| 00:00 | <jwalden> | who then? I didn't know ms had fanboys, and I can't think of anyone else who'd presume to speak on this |
| 00:01 | <Philip`> | Everyone has fanboys :-) |
| 00:22 | <Hixie> | annevk3: yt? |
| 00:23 | <Hixie> | annevk3: XHR should define what happens when (a) its owner document stops being fully active (probably the events get buffered up somehow) and (b) what happens when its owner document is reset (probably all buffered events are discarded, and the i/o is silently aborted without any further events firing, but you'll have to check IE to be sure) |
| 00:24 | <Hixie> | (a document is reset when document.open() is called upon it) |
| 00:24 | <Hixie> | annevk3: also you'll have to define that a document has an implied strong reference to the XHR object, otherwise the object will get garbage collected before XHR has completed |
| 00:25 | <Hixie> | maybe instead of owner document for the latter case you should talk in terms of the owner scripting context, since we're adding xhr to workers too |
| 00:41 | <Hixie> | actually (a) might be better handled by saying that you just block while the scripting context of the event listeners is frozen... |
| 00:41 | <Hixie> | and then we can define freezing of scripting contexts... |
| 00:41 | <Hixie> | hmm... |
| 00:41 | <Hixie> | document.open() sucks |
| 01:24 | <MikeSmith> | Hixie: about the term "void element" -- you went with that because "empty element" is ambiguous? That is, it could mean both, "An element that by design must not contain any contents", and "An instance of element that can contain contents but that happens not to contain any." e.g., <script src=foo.js></script> is an empty element |
| 01:24 | <MikeSmith> | was that the reason, or part of the reason, or was it for a different reason(s) altogether? |
| 01:44 | <roc> | there are a lot of MS fanboys. The MS MVP program is institutionalized fanboys |
| 01:49 | <roc> | it's going to be a real nightmare figuring out why (or not) a site is being displayed with the IE8 layout engine |
| 01:56 | <roc> | "As for CSS compliance I have a question: the CSS specification says that illegal property-values should be ignored. Why does IE throw an exception when an illegal property-value is assigned through script (e.g. setting element.style.width='-10px' or something)?" |
| 01:56 | <roc> | urp |
| 01:57 | <MikeSmith> | roc: what was the answer to that? |
| 01:58 | <roc> | there was no answer |
| 02:00 | <Hixie> | MikeSmith: that was the only reason that i recall |
| 02:01 | <MikeSmith> | Hixie: OK, thanks |
| 02:01 | <MikeSmith> | roc: that was from a recent "chat with the MS team" thing? |
| 02:01 | <roc> | comment in the IEblog |
| 02:03 | <MikeSmith> | I se |
| 02:03 | <MikeSmith> | see |
| 02:29 | <MikeSmith> | roc: btw, I made a number of changes to the script I use for generating the "Typical default display properties" sections, to deal with problems you mentioned |
| 02:29 | <roc> | cool |
| 02:29 | <MikeSmith> | see http://www.w3.org/html/wg/markup-spec/#audio-display |
| 02:29 | <MikeSmith> | or better: |
| 02:29 | <MikeSmith> | http://www.w3.org/html/wg/markup-spec/#input-display |
| 02:37 | <roc> | MikeSmith: I don't think 'box-align' is a standard property (yet) |
| 02:38 | <MikeSmith> | roc: yeah, I know. but is that not also true about box-orient and box-margin? |
| 02:38 | <roc> | yes |
| 02:38 | <roc> | I didn't notice those |
| 02:39 | <roc> | appearance and box-sizing are too |
| 02:39 | <ajnewbold> | where are all of the circle-based properties? |
| 02:39 | <ajnewbold> | the circle doesn't get any respect :( |
| 02:40 | <MikeSmith> | roc: yeah, and border-radius, and margin-start, and padding-start, and user-select |
| 02:40 | <roc> | well, those ones are in CSS3 |
| 02:40 | <roc> | although Webkit's syntax might not match the CSS3 syntax for border-radius |
| 02:41 | <MikeSmith> | OK |
| 02:41 | <roc> | user-select might not be in any current CSS3 draft but it has been in the past |
| 02:41 | <roc> | so has box-sizing I suppose |
| 02:41 | <MikeSmith> | OK, well, I guess I can switch those back to e.g., _vendor_-box-orient |
| 02:41 | <roc> | box-align/box-orient/box-margin are the most glaring annoyances because they've never been in any CSS spec |
| 02:42 | <MikeSmith> | OK |
| 02:42 | <roc> | I don't think it makes sense to include non-standard properties at all, to be honest |
| 02:42 | <MikeSmith> | hmm |
| 02:42 | <roc> | most readers aren't going to know that those box properties are XUL related |
| 02:42 | <MikeSmith> | webkit uses them also |
| 02:43 | <roc> | yeah, because it implements parts of XUL layout |
| 02:43 | <MikeSmith> | ah |
| 02:43 | <roc> | these rules are really UA-specific ways of achieving a certain rendering for the replaced element |
| 02:43 | <MikeSmith> | so I'll just have the script drop the box- properties |
| 02:43 | <MikeSmith> | roc: OK, I see |
| 02:44 | <roc> | IMHO we don't need or really want to display "typical UA properties" for replaced elements |
| 02:44 | <MikeSmith> | yeah |
| 02:44 | <roc> | they're implementation details |
| 02:45 | <MikeSmith> | OK, but what about keeping the -start properties ? |
| 02:46 | <MikeSmith> | I could just have it change those to -left, but that would kind of misleading also |
| 02:46 | <Lachy> | MikeSmith, is your HTML markup language draft somewhere in CVS? |
| 02:47 | <Lachy> | I don't see it with the rest of them on dev.w3.org |
| 02:47 | <MikeSmith> | Lachy: it's in www.w3.org CVS |
| 02:47 | <Lachy> | ok |
| 02:48 | <Lachy> | just out of interest, why didn't it get put with the others? |
| 02:49 | <MikeSmith> | because systeam is asking me repeatedly not to put any more "documents" in dev.w3.org. claim that dev.w3.org is intended for hosting code and that we are essentially abusing it |
| 02:50 | <MikeSmith> | problem is that dev.w3.org is one puny machine, not load-balanced or mirrored like www.w3.org is |
| 02:50 | <roc> | IIRC the "start" and "end" variants of properties are something that is being standardized |
| 02:50 | <MikeSmith> | they have had some serious load problems with spiders crawling the HTML5 draft hosted on dev.w3.org |
| 02:50 | <heycam> | MikeSmith, shouldn't the solution to that problem be to allow CVS access to certain areas of www.w3.org by WG members? |
| 02:50 | <roc> | I'd just leave them in |
| 02:50 | <MikeSmith> | heycam: yeah |
| 02:51 | <MikeSmith> | roc: OK, thanks. What do you think about keeping "appearance"? |
| 02:51 | <MikeSmith> | values of appearance are mostly obvious |
| 02:52 | <roc> | all your uses of 'appearance' are for replaced elements |
| 02:52 | <roc> | like I said, I wouldn't include any "default display properties" for replaced elements |
| 02:52 | <Lachy> | ok, so does the sys team have any intention of shifting specs to the other CVS server, or moving dev.w3.org onto the load balanced servers? |
| 02:53 | <MikeSmith> | roc: I guess I'm not clear on on what a replaced element is. seems like 'input, input[type="password"], input[type="search"]' is not selecting a replaced element.. or is it? |
| 02:54 | <roc> | replaced elements are elements like form controls where (at least in principle) CSS does not define the rendering of the element |
| 02:54 | <MikeSmith> | OK |
| 02:54 | <roc> | CSS treats replaced elements as black boxes basically |
| 02:55 | <roc> | it just so happens that in Gecko and Webkit, at least, we implement replaced elements using CSS internally as well |
| 02:55 | <roc> | or mostly so |
| 02:55 | <roc> | but that is an implementation detail |
| 02:56 | <MikeSmith> | I see. Is that true for "user-select" also? |
| 02:56 | <MikeSmith> | Lachy: we can shift specs ourselves to www.w3.org and use ACLs to control access. but it's maintenance PITA, and no good reason to do it except for the dev.w3.org load issues |
| 02:56 | <roc> | user-select is a CSS property, not an element |
| 02:57 | <roc> | you're listing it on 'input' and 'textarea', which are both replaced elements |
| 02:57 | <MikeSmith> | roc: yeah, I know -- I mean the actual property |
| 02:57 | <MikeSmith> | OK |
| 02:57 | <roc> | <audio> and <video> are also replaced elements |
| 02:58 | <MikeSmith> | roc: OK, I'll have it just drop what's there for audio |
| 02:58 | <roc> | listing "typical UA stylesheet rules" for non-replaced elements makes some sense; authors can use that as a guide for how their stylesheets are likely to interact with typical UAs |
| 02:59 | <MikeSmith> | roc: right, that's what my main intent was for including it |
| 02:59 | <roc> | for replaced elements, for which CSS does not define how author styles work, and in practice author styling only works in limited ways, it's less useful |
| 02:59 | <MikeSmith> | yeah, I'm understanding that now |
| 03:17 | <billyjackass> | roc: updated with all the replaced-elements stuff dropped - |
| 03:17 | <billyjackass> | http://www.w3.org/html/wg/markup-spec/ |
| 03:17 | <roc> | cool |
| 03:18 | <roc> | you still have <select>, which is a replaced element |
| 03:18 | <billyjackass> | well, not all of it |
| 03:18 | <billyjackass> | yeah |
| 03:18 | <roc> | ditto option and optgroup |
| 03:19 | <billyjackass> | yep yep |
| 03:19 | <billyjackass> | .me goes back to tweak more |
| 03:19 | <roc> | and input |
| 03:20 | <roc> | and textarea |
| 03:20 | <roc> | hmmmm |
| 03:20 | roc | checks his cache |
| 03:20 | <billyjackass> | button also, right? |
| 03:37 | <roc> | yeah |
| 03:38 | <joshworm> | malware? |
| 03:38 | <malware> | joshworm: malware = MikeSmith |
| 03:38 | <joshworm> | malware: I guessed that much |
| 03:38 | malware | is having network issues, IRC sessions getting hung open |
| 03:39 | <joshworm> | I was just wondering about the peculiar choice of nickname |
| 03:39 | <malware> | old handle I used when I was working at Opera |
| 03:40 | jcranmer | shouldn't accuse too much... |
| 03:41 | <malware> | I think madrobby gots the best handle of all |
| 03:41 | <jcranmer> | malware not found |
| 03:42 | <malware> | malware keeps people on their toes |
| 04:51 | <jwalden> | !summon othermaciej |
| 08:03 | <hsivonen> | "JavaScript and its more sophisticated cousin Ajax" |
| 08:04 | <hsivonen> | source: http://news.cnet.com/8301-17939_109-10113196-2.html |
| 08:09 | <MikeSmith> | bastard stepchild |
| 08:14 | <yecril71> | How an URL gets decoded is defined in the RFC, there is no need to repeat it for HTML. |
| 08:15 | <yecril71> | (Especially Wikipedia has various funny marks in fragment identifiers, so it uses this feature extensively) |
| 08:16 | <yecril71> | id="foo%20bar" stands for just that, no decoding. |
| 08:17 | <yecril71> | You can say id="foo bar" if that does not suit your needs ;-) |
| 08:18 | <yecril71> | That will not validate of course. |
| 08:18 | <yecril71> | But neither does %20. |
| 08:19 | <yecril71> | The authors are expected to double check their pages, not to rely on the user agent to fix their nonsense each time. |
| 08:20 | <hsivonen> | Sun hasn't done a great job in explaining how much JavaFX is just libraries and how much it's runtime environment features you couldn't do as libraries |
| 08:20 | <yecril71> | Of course, there are exceptions, but a mismatched identifier is not worth fixing IMHO. |
| 08:21 | <yecril71> | The reader can find the interesting fragment herself. |
| 08:24 | <hsivonen> | The first Java Web Start-based demo I tried is so slow it isn't even funny |
| 08:24 | <hsivonen> | so slow to start that is |
| 08:25 | <yecril71> | Content of a BR element cannot be loaded from text but it can be constructed from script. |
| 08:25 | <yecril71> | In that case, it should be ignored. |
| 08:31 | <hsivonen> | So Sun now ships VP6 video decoder, too. |
| 09:11 | <nessy> | hsivonen: where did you get that news from? |
| 09:13 | <hsivonen> | nessy: Slashdot, On2's site and Wikipedia |
| 09:13 | <nessy> | thx |
| 09:19 | <jwalden> | hsivonen: did you see "Firefox is owned by Google, at this point." in that article? |
| 09:19 | <hsivonen> | jwalden: I did |
| 09:20 | <jwalden> | I suppose I shouldn't expect more from a renewed attempt to push a technology that was basically DOA, but I do |
| 09:20 | <maikmerten> | the JavaFX video demonstration plays rather horribly here |
| 09:20 | <maikmerten> | (and it won't resize... great) |
| 09:22 | <maikmerten> | and right, they licensed VP6 (On2 has a Java decoder for it) - not sure what their choice for audio is, but I'd assume MP3 perhaps to be compatible with the usual FLV files |
| 09:24 | <hsivonen> | maikmerten: I think a FAQ said MP3 |
| 09:25 | <maikmerten> | ah |
| 09:25 | <hsivonen> | they aren't naming their files .flv, though |
| 09:25 | hsivonen | already closed the window with the new filename extension |
| 09:25 | <maikmerten> | def mediaUrl:String ="http://capra.sfbay.sun.com/~jm158417/javafx_videos/big_buck_bunny_512x288_h264.flv";; |
| 09:25 | <hsivonen> | the first demo I tried was underwhelming |
| 09:25 | <maikmerten> | http://javafx.com/samples/SimpleVideoPlayer/index.html |
| 09:26 | <maikmerten> | yeah, it looked not exactly that nice and wasn't fluid here |
| 09:26 | <hsivonen> | It took around 10 minutes for Java Web Start to tell me that it tried to load and failed |
| 09:26 | <maikmerten> | oh, took only like 15 seconds here |
| 09:26 | <maikmerten> | their site was basically DDOSed by the JavaFX announcement |
| 09:28 | <maikmerten> | ooooh, *now* I see why that video wasn't fluid |
| 09:28 | <maikmerten> | the *encoded file* has dropped frames everywhere |
| 09:29 | <maikmerten> | yup, the audio track is 22050 Hz stereo at 64 kbit/s MP3 |
| 09:30 | <maikmerten> | why would they *showcase* such a crappy encoding (not so much the audio quality, but the video) |
| 09:39 | <annevk3> | Hixie, meh |
| 09:39 | <annevk3> | Hixie, e-mail? |
| 09:42 | <annevk3> | oh you did |
| 09:42 | <annevk3> | meh |
| 09:55 | <zcorpan__> | now i've seen at least two people think that <section><h1>foo</h1><section>bar</section></section> is the correct way |
| 09:55 | <zcorpan__> | maybe the spec should call it out explicitly |
| 09:55 | <Lachy> | zcorpan__, link? |
| 09:56 | <Lachy> | oh, nevermind. that's what that guy asked about on help@whatwg |
| 09:56 | <Lachy> | I'll make sure I mention that in the authoring guide |
| 09:57 | Philip` | wonders what's wrong with that markup |
| 09:58 | <zcorpan> | Philip`: 'bar' isn't intended to be a subsection |
| 09:58 | <Philip`> | Ah |
| 09:59 | <Hixie> | zcorpan: yeah... not sure what to do about that. send me mail with advice? :-) |
| 09:59 | <zcorpan> | Hixie: Lachy's response on help⊙wo seemed like something that could be converted to examples in the spec |
| 10:01 | <BenMillard> | sectioning elements make sectioning more trouble than it's worth, imho, but I haven't studied this in detail yet |
| 10:02 | <Lachy> | BenMillard, it can make things a little more verbose than just using numbered headings |
| 10:04 | <zcorpan> | i'd only use them for wrapper divs on a page, not within an article |
| 10:04 | <BenMillard> | replacing <div> with <section> throughout a page because it's "more semantic" is a permathread waiting to happen! :P |
| 10:04 | <Philip`> | Maybe it would be worth telling people that they don't actually need to use <section> |
| 10:05 | <Philip`> | so they shouldn't try to mark up every single 'section' (in some very loose definition) of their page, because there's no point |
| 10:06 | <zcorpan> | we need an outline tool so authors can see the result of their markup |
| 10:06 | <krijn> | BenMillard: and some <div>'s would become <article> because that's even more semantic! |
| 10:07 | <Philip`> | And spans become <section style="display:inline">! |
| 10:07 | <Lachy> | it would be better if its use could be optimised so you can replace sequential sections like this: |
| 10:07 | <Lachy> | <section><h1/><p>...</section><section><h1/><p>...</section> |
| 10:07 | <Lachy> | with: |
| 10:07 | <Lachy> | <section><h1/><p>... <h1/><p>...</section> |
| 10:07 | <Philip`> | With these changes, we're half way to the Semantic Web already |
| 10:08 | <Lachy> | actually, I think that already works |
| 10:09 | <zcorpan> | Lachy: it does. you get a new implied section after the <section> element |
| 10:10 | <zcorpan> | so it's as if it was <section><h1/><p>...</section><::implied-section><h1/><p>...</::implied-section> |
| 10:10 | <zcorpan> | in the outline |
| 10:10 | <krijn> | Can't html5.validator.nu include the outline algorithm? :) |
| 10:11 | <BenMillard> | jgraham, you implemented the document outline algorithm IIRC? |
| 10:11 | <zcorpan> | krijn: http://bugzilla.validator.nu/show_bug.cgi?id=65 |
| 10:11 | <krijn> | Ah okay |
| 10:12 | <krijn> | Would be awesome :) |
| 10:26 | <BenMillard> | krijn, the Semantic Data Extractor from W3C works well for <h1>-<h6>, plus a few other things: http://xrl.us/ozrwg |
| 10:28 | <krijn> | Yeah, but I want to use it for something like http://fronteers.nl/bijeenkomsten/planning.html5 |
| 10:29 | <krijn> | (Which is a bad example, regarding <section>, but still) |
| 10:29 | <Hixie> | since the point of <section> is to make the headers easier, maybe we should just require each section to have a header |
| 10:30 | <MikeSmith> | Hixie: yes |
| 10:30 | <hsivonen> | Hixie: I think the key is getting a selector |
| 10:30 | <Hixie> | that too |
| 10:30 | <MikeSmith> | <section> doesn't seem particularly useful without a header |
| 10:32 | <hsivonen> | my knee jerk reaction is that trying to require stuff to be present is bad, because it makes it harder for editors to maintain continuous validity and because someone always has a legitimate counter case |
| 10:34 | <Hixie> | yeah |
| 10:34 | <krijn> | Would http://fronteers.nl/congres/2008/english.html5 be bad use of <section> then? |
| 10:34 | <hsivonen> | the best things that could happen with section: |
| 10:35 | <annevk3> | if we don't get the CSS issue sorted out <section> is sort of useless |
| 10:35 | <hsivonen> | 1) a major browser changing its UA style sheet to style by outline depth by default |
| 10:35 | <hsivonen> | 2) a major browser exposing the outline in the UI for quick navigation |
| 10:35 | <hsivonen> | annevk3: I agree |
| 10:36 | <MikeSmith> | if we don't require a heading for section, then I guess I'd question whether we want to have section at all. because the current content model for section is exactly the same as div; the differ only in semantics |
| 10:36 | <Hixie> | krijn: looks ok to me. <div>s in the <nav> could be sections too. and the sponsors could be <aside>. fwiw. |
| 10:37 | <BenMillard> | I think the only elements which should participate in the document outline are <h1>-<h6>. Even that much complexity is beyond a lot of authors at the moment. Also, <h1>-<h6> seem adequate for ATs when they are present and used in any sensible manner, without special sectioning elements and an outline depth selector. |
| 10:37 | <annevk3> | MikeSmith, <section> allows for nested headers without having to care about which <hx> element you use |
| 10:37 | <Hixie> | MikeSmith: lots of elements have the same content model as <div> |
| 10:37 | <Hixie> | anyway bed time |
| 10:37 | <Hixie> | nn |
| 10:37 | <BenMillard> | once I find a new source for funding, headings are the intended outline of documents is something I want to study |
| 10:37 | <BenMillard> | cya hixie |
| 10:38 | <krijn> | Fixed |
| 10:38 | <BenMillard> | s/headings are/headings and/ |
| 10:39 | <MikeSmith> | anyway, moot point until browsers show some interest in actually implementing support for <section> natively |
| 10:39 | <Philip`> | BenMillard: Do you have any idea whether many authors view HTML documents as a linear sequence of items, versus as a nested tree structure? |
| 10:40 | <BenMillard> | Philip`, interesting question. |
| 10:40 | <Philip`> | I'd guess the first viewpoint fits better with <h1>-<h6>, and the second with (nested) <section>s |
| 10:40 | <Lachy> | Hopefully Selectors 3 will get to PR soon enough and the CSS WG can start solving issues like sections and heading levels in Selectors 4 |
| 10:41 | <Philip`> | I suppose as soon as you start doing any kind of non-trivial layout, either with tables or with CSS, you've got to view the document as a tree otherwise it just won't make sense |
| 10:41 | <MikeSmith> | Hixie: as far as I can see, <section> is in fact the only element that has the same content model as <div> |
| 10:41 | <krijn> | BenMillard: hmm, you need funding for this stuff? :) |
| 10:41 | <BenMillard> | Philip`, my experience is that typical authors don't think about semantics, so their mental model of the document is simply how it looks on screen. |
| 10:41 | <MikeSmith> | they're the only ones that can start with style@scoped |
| 10:41 | <krijn> | Agreed with BenMillard |
| 10:42 | <krijn> | Most authors think of a way to make a design work |
| 10:42 | <krijn> | Most of the times that's where you start writing HTML |
| 10:43 | <krijn> | Unless there's some wireframe/interaction design you start with |
| 10:43 | <BenMillard> | krijn, my bank balance has this column where some values are positive some are negative. The sum of those numbers must be positive, otherwise I start getting worried. Without funding, it's hard to get enough positive values into that column whilst still doing significant amounts of HTML5 stuff. :) |
| 10:44 | <Philip`> | Hmm, so does that mean the tree view makes sense when it has a clear relationship with the layout (e.g. deeply nested tables/divs), but not so much when it's largely independent of layout (e.g. invisible <section>s)? |
| 10:46 | <krijn> | The tree view is important for CSS selectors as well |
| 10:46 | <krijn> | But as an author, you mostly control the rough layout parts.. Inside the body of an article, most of the stuff comes from a CMS, where you can't control anything |
| 10:46 | <krijn> | At least, that's my experience |
| 10:49 | <MikeSmith> | hmm, hsivonen looking at the HTML5 spec, I see that it currently seems to allow style@scoped in any flow content -- not just in div and section |
| 10:49 | <MikeSmith> | but the whattf.or schema restricts style@scoped to just being in div and section |
| 10:52 | <MikeSmith> | btw, do any UAs yet support style@scoped? (style in body content) |
| 10:53 | <annevk3> | no |
| 10:53 | <annevk3> | (style in body content they do, but that's not the same) |
| 10:54 | <MikeSmith> | annevk3: style in body content just gets moved to the DOM head and is unscoped, right? |
| 10:54 | <MikeSmith> | that is, @scoped is just ignored if it's present |
| 10:54 | <BenMillard> | MikeSmith, yeah. I've used <style> in body content to hide content further up the page in a demo in Firefox 2 and that worked. (So the <style> was not scoped.) |
| 10:55 | <MikeSmith> | I see |
| 10:55 | <annevk3> | MikeSmith, it should not get moved to the head I think |
| 10:56 | <BenMillard> | hiding #breadcrumb: http://calthorpepark.hants.sch.uk/temp/test/footer.htm |
| 10:56 | <BenMillard> | Firefix 2 moves the <style> to the <head> |
| 10:57 | <BenMillard> | s/Firefix 2/Firefox 2/ |
| 10:57 | <annevk3> | news at eleven, browsers have bugs! |
| 10:58 | <BenMillard> | hehe :D |
| 10:59 | <krijn> | They do? Wtf |
| 11:00 | <Philip`> | krijn: It's just a conspiracy by the browser developers to keep their QA people employed in this period of economic hardship |
| 11:02 | <krijn> | :) |
| 11:03 | <|tbb|> | anyone know where (irc server/channel) i can get help for -webkit-animation questions? |
| 11:03 | <MikeSmith> | unemployed QA people are know to cause all manner of havoc |
| 11:03 | <annevk3> | or maybe it's the QA and spec people keeping themselves employed |
| 11:03 | <annevk3> | |tbb|, #webkit ? |
| 11:03 | <MikeSmith> | like the Bathists that the US kicked out of their jobs in Iraq |
| 11:04 | <|tbb|> | annevk3: thx, sometimes it is so easy ;) |
| 11:12 | <jgraham> | Hmm Ben has gone so I can't argue with him about HTML4 headings being understandable by authors |
| 11:13 | <krijn> | Are they? |
| 11:14 | <|tbb|> | anyone has experience with webkit-animation ? |
| 11:15 | <Philip`> | jgraham: Are you wanting to argue in favour of or against the view that they are understandable? |
| 11:15 | <MikeSmith> | |tbb|: ping dino_ on #webkit |
| 11:16 | <jgraham> | Philip`: Against, otherwise it would be a kind or boring argument where we agreed |
| 11:17 | <krijn> | Hmm, http://fronteers.nl/congres/2008/english.html5 and http://fronteers.nl/congres/2008/english.xhtml5 work remarkably well.. (my first HTML5 test with a real site) |
| 11:17 | <Philip`> | jgraham: But he never claimed they were understandable - he said they were "beyond a lot of authors at the moment" |
| 11:21 | <jgraham> | Philip`: Oh well. I still disagree with the premise that that means we should not have <section> in html5 |
| 11:23 | <MikeSmith> | |tbb|: you might also try posting a question to the webkit-dev mailing list |
| 13:05 | <hsivonen> | http://agilewebdevelopment.com/plugins/assert_valid_markup_nu |
| 13:05 | <hsivonen> | anyone here tried it yet? |
| 13:18 | <BenMillard> | jgraham, I'm here now. :) |
| 13:26 | <jgraham> | BenMillard: You're wrong :p |
| 13:27 | <jgraham> | BenMillard: I did implement the HTML5 outline algorithm. There is a shoddy UI on james.html5.org |
| 13:27 | <BenMillard> | jgraham, cool |
| 13:28 | <annevk3> | so far over half of the requests to html5.org in December resulted in 403 (53046 requests) |
| 13:29 | <annevk3> | amount of requests declined enormously after I implemented that change as well |
| 13:29 | <BenMillard> | annevk3, I have similarly annoying stats on Project Cerbera |
| 13:29 | <BenMillard> | but for 404s |
| 13:29 | <annevk3> | first day 26000, implemented change, second day 13000 requests, third and fourth day 3000 requests each |
| 13:29 | <BenMillard> | like, things are requesting completely bogus URLs from my site which have never existed |
| 13:30 | <annevk3> | BenMillard, this is something else |
| 13:30 | <annevk3> | a bot was hitting html5.org quite hard |
| 13:30 | <annevk3> | using a distributed proxy network of some sorts |
| 13:30 | <BenMillard> | annevk3, do you think it was malicious or just dumb? |
| 13:31 | <annevk3> | no idea |
| 13:35 | <BenMillard> | jgraham, about a year ago I summarised some of the common approaches to using <h1>-<h6> amongst accessibility professionals from accessify.com: http://projectcerbera.com/web/articles/heading-numbers#interpretations |
| 13:36 | <BenMillard> | jgraham, I've collected examples of how headings are done on the web here: http://projectcerbera.com/web/study/2008/headings |
| 13:37 | <BenMillard> | jgraham, the biggest problem from an accessibility point of view is heading text which is not marked up using heading elements. |
| 13:43 | <jgraham> | BenMillard: When I implemented a Firefox extension for document outlines, the biggest problem apart from simply missing headings was people who didn't connect their use of headings to the logical structure of their document, even though that logical structure was often expressed in <div>s |
| 13:43 | <BenMillard> | krijn, here's an HTML5 outline algorithm thingy from jgraham: http://james.html5.org/outliner.html |
| 13:44 | <BenMillard> | jgraham, really? Nearest I've found to a fully sectioned document are ones which use <blockquote> as if it were a section element, so they get indented sections. |
| 13:44 | <BenMillard> | for example: http://www.opengroup.org/onlinepubs/000095399/functions/glob.html |
| 13:45 | <BenMillard> | what sort of disconnects did you find? an <h2> before the first <h1>, things like that? |
| 13:48 | <jgraham> | BenMillard: Things like people using <h2> for all their subheadings and then <h4> for all the sidebar headings "because they are less important", so the sidebar ends up under the final subsection |
| 13:48 | <BenMillard> | jgraham, yeah I've seen things like that as well |
| 13:49 | <krijn> | BenMillard, jgraham: doesn't work too well, but could be me.. |
| 13:49 | <BenMillard> | krijn, yeah I entered a URL and clicked "Load" and still waiting... |
| 13:49 | <krijn> | :) |
| 13:49 | <jgraham> | Well it worked once upon a time |
| 13:50 | <krijn> | It also doesn't like <h1><img alt="Foo"></h1> |
| 13:50 | <BenMillard> | jgraham, that's basically fine for the user experience in practice, so long as it's done consistently across the website |
| 13:50 | <jgraham> | BenMillard: Not for visual users displaying the headings in a tree |
| 13:50 | <jgraham> | At least for me it sucked |
| 13:50 | <BenMillard> | jgraham, who would do that? If you can see the document, you don't need a |
| 13:50 | <BenMillard> | oops |
| 13:51 | <BenMillard> | if you can see the document, you don't need a document outline to show you what you can already see? |
| 13:51 | <BenMillard> | (yay arguments! :P) |
| 13:52 | <BenMillard> | having heading elements at all creates a dramatic improve in accessibility since "next heading" and "previous heading" are a boon for zooming around the page non-visually |
| 13:53 | <jgraham> | BenMillard: http://gnuplot.info/docs_4.0/gnuplot.html |
| 13:53 | <jgraham> | BenMillard: Also if it is useless it is pretty surprisingly common in editors/viewers for other languages e.g. word, many pdf readers |
| 13:55 | <BenMillard> | jgraham, that's a really long document with many sections...not very typical of web pages I find through random searching and browsing. :) |
| 13:55 | <BenMillard> | jgraham, does your Firefox extension support Firefox 2? if so I could try it out |
| 13:55 | <zcorpan> | hsivonen: what is the plugin for? |
| 13:56 | <zcorpan> | hsivonen: i.e. which app? |
| 13:56 | <jgraham> | BenMillard: Officially it only supports 2 because I didn't update it for 3 |
| 13:56 | <hsivonen> | zcorpan: looks like its for running unit tests in the context of Rails |
| 13:56 | <BenMillard> | jgraham, cool got a link for it? |
| 13:57 | <BenMillard> | jgraham, the Information > Show Document Outline feature in the Web Developer Toolbar for Firefox works on that document, just with <h1>-<h6>, although you can't click it's output to scroll the document it came from. |
| 13:57 | <BenMillard> | s/Show Document Outline/View Document Outline/ |
| 13:57 | <jgraham> | https://addons.mozilla.org/en-US/firefox/addon/475 |
| 13:58 | jgraham | notices that some people seem to like it which suggests non-disabled users have a use for this kind of feature |
| 13:59 | <BenMillard> | jgraham, I can totally see the handiness of this for people who frequently browse massive documents, like specs and big computery manual thingies. |
| 13:59 | <BenMillard> | jgraham, and it works without specialist sectioning markup. ;) ;) |
| 14:01 | <jgraham> | BenMillard: It needs <h1>-<h6> |
| 14:01 | <jgraham> | But that design just lends itself to brokenness |
| 14:01 | <BenMillard> | jgraham, by "specialist sectioning markup" I mean things like <section>, <article>, etc |
| 14:02 | <BenMillard> | jgraham, do you think requiring more markup from authors would reduce the frequency of brokenness on the web, then? |
| 14:02 | <krijn> | Just <h> (or now <h1>) with good selector support would've saved so much trouble :/ |
| 14:02 | <BenMillard> | sectioning elements + heading elements is more markup than heading elements alone |
| 14:03 | <jgraham> | BenMillard: I think markup that corresponds to the structure being represented should lead to less brokenness |
| 14:04 | <jgraham> | Headings were badly designed in the first place, presumably because there was no easy way to style based on nesting level but different elements were easy to style differently |
| 14:04 | <jgraham> | I think it is worth trying to fix both those problems, subject to the usual constraints |
| 14:05 | <hendry> | can i confirm that inputmode has been dropped from HTML5? |
| 14:06 | <BenMillard> | jgraham, I'm not sure we agree what "brokenness" is. |
| 14:06 | <BenMillard> | I consider <p><strong> for <h4> to be broken. I consider or using <h1> for big bold text to be broken. |
| 14:06 | <BenMillard> | s/or using/using/ |
| 14:06 | <jgraham> | No disagreement so far |
| 14:07 | <BenMillard> | I consider <h1> for main content heading with <h2> for company name in the header to be fine |
| 14:07 | <BenMillard> | (so an <h2> before the first <h1>) |
| 14:08 | <annevk3> | hendry, yes |
| 14:08 | <jgraham> | I consider that particular example suboptimal but of little consequence |
| 14:08 | <jgraham> | (HTML5 deals with it fine) |
| 14:08 | <BenMillard> | jgraham, fair enough. |
| 14:08 | <BenMillard> | jgraham, this document has no <h1> so is that broken? http://gnuplot.info/docs_4.0/gnuplot.html |
| 14:09 | <hendry> | annevk3: ta |
| 14:10 | <jgraham> | BenMillard: I don't really carewhich headings people use an an absolute sense, so long as they have the correct relationship to each other |
| 14:11 | <BenMillard> | jgraham, interesting. |
| 14:13 | <jgraham> | BenMillard: Unless you can demonstrate tha it is practical to extract some information from markup I'm not that bothered by how it is used. I doubt you can extract much meaningful information from absolute header numbers that you could not get by making a tree |
| 14:16 | <BenMillard> | jgraham, do you agree that using any heading element around heading text is better than using no heading element around that text? |
| 14:19 | <jgraham> | Yes |
| 14:20 | <BenMillard> | jgraham, do you agree that heading elements are uncommon even though heading text is common? |
| 14:20 | <jgraham> | BenMillard: I don't know. I suspect that is true |
| 14:22 | <BenMillard> | jgraham, find-in-page for "h1" to "h6": http://canvex.lazyilluminati.com/survey/2007-07-17/analyse.cgi/index |
| 14:22 | <BenMillard> | and use common sense to realise that just about ever page has multiple runs of heading text :) |
| 14:23 | <BenMillard> | s/ever/every/ |
| 14:23 | <BenMillard> | the "Tag names (page)" bars are the best for seeing this |
| 14:28 | <BenMillard> | jgraham, am I right in saying we agree the biggest problem is getting authors to indicate heading-ness in the first place, and indicating the correct level is the second biggest problem? |
| 15:06 | <krijn> | The biggest problem is tools, letting authors produce something like <p><strong>Some sentence without a period</strong></p> |
| 15:06 | <krijnh> | Doh |
| 15:06 | <krijnh> | BenMillard: still discussing headings? :) |
| 15:07 | <jgraham> | BenMillard: I agree with the problems. I think <section> is more likely to get the relative ordering right. I don't know what is more likely to get people using headings at all but I guess it doesn't make too much difference |
| 15:38 | <BenMillard> | jgraham, <section>+<h1> requires more elements and a special CSS selector. That dooms its prospects of becoming widespread, from the authoring practices I've seen. The bigger problem is headingness, so we shouldn't make headings in HTML more complicated by adding a doomed solution to a lesser problem. |
| 15:41 | <krijnh> | h2, h3 and h4 also still work, which don't require more markup, I think <section><h1 /></section> solves the <div><h4 /></div> problem |
| 15:41 | <krijnh> | Or am I not getting it? |
| 15:41 | <BenMillard> | (<section>+<h1> is a shorthand for the combination of <article>, <aside> and <section> with <h1> (or <h1>-<h6>) in HTML5) |
| 15:42 | <BenMillard> | krijnh, if they work then why introduce section elements and the ability to only use <h1> when sectioning elements are present? |
| 15:42 | <BenMillard> | (my point is that there's no reason to introduce these things that outweighs the complexity of adding them) |
| 15:44 | <annevk3> | BenMillard, <section>+<h1> is crucial if you have an article you want to use in several contexts, e.g. on blogs |
| 15:45 | <BenMillard> | annevk3, I'd hardly call it crucial since people get by without it fine already. |
| 15:45 | <annevk3> | BenMillard, modifying the value of x in <hx> properly when publishing an article in several contexts is just not trivial |
| 15:45 | <annevk3> | BenMillard, I avoid the use of headers in blog posts exactly because of this |
| 15:45 | <jgraham> | BenMillard: Especially when syndicating from multiple sources |
| 15:45 | <annevk3> | yeah, for planet.intertwingly the problem is even worse |
| 15:48 | <jgraham> | BenMillard: For the kind of people who care enough to use headings at all I think the complexity of <section> et.al. is small. Certianly people have said to me in person that they want to use the new elements |
| 15:48 | <BenMillard> | annevk3, doing trivial string replacement to change heading levels automatically has worked fine for entries I write on my blog for over a year. |
| 15:49 | <BenMillard> | jgraham, people get seduced by novelty. I'd rather not do a new study in 20 years time just to say "See! I told you so!" :P |
| 15:49 | <jgraham> | OTOH the complexity of <h1>-<h6> is kind of hidden. They look simple but people always use them in odd ways |
| 15:50 | <BenMillard> | jgraham, playing fast an loose with heading levels doesn't matter if it's done consistently across a website. |
| 15:50 | <krijnh> | Not saving the <hx> markup anywhere in a database (only the heading level in the current context) makes it a lot easier as well |
| 15:51 | <BenMillard> | as for syndicating from multiple sources, I suggest that is non-trivial to start with and that there are many corrections other than heading depth which must be made to do it "properly" |
| 15:51 | <gsnedders> | What Lachy was getting me to write was a script to replace all <hx> elements with their correct depth, ignoring the fact that <section> resets the x required |
| 15:52 | <annevk3> | BenMillard, string replacement is not really acceptable for markup |
| 15:52 | <BenMillard> | annevk3, tell that to the web. :) It's been faultless on my blog, and I suck at programming... |
| 15:52 | krijnh | shakes BenMillards hand :) |
| 15:53 | BenMillard | laughs! |
| 15:58 | <BenMillard> | wow, Yahoo! Slurp is still accessing URLs I created permanent redirects for about 3 years ago... |
| 15:58 | <gsnedders> | BenMillard: The tools won't save us. HTTP is badly implemented. |
| 15:59 | <krijnh> | Slurp is slurping all my sites as well, without ever providing any visitors :/ |
| 15:59 | <krijnh> | (And it's the only one slurping rel="alternate" type="application/xhtml+xml") |
| 16:01 | <BenMillard> | krijnh, interesting. :) |
| 16:01 | <BenMillard> | Yahoo! is usually my 3rd biggest traceable referrer (or 2nd if you count Google and Google Images as 1 referrer): http://projectcerbera.com/blog/2008/04/bandwidth |
| 16:02 | <BenMillard> | gsnedders, yeah I've seen other people complain about permanent redirects not being treated as permanent....I expected a search engine to get that right though. |
| 16:03 | <gsnedders> | BenMillard: SimplePie doesn't at all. |
| 16:03 | <Philip`> | Google is the only one I've seen that crawls URLs from scripts as well as from the HTML |
| 16:05 | <krijnh> | Google Reader, Firefox Live Bookmarks.. Also not respecting 301's.. |
| 16:06 | Philip` | doesn't entirely understand why search engines all seem to have completely different lists of places from which they extract URLs to crawl, and haven't yet converged to anything consistent |
| 16:06 | <gsnedders> | http://diveintomark.org/tests/client/http/ |
| 16:06 | BenMillard | wonders what proportion of requests that look like they come from search bots actually come from spambots and other malware using common search bot UA strings... |
| 16:10 | <Philip`> | BenMillard: I suppose that would be easy to determine, by just seeing who the IP address belongs to |
| 16:10 | <yecril71> | getElementByIdInSubtree is not that good really. |
| 16:11 | <yecril71> | It solves only the part of the problem. |
| 16:11 | <yecril71> | a part. |
| 16:11 | <yecril71> | What is needed here are local identifiers. |
| 16:11 | <yecril71> | They need not be unique throughout a document. |
| 16:12 | <krijnh> | IE5.5 requesting 10 pages all in the exact same second sure does look like a bot.. And then it also fetches stylesheets and javascripts :\ |
| 16:12 | BenMillard | just traced that a 404 was being caused by Slup trying to access a REALLY REALLY old URL, which goes through 3 layers of rewriting that ends up removing a necessary dash. |
| 16:13 | <Philip`> | (Since "descendant" is hard to spell, maybe the term "offspring" or "progeny" would work better) |
| 16:13 | <yecril71> | The local identifier, when it is referred to by an attribute, should be located in the subtree of the calling element first. |
| 16:13 | BenMillard | resolves to refactor his legacy .htaccess rules...which won't be fun. |
| 16:14 | <krijnh> | BenMillard: you should get yourself better tools ;) They'll save you! |
| 16:14 | <yecril71> | If that fails, the parent nodes should be considered, until a subtree with a matching descendant is found. |
| 16:15 | <yecril71> | That would be a lifeboat for people using server-side templates. |
| 16:15 | <Philip`> | BenMillard: You should have unit tests so you can refactor your .htaccess safe in the knowledge that you won't accidentally break any links :-) |
| 16:16 | <Philip`> | yecril71: That would probably be incompatible with existing content, which relies on it always returning the first element (in document order) with that id |
| 16:17 | <krijnh> | http://projectcerbera.com/blog/2008/04/bandwidth/foo/bar/baz/hmm/i/need/a/htaccess/update/k10x :) |
| 16:17 | <BenMillard> | Philip`, that's a really good idea. Evidently my previous URL changes have caused little bits of breakage. Is there a guide to doing that well? |
| 16:17 | <BenMillard> | krijnh, whoa...what the hell... |
| 16:18 | <krijnh> | Err, yeah :) |
| 16:19 | <Philip`> | BenMillard: I'm not aware of any - it's the kind of thing that I've thought I should do, but never got around to |
| 16:20 | <Philip`> | http://projectcerbera.com/blog///////////////////////// is fun |
| 16:20 | <BenMillard> | Philip`, I'll have a look around when I bite the bullet to try and fix this once and for all |
| 16:21 | <krijnh> | Well Yahoo Slurp, eat your heart out :) |
| 16:21 | <BenMillard> | LOL...how are those even working? |
| 16:33 | <BenMillard> | zcorpan, Yahoo! Slup still enters from the .html version of my URLs, so it always starts by going through the permanent redirect we designed nearly 2 years ago when I moved to extensionless URLs. :| |
| 16:39 | <gsnedders> | http://www.詹姆斯.com/feed breaks pretty much everything |
| 16:40 | <gsnedders> | Sorry, I'm just trying out other feed readers |
| 16:40 | <gsnedders> | Suggestions? |
| 16:40 | <BenMillard> | Sage is an add-on for Firefox which I've used for a long time: https://addons.mozilla.org/en-US/firefox/addon/77 |
| 16:41 | <gsnedders> | BenMillard: Main issue with that: I don't use Firefox. |
| 16:41 | <BenMillard> | gsnedders, not even for testing? |
| 16:41 | <gsnedders> | BenMillard: Yes, but I do little testing at the moment :) |
| 16:43 | <BenMillard> | gsnedders, well Sage has served me...I don't recall seeing a broken feed in it but then I have a tiny number of quite predictable feeds. |
| 16:43 | Philip` | wonders in what way that feed breaks |
| 16:43 | <gsnedders> | BenMillard: I am subscribed to a number of feeds that do horrible things just to break stuff |
| 16:43 | <gsnedders> | Philip`: It's because the XHTML namespace is stored as an entity |
| 16:44 | <gsnedders> | Philip`: I don't know why that breaks a lot of feed readers seeming most do at least use an XML parser |
| 16:44 | <Philip`> | gsnedders: Oh - it seems to work fine in Opera at least |
| 16:44 | <gsnedders> | Philip`: It completely fucks up Safari, and Google Reader doesn't take kindly to it either |
| 16:46 | <BenMillard> | gsnedders, that page for Sage claims 2,184,552 total downloads, which I guess makes it a fairly popular choice (over 100,000 actual users?) |
| 16:46 | <gsnedders> | http://simplepie.org/demo/?feed=www.詹姆斯.com%2Ffeed works :P |
| 16:55 | <Philip`> | gsnedders: If I go to http://simplepie.org/404/ then the big search box gives an empty page when I try to use it :-( |
| 16:59 | <gsnedders> | Philip`: EPIC FAIL. |
| 17:01 | <Philip`> | gsnedders: It's a disaster |
| 17:02 | <gsnedders> | Philip`: I know, I'm so sad :( |
| 19:08 | <yecril71> | getElementByID is not called on an Element at present, so there are no backward compatibility issues involved. |
| 22:09 | gsnedders | yawns |
| 22:10 | gsnedders | is trying to convince someone that he does actually want his CMS to output HTML |
| 22:10 | <krijnh> | Pick me, pick me :) |
| 22:11 | <krijnh> | gsnedders: a customer of yours? |
| 22:12 | <gsnedders> | krijnh: No, I'm just looking for a CMS to use for a website, and I was asking in #drupal-support whether Drupal could output HTML |
| 22:12 | <smedero> | I imagine you got a lot of blank stares. |
| 22:13 | <Dashiva> | It can output broken XHTML, why would you ever want HTML? |
| 22:13 | <krijnh> | I picked HTML output for my CMS 3,5 years ago, and still happy with it (as are my customers) |
| 22:14 | <krijnh> | OTOH is outputting 'broken XHTML' means a solidus here and there, who cares |
| 22:14 | <krijnh> | *if |
| 22:17 | <krijnh> | gsnedders: take a crash course Dutch and you can use mine for free ;) |
| 22:17 | <gsnedders> | krijnh: :) |
| 22:17 | <gsnedders> | krijnh: I think I have too much on my plate with four AHs. I have a computing project which I need to do in the next two weeks, and an English dissertation in the next four :) |
| 22:18 | <krijnh> | Heh |
| 22:18 | <krijnh> | To get to know Drupal, 2 weeks isn't enough :) |
| 22:18 | <gsnedders> | I'm not planning on doing any such thing. :) |
| 22:19 | <gsnedders> | What I'm doing is a module for Anolis 1.1 disguised as being a program in itself :) |
| 22:20 | <gsnedders> | Anyone know of a CMS that can take HTML input (and sanitize it) and that uses a serializer? |
| 22:44 | <aboodman> | Hixie: what are the latest thoughts around adding crypto primitives? |