivonen> | Philip`: I'd expect lots of bogosity and bugs in that area |
| 12:46 | <zcorpan_> | hsivonen: not sure |
| 12:46 | <Philip`> | http://trac.webkit.org/projects/webkit/browser/trunk/WebCore/html/HTMLCanvasElement.cpp#L238 - ah, that's using "DeviceRGB" |
| 12:47 | <hsivonen> | Philip`: yeah, but DeviceRGB isn't really the *device* but the OS. |
| 12:47 | <Philip`> | (Oh, did someone add Qt support to WebKit <canvas> recently?) |
| 12:47 | <hsivonen> | there's a transfer function between DeviceRGB and the real device |
| 12:48 | <Philip`> | (I guess that'll provide a whole 'nother set of bugs...) |
| 12:49 | <hsivonen> | Philip`: I wouldn't expect bug opportunities there. only miscalibratin opportunities |
| 12:49 | <hsivonen> | AFAIK, as far as apps are concerned, DeviceRGB is the "device" |
| 12:49 | <krijnh> | zcorpan_: IE does sniff when using Content-Type: text/plain;charset=utf-8 |
| 12:50 | <hsivonen> | but what it really means is configurable system-wide |
| 12:50 | <zcorpan_> | krijnh: ok |
| 12:50 | <krijnh> | IE7 as well |
| 12:50 | <Philip`> | Oh, sorry, I was thinking of bugs in WebKit-Qt (since that has totally new rendering code) |
| 12:50 | <Philip`> | Colour spaces seem confusing enough that I'll probably ignore them for now and be happy :-) |
| 12:50 | <zcorpan_> | krijnh: in ie7, there's a security setting to disable content type sniffing |
| 12:52 | hsivonen | wonders if apps can circumvent DeviceRGB via OpenGL on Mac OS X |
| 12:52 | <krijnh> | zcorpan_: What's it called? |
| 12:53 | <zcorpan> | krijnh: "Open files based on content, not file extension" |
| 12:53 | <krijnh> | Open files based on content, not file.. |
| 12:53 | <krijnh> | Indeed :) |
| 12:53 | <krijnh> | Wow, that's a crappy label |
| 12:53 | <zcorpan> | yeah |
| 12:54 | <krijnh> | It has nothing to do with the extension |
| 12:54 | <krijnh> | And I don't think it checks on the extension |
| 12:54 | <krijnh> | Er, sniffs |
| 12:55 | <hsivonen> | probably a case of the UI folks not understanding the implementation |
| 12:55 | <hsivonen> | on Mac OS X "Rich Text" is translated to "RTF" in Finnish in Mail.app even though "Rich Text" in mail is HTML |
| 12:55 | <krijnh> | Rhat the fuck |
| 12:57 | <hsivonen> | krijnh: was that in response to what I said? |
| 12:57 | <krijnh> | Nah, I say rhat the fuck all the time |
| 12:58 | <krijnh> | I'll be silent again now ;) |
| 13:00 | hsivonen | is slow. only gets it now |
| 13:07 | <zcorpan> | who has access to the whatwg blog source? ".navigation div { position:relative; }" needs to be added to the style sheet for opera |
| 13:12 | <zcorpan> | Lachy: yt? |
| 13:17 | <Lachy> | yo |
| 13:17 | <zcorpan> | Lachy: see above :) |
| 13:17 | <Lachy> | yeah, I just noticed |
| 13:18 | <zcorpan> | perhaps i should also try to make a test case out of it, if it's actually a bug in opera |
| 13:18 | <Lachy> | it probably is a bug |
| 13:20 | <Lachy> | I don't see what the bug is |
| 13:20 | <zcorpan> | it's not always reproducable |
| 13:21 | <Lachy> | how can I reproduce it at all? I've tried reloading several times and don't see it |
| 13:22 | <MikeSmith> | hsivonen - you aware of any Linux PDF viewers that aren't based on xpdf? (just wondering if there are any) |
| 13:23 | <zcorpan> | Lachy: dunno. sometimes the links aren't clickable and then if you scroll down and up again the links disappear |
| 13:23 | <Lachy> | which version of Opera? |
| 13:23 | <Lachy> | I have 9.20 |
| 13:24 | <zcorpan> | 9.21 |
| 13:24 | <Lachy> | ok. updated stylesheet is published now |
| 13:25 | <zcorpan> | thanks |
| 13:35 | <zcorpan> | http://simon.html5.org/test/opera/unselectable-text.htm |
| 13:47 | <zcorpan> | submitted bug #266507 |
| 13:51 | Lachy | is about to publish a rather controversial blog post about the HTMLWG |
| 13:54 | annevk | will go to reboot for a day! |
| 13:55 | <met_> | Lachy: about what? htmlwg dead? 8-) |
| 13:56 | <Lachy> | it's about the issues with the HTMLWG, giving an overview of all the major debates that have gone on |
| 13:56 | <Lachy> | it was originally an email to Molly that she asked me to write, and she requrested that I make it public |
| 14:01 | <Philip`> | Conclusion from testing <canvas> in WebKit-Qt: it's currently really quite broken (e.g. it doesn't even seem to do rotation correctly), and also I can't copy-and-paste the test results and then it had a segmentation fault and lost the data :-( |
| 14:03 | <annevk> | what's WebKit-Qt? |
| 14:03 | <Philip`> | It's just WebKit, but compiled with Qt :-) |
| 14:03 | <Philip`> | (I don't know if it has an official name) |
| 14:03 | <Philip`> | and running on Linux |
| 14:04 | <met_> | sounds like khtml 8-) |
| 14:05 | <Lachy> | http://lachy.id.au/log/2007/05/htmlwg |
| 14:06 | <Philip`> | It sounds like there are plans to make it possible to use WebKit as the HTML component in KDE4, as an alternative to the normal KHTML |
| 14:06 | <annevk> | btw, the easiest way to make public record of an e-mail is to cc www-archive⊙wo |
| 14:07 | <Lachy> | I thought about that, but I decided marking it up and blogging it was more appropriate in this case |
| 14:07 | <MikeSmith> | annevk - |
| 14:07 | <MikeSmith> | http://dot.kde.org/1152645965/ |
| 14:07 | <MikeSmith> | http://labs.trolltech.com/blogs/category/labs/internet/webkit/ |
| 14:08 | <MikeSmith> | George Staikos and Zack Rusin |
| 14:09 | <MikeSmith> | and Lars Knoll |
| 14:11 | <annevk> | looks interesting |
| 14:12 | <annevk> | http://www.reboot.dk/artefact-2493-en.html |
| 14:14 | <annevk> | btw |
| 14:14 | <annevk> | how about |
| 14:15 | <annevk> | r = new XMLHttpRequest(uri); r.onload = ...; r.onerror = ... |
| 14:15 | <annevk> | to do a simple asynchronous GET request without hassle |
| 14:16 | <MikeSmith> | I guess that when completed the WebKit Qt port can run within Trolltech's Qtopia (what used to be called Qt/Embedded) on mobile devices |
| 14:18 | <jonbarnett> | anne: would/coult any of those events have the XMLRequestObject as one of the *target properties of the event object passed to the handler? |
| 14:18 | <annevk> | this should refere to XHR in theory |
| 14:19 | <annevk> | and does in Opera |
| 14:19 | <annevk> | and I think it does in WebKit builds too |
| 14:20 | <jonbarnett> | the draft doesn't say so. which property (currentTarget?) |
| 14:20 | <annevk> | it's all implied by making XMLHttpRequest implement EventTarget |
| 14:20 | <jonbarnett> | got it |
| 14:41 | <jonbarnett> | anne: what version of Opera does so? |
| 14:41 | <annevk> | supports what, exactly? |
| 14:42 | <jonbarnett> | passes something to the event handler. I'm testing with 0 |
| 14:42 | <jonbarnett> | 9 |
| 14:42 | <annevk> | using request.onreadystatechange = function() { alert(this.readyState) } |
| 14:42 | <annevk> | should work |
| 14:43 | <jonbarnett> | I assumed that function(e) { alert(e); } should give something |
| 14:44 | <annevk> | I don't think we have made it an EventTarget yet |
| 15:19 | <hsivonen> | MikeSmith: ggv is Ghostscript-based and Adobe Reader is, of course, using Adobe's impl. But yeah, all the nice ones that are Free Software tend to be xpdf-based. |
| 15:20 | <MikeSmith> | hsivonen - k |
| 16:55 | <hsivonen> | HTML5 conformance checking is now on the XML radar: http://www.oreillynet.com/xml/blog/2007/05/fake_realtime_blog_from_xtech.html |
| 17:04 | <MikeSmith> | hsivonen - interesting -- especially part about error messages |
| 17:05 | <MikeSmith> | funny that he writes "they" in several places |
| 17:05 | <MikeSmith> | maybe he thinks you're not a real person ... |
| 17:05 | <MikeSmith> | but some kind of collective |
| 17:05 | <MikeSmith> | the Henri Sivonen Collective |
| 17:05 | <hasather> | hehe |
| 17:06 | <MikeSmith> | looking forward to seeing what he comes up with as far as trying to implement table-integrity checking solely in Schematron |
| 20:16 | <Jero> | ok wtf, in my HTML5 parser the code "<a><p>X<a>Y</a>Z</p></a>" gets transformed to "<a><p>X</p><a>YZ</a></a>" instead of "<a></a><p><a>X</a><a>Y</a>Z</p>" |
| 20:17 | <Jero> | is that an error in parser the start tag token of the second A element, an error in parsing the adoption agency, or an error in reconstructing the active formatting elements? |
| 20:26 | <Philip`> | Jero: http://hasather.net/html5/parsetree/parsetree.py?source=%3Ca%3E%3Cp%3EX%3C%2Fp%3E%3Ca%3EYZ%3C%2Fa%3E%3C%2Fa%3E says it should give <a><p>X</p></a><a>YZ</a> which is neither of the ones you mentioned |
| 20:27 | <Philip`> | (but I don't understand the parser at all, so I have no idea why yours wouldn't be closing the first <a> at the right point) |
| 20:28 | <gsnedders> | Philip`: that tree looks correct, under my memory of the spec |
| 20:28 | <Jero> | interesting |
| 20:29 | <Jero> | http://html5lib.googlecode.com/svn/trunk/tests/tree-construction/tests1.dat gives my a different output tree (search for "<a><p>X<a>Y</a>Z</p></a>") |
| 20:35 | <Philip`> | Oh, okay - when I run with a (probably newer?) version of html5lib, it does give <a/><p><a>X</a><a>Y</a>Z</p> which matches that test |
| 20:50 | <Jero> | Step 3 of the adoption agency: If there is no furthest block ... pop all the nodes from the bottom of the stack of open elements, from the current node up to the formatting element... |
| 20:52 | <Jero> | because of "up to the formatting element" i didn't include to pop the formatting element from the stack which resulted in not closing the first A element |
| 20:53 | <Jero> | so maybe it should explicitly say "including the formatting element"? |
| 20:54 | <Hixie> | if that would help please do send mail so i remember to fix it |
| 20:54 | <Hixie> | ian⊙hc or whatwh⊙wo |
| 20:54 | <Hixie> | whatwg.org |
| 20:59 | <Jero> | thanks, I will |
| 21:09 | <Jero> | Philip`: my parser now treats "<a><p>X<a>Y</a>Z</p></a>" as "<a></a><p><a>X</a><a>Y</a>Z</p>", instead of "<a><p>X</p></a><a>YZ</a>" like http://hasather.net/html5/parsetree/ gives as output |
| 21:10 | <Jero> | oh, i see you updated the parser, sorry |
| 21:11 | Philip` | wonders who controls that parsetree page and what version of html5lib it's running |
| 21:11 | <Jero> | oh, i thought you were :p |
| 21:13 | <Philip`> | At least it sounds like your one is correct now, if it matches the output from the test cases |
| 22:31 | <annevk> | hasather, ping, see above |
| 22:31 | <annevk> | hasather, your parsetree is out of date |
| 22:54 | <hasather> | done |
| 22:54 | <hasather> | thanks for the notice |
| 22:55 | <Jero> | thanks, your parsetree really is a useful tool for testing |
| 22:56 | <hasather> | not mine |
| 22:56 | <hasather> | James Graham made it |
| 22:56 | <hasather> | just hosting :) |
| 23:10 | <Jero> | well, thanks for hosting then :) |
| 23:46 | <zcorpan> | http://justinsomnia.org/2007/05/ivan-came-in-yesterday-wearing-a-totally-radical-webgeek-tshirt-from-a-list-apart-of-course-i-had/ |
| 23:47 | <zcorpan> | (nowadays you almost don't need to follow links -- you can read the full story by just reading the URL) |
| 23:49 | <Dashiva> | It's only missing "to update it: HTML5 fist" |
| 23:50 | <Hixie> | there's something fundamentally wrong about following links and finding they're only three clicks from my blog |