| 00:00 | <zcorpan_> | so you can do <h1>foo</h1><h2>bar</h2><hgroup><h3>baz</h3><h4>quux</h4></hgroup> |
| 00:01 | <zcorpan_> | where baz is a subsection of bar |
| 00:01 | <Hixie> | non-issue in practice, imho, <hgroup> is almost always going to be the first heading |
| 00:01 | <Hixie> | and the rest of the time, people can use <section> |
| 00:01 | <Hixie> | no? |
| 00:03 | <zcorpan_> | would be nice if it worked as people expect when using implied sections |
| 00:03 | <Hixie> | i guess |
| 00:03 | <Hixie> | ...send mail :-) |
| 00:04 | <zcorpan_> | jgraham: btw your outliner seems to use the wrong heading in the example above (it uses foo twice) |
| 00:15 | <MikeSmith> | Hixie: (or anybody else using tircd) - http://impressive.net/people/gerald/2009/05/xchat-inputcount.pl |
| 00:16 | <MikeSmith> | [[ |
| 00:16 | <MikeSmith> | # xchat-inputcount.pl: add a character count next to xchat's input box. |
| 00:16 | <MikeSmith> | # (useful when twittering, to see if you are within 140 characters) |
| 00:16 | <MikeSmith> | ]] |
| 00:48 | <zcorpan_> | ↕The start tag SHOULD contain ↕ a ↕version ↕↕ attribute that declares ↕ the ↕version of XHTML in use. The version of this version of XHTML is ↕↕ -//W3C//DTD XHTML 1.1//EN ↕. |
| 00:48 | <zcorpan_> | http://www.w3.org/TR/2009/PER-xhtml11-20090507/xhtml11-rec-diff.html |
| 00:49 | <zcorpan_> | doctype is MAY |
| 00:50 | <Hixie> | i'm amused as to how the main effect HTML5 has had on the XHTML2 WG is that the XHTML2 WG keeps releasing new XHTML 1.x drafts instead of working on XHTML2 |
| 00:51 | <zcorpan_> | 1.1 adds lang="" |
| 00:52 | <Hixie> | does it define how it works? |
| 00:52 | <Hixie> | http://www.whatwg.org/specs/web-apps/current-work/#the-lang-and-xml:lang-attributes |
| 00:52 | <zcorpan_> | The ↑↑ lang ↑↑ attribute is defined in [ ↑↑ HTML4 ↑↑ ]. When this attribute and the ↑↑ xml:lang ↑↑ are specified on the same element, the ↑↑ xml:lang ↑↑ takes precedence. ↑↑ When both ↑↑ lang ↑↑ and ↑↑ xml:lang ↑↑ are specified on the same element, they SHOULD have the same value. ↑ |
| 00:53 | <Hixie> | so "no" then? |
| 00:54 | <Hixie> | i guess they just defer to HTML4 |
| 00:54 | <Hixie> | as for everything else |
| 00:56 | <zcorpan_> | img& usemap (IDREF) |
| 00:56 | <Hixie> | they broke it again? |
| 00:56 | <Hixie> | am i going to have to update my blog post |
| 00:56 | <zcorpan_> | was it ever fixed? |
| 00:56 | <Hixie> | http://ln.hixie.ch/?start=1172653243&count=1 |
| 00:57 | <zcorpan_> | "...new version of this specification which did not fix one of its simplest problems" |
| 01:03 | <Hixie> | woah, creativecommons.org is branding BSD and GPL licenses as "cc"? |
| 01:03 | <zcorpan_> | hmm, 1.1 made the doctype "may" but in 1.0 it's still "must" |
| 01:12 | <zcorpan_> | http://lists.w3.org/Archives/Public/www-style/2009May/0065.html |
| 01:14 | zcorpan_ | is confused as to why there should be pseudo-classes for aria |
| 01:14 | <zcorpan_> | oh well |
| 01:14 | <zcorpan_> | nn |
| 01:45 | <hsivonen> | Hixie: CC has called the GPL CC-GPL as early as when CC 1.0 series came out |
| 01:47 | <Hixie> | that's... surprising |
| 01:49 | <hsivonen> | I think non-program work and program licensing would be more interoperable if insteat of drafting CC-by and CC-(by)-sa CC had promoted CC-BSD and CC-GPL for all kinds of works |
| 01:49 | <hsivonen> | *instead |
| 01:49 | <Hixie> | yes, imho cc has done much harm through its proliferation of licenses |
| 01:50 | <hsivonen> | although I guess it's considered a feature that CC-by-sa doesn't require the preferred form of making modifications to be made available |
| 01:51 | <Hixie> | ironically, none of the CC licenses these days are appropriate for what I want at all (since they all require attribution) |
| 01:51 | <hsivonen> | given current state of the Internet and codecs, one would want to avoid distributing the preferred form of making modifications to many videos |
| 01:51 | <hsivonen> | yeah, attribution on the license level is a big problem |
| 01:51 | <hsivonen> | particularly for massively collaborative works like wikis |
| 01:52 | <hsivonen> | (when you'd want to integrate external works that don't specify the wiki as the party who receives attribution) |
| 02:45 | <roc> | the preferred form of making modifications to videos is the raw data |
| 05:12 | <olliej> | Philip`: ping? |
| 08:01 | <Philip`> | olliej: Pong |
| 08:02 | <olliej> | Philip`: you need to update your canvas compat chart :D |
| 08:04 | <Philip`> | olliej: I need to update the tests too |
| 08:04 | <olliej> | true |
| 08:04 | <Philip`> | particularly so I can discover how badly everyone fails at text rendering :-) |
| 08:04 | <olliej> | :p |
| 08:05 | <Philip`> | but that'll take quite a bit of time, and I haven't been sufficiently motivated to do it |
| 08:14 | <zcorpan_> | hgroup > h1:heading-level(3) ~ h2 { /* same styles as <h4> */ } |
| 08:16 | <MikeSmith> | whoa, what is this heading-level() magic? |
| 08:16 | <MikeSmith> | is that already supported in browsers? |
| 08:16 | <zcorpan_> | no |
| 08:17 | <zcorpan_> | it's something i made up this morning |
| 08:17 | <MikeSmith> | damn lazy browser developers, get to work implementing that already! |
| 08:18 | <zcorpan_> | the outline algorithm needs to be fixed first, at least :) |
| 08:19 | <MikeSmith> | zcorpan_: well, I hope your idea doesn't take |
| 08:19 | <MikeSmith> | because I like putting stuff into stylesheets like |
| 08:20 | <MikeSmith> | > ul > li > ul > li > ul > li > ul > li > ul > li |
| 08:20 | <MikeSmith> | when I look at it in vim, it shows up in red and yellow |
| 08:20 | <MikeSmith> | like christmas tree lights |
| 08:24 | <MikeSmith> | div[class="section"] > div[class="section"] > div[class="section"] > div[class="section"] > div[class="section"] > h2 |
| 08:24 | <MikeSmith> | ... is fun too |
| 08:24 | <MikeSmith> | that one adds pink |
| 09:58 | <jgraham> | zcorpan_: You can imagine doing something like hgroup > *:not(:heading-level(*)) to match subheadings |
| 09:58 | <jgraham> | But it is not really obvious or pretty |
| 10:04 | <Philip`> | Hmm, I've never tried WEFT (Microsoft's tool for generating EOTs) until now, but now that I have I can understand why EOTs never became popular |
| 10:05 | <Philip`> | I guess whoever wrote it really wanted to be writing an IDE, not a tool that simply needs two inputs (a font file, and the characters you want to include in it) and one output (a new font file) |
| 10:06 | <Philip`> | so it's all got customisable UI and it hooks into IE's rendering engine to extract the fonts and characters you're using, and can integrate with something on the web server to upload the new font files and new HTML/CSS and all that stuff |
| 10:07 | <Philip`> | plus it's got a project system, and a wizard, and a Windows menu with Tile Vertically and Tile Horizontally and Cascade Windows and all those really useful features |
| 10:08 | <jgraham> | Ah, you know you've gone really wrong when you find you have implemented a project system |
| 10:08 | <Philip`> | (This is version 5.3.2 of the tool) |
| 10:09 | <Philip`> | (from 2003) |
| 10:09 | <Philip`> | It's even got a mottled stone background in its toolbar |
| 10:10 | <Philip`> | http://www.em2-solutions.com/projs/weft.html - apparently it's those Swedish people's fault |
| 10:18 | <zcorpan_> | silly swedish people |
| 10:19 | <jgraham> | You're telling me :p |
| 10:19 | <zcorpan_> | jgraham: also has weird results when you have invalid markup like <hgroup><center><h1> |
| 10:20 | <Philip`> | Also, the fonts it generates seem a bit strange - e.g. if I have a page using normal Calibri, it doesn't draw any ligatures; but if I use WEFT to create an EOT for that page, the new font does draw ligatures |
| 10:20 | <Philip`> | except it doesn't really draw them properly - I have the word "fluffily" and it draws a 'fl' ligature, 'ff' ligature, and then it tries to draw 'i' except that glyph isn't in the font so it falls back on a default font instead |
| 10:21 | <zcorpan_> | is the "fi" ligature in the font? |
| 10:21 | <zcorpan_> | and the "f" glyph? |
| 10:21 | <jgraham> | zcorpan_: In the outliner. Weird. Thanks |
| 10:21 | <jgraham> | s/./?/ |
| 10:22 | <Philip`> | (Also it doesn't handle substitutions properly, e.g. i followed by combining acute accent gets drawn like an i with an accent on its left, rather than a dotless i with an accent where the dot would go) |
| 10:23 | <Philip`> | zcorpan_: 'f' is; 'i' isn't; 'fi' isn't |
| 10:23 | <zcorpan_> | Philip`: then it would have to fall back on a default font anyway (since i and fi aren't present), no? |
| 10:24 | <Philip`> | Also, when I tested this yesterday it alternated randomly between two renderings when I refreshed the page, but now it seems more consistent... |
| 10:24 | <Philip`> | zcorpan_: No, it should use the 'ffi' ligature |
| 10:24 | <zcorpan_> | oh, forgot about ffi |
| 10:25 | <Philip`> | I guess it thinks there's an 'ffi' when it's deciding which glyphs to include in the font, but then forgets about the substitution rule for that ligature and so the characters "ffi" get turned into the glyphs 'ff' and 'i' instead of into 'ffi' |
| 10:26 | <Philip`> | or something like that |
| 10:26 | <Philip`> | Anyway, it lets me conclude that my font subsetting code is not a waste of time repeating what their tool/API already does, because I avoid most of these bugs :-) |
| 10:27 | <jgraham> | Philip`: But you don';t have a project mode! |
| 10:28 | <Philip`> | jgraham: I'm too lazy to do that, so I'll stick with my checkbox-to-select-font and textarea-to-select-characters and button-to-generate-the-output UI for now |
| 10:32 | Philip` | thinks someone should make an EOT-to-TTF service |
| 11:45 | <zcorpan_> | i think smylers might be on to something (re header and hgroup) |
| 12:06 | <zcorpan_> | http://forabeautifulweb.com/blog/about/lead_pipe/ - the two paragraphs talking about leading seems to be a reasonable use case for style="" |
| 12:06 | <jgraham> | zcorpan_: Yeah, it sounds plausible. My only concern is that it violates "explicit is better than implicit" by making <header> have surprising magic properties |
| 12:07 | <jgraham> | (which could be bad if we don't accuratley anticipate how <header> will be used) |
| 12:08 | <jgraham> | (although you could argue that the table headers algorithm is also rather magic, but at least that was based on the failure of the explicit solutions + research) |
| 12:08 | <jgraham> | (but I guess we can do research on this case) |
| 12:10 | <jgraham> | Philip`: If you have time it might be nice to look at how headers are used as children of <div class="header">. Like collect all instances of <div id/class="header"> and filter just <hx> desendants and <div> descendants |
| 12:24 | <zcorpan_> | <header><h1>x</h1><nav><h1>y</h1></nav><h1>z</h1></header> - what would you make of this? |
| 12:34 | <jgraham> | zcorpan_: I would assume that the second <h1> child of the <header> represented a different subsection |
| 12:35 | <jgraham> | So: |
| 12:35 | <jgraham> | +--x |
| 12:35 | <jgraham> | +--y |
| 12:35 | <jgraham> | +--z |
| 17:59 | <Philip`> | jgraham: Haven't you set something up so you can collect that kind of data yourself? :-) |
| 18:00 | <Philip`> | by which I mean I probably won't be motivated to find time to look at it myself, since I'd probably have to write lots of Java to extract the right data, and I don't like writing DOM/SAX code in Java |
| 18:21 | gsnedders | sighs at some of the arguments around the intarwebs about HTML 5 |
| 18:21 | <gsnedders> | Would it not be more productive to send substantial comments about the spec rather than write substantial blog posts bitching about the process (whether fair or not). |
| 18:24 | <Philip`> | gsnedders: No, because substantial comments about the spec would get disregarded due to the problems with the process |
| 18:24 | <Philip`> | and the solution to such a problem is not to send more substantial comments that will all get disregarded |
| 18:24 | <gsnedders> | OK, then bitch to W3C management and not on blogs? |
| 18:25 | <gsnedders> | Fork the spec? |
| 18:25 | <inimino> | gsnedders: not for people who have disengaged with the process by choice |
| 18:25 | <gsnedders> | They all seem more productive things to do in that case. |
| 18:25 | <inimino> | forking the spec sounds like work |
| 18:25 | <gsnedders> | Bitching on blogs will _not_ fix any problems. |
| 18:25 | <inimino> | which doesn't really satisfy the same human urges that bitching does ;) |
| 18:26 | <Philip`> | gsnedders: Those are worse options in the cost/benefit ratio, because complaining on blogs and is easy and it might possibly eventually cause some change, whereas the other options require a lot more effort |
| 18:27 | <Philip`> | The current evidence indicates it takes years of work for an unofficial fork of HTML to be seen as legitimate |
| 18:28 | <inimino> | and that's with the backing of major browser developers |
| 18:29 | <Philip`> | It's like the current HTML 5 is an asteroid heading catastrophically towards Earth, and blog comments are like bouncing little rocks off it in an attempt to deflect it onto a course that won't destroy the things those people care about |
| 18:30 | <Philip`> | whereas you're suggesting building an entire new asteroid and making sure that hits Earth first and knocks it out of the way so the other asteroid won't have any detrimental impact |
| 18:32 | <Philip`> | and your other suggestion is that those people should complain to God so that He declares the asteroid as unholy in the hope that the asteroid will realise the error of its ways |
| 18:32 | gsnedders | stops bitching about the bitchers |
| 18:32 | <Philip`> | I like this analogy |
| 18:33 | <inimino> | people who see the standards process as a way to strong-arm browser developers are unlikely to find satisfaction in the current process |
| 18:33 | gsnedders | wonders how Content-Type is parsed |
| 18:34 | <inimino> | (or in any other process that has a realistic chance to displace it) |
| 18:34 | <gsnedders> | Short immediately apparent answer: not how the spec says. |
| 18:35 | <inimino> | so I think there is some fallout from the breakdown of unrealistic expectations that had seemed reasonable for a few years |
| 18:36 | <gsnedders> | How boring. That page now has sane headers. |
| 18:37 | <gsnedders> | Garbage at the end is ignored. |
| 18:40 | <jgraham> | Philip`: Er, the point is that I haven't finished writing the data parsing tools |
| 18:41 | <jgraham> | Particularly the more advanced stuff like integrating the validator.nu parser |
| 18:41 | <Philip`> | jgraham: Sounds like you've now got an excellent reason to finish them ;-) |
| 18:41 | <gsnedders> | Actually, garbage in general is ignored. |
| 18:42 | <jgraham> | Philip`: so would it obviously be horrendously slow to pipe sax output from validator.nu to other, saner, programming languages? |
| 18:43 | <jgraham> | Or is there some other reason that is a bad idea? |
| 18:43 | <Philip`> | jgraham: That does sound horrendously slow |
| 18:43 | <Philip`> | jgraham: You could use Jython |
| 18:43 | <Philip`> | (which shouldn't be as horrendously slow) |
| 18:43 | <jgraham> | Yeah, maybe Jython would be a good idea |
| 18:44 | <Philip`> | The overhead of serialisation and IPC and parsing seems like a really bad idea when there's so much data |
| 18:44 | <Philip`> | so processing the data as close to the parser as possible is a good thing |
| 18:44 | <Philip`> | Ooh, you could use Scala! |
| 18:45 | <jgraham> | Philip`: The idea, I guess, would be to pre-filter the data s much as possible (e.g. using regexps) |
| 18:45 | <Philip`> | I don't know what that is but people have talked about it so it must be alright |
| 18:45 | <Philip`> | jgraham: Regexps are slow :-p |
| 18:45 | <jgraham> | Faster than parsing :-p |
| 18:45 | <Philip`> | Hmm, not sure how much difference there'd be |
| 18:46 | <Philip`> | (particularly for streaming SAX parsing) |
| 18:46 | <jgraham> | Yeah but with the SAX parsing you still have to do more work to get useful information out |
| 18:47 | <Philip`> | Depends what information you want to get out |
| 18:47 | Philip` | 's Java stuff uses SAX for some things and DOM for other things, depending on which seemed easier |
| 18:48 | jgraham | generally wants to end up writing processing code in something less painful than java |
| 18:48 | <jgraham> | Since it will just be a bunch of one-shot filters |
| 18:48 | <Philip`> | I haven't found it painful to write processing code in Java |
| 18:49 | <jgraham> | Philip`: You have found it painful enough that you don't want to do this for me :p |
| 18:49 | <Philip`> | It helps that the Java code is just extracting information and outputting a giant XML stream, and then I use command-line tools and Perl to do the rest of the analysis and summarisation |
| 18:50 | <Philip`> | jgraham: That would be equally painful in any language :-) |
| 18:50 | <Philip`> | because I'd have to work out what you actually want, and then how to extract it from a document tree, and then how to present it, and all of those things are language-independent |
| 19:22 | <jgraham> | Hmm, Jyhton seems to have the options of a stupidly out of date table version or a reasonably up to date beta :( |
| 19:22 | <jgraham> | s//s/ |
| 19:23 | <Philip`> | The beta worked for me |
| 19:23 | <Philip`> | Actually it was an alpha when I used it |
| 19:25 | <Philip`> | I found it worked when I basically wrote something like |
| 19:25 | <Philip`> | from nu.validator import htmlparser; from org.xml.sax import InputSource; from org.xml.sax.helpers import DefaultHandler |
| 19:25 | <Philip`> | class TestContentHandler (DefaultHandler): def startElement(self, uri, localName, qName, attributes): ... |
| 19:25 | <Philip`> | parser = htmlparser.sax.HtmlParser(htmlparser.common.XmlViolationPolicy.ALLOW); parser.setContentHandler(TestContentHandler()); parser.parse(InputSource(java.io.ByteArrayInputStream(body))) |
| 19:25 | <Philip`> | so it's not too complex really |
| 19:26 | <Philip`> | (Uh, and I think 'body' was a byte[] read from disk somehow) |
| 19:28 | <jgraham> | Oh well if it is stable, maybe I will try it |
| 19:29 | gsnedders | wonders whether grepping over the entire moz-central is sane |
| 19:29 | <jgraham> | gsnedders: No |
| 19:30 | jgraham | wonders why gsnedders is trying to do that |
| 19:30 | <Philip`> | Handy hint: Don't use grep in a utf8 locale |
| 19:30 | <Philip`> | (It's really slow) |
| 19:30 | <jgraham> | Handy hint:mxr.mozilla.org |
| 19:30 | Philip` | therefore uses 'LANG=C grep', which is something stupid like a hundred times faster |
| 19:30 | gsnedders | follows Philip`'s advice of how to be stupid quickly |
| 21:45 | gsnedders | wonders if it is a bad idea to go to the theatre by himself |
| 21:49 | <jgraham> | gsnedders: Why would that be a bad idea? |
| 21:49 | <gsnedders> | jgraham: geeky loneliness? |
| 21:49 | <jgraham> | Nah. Just make sure you are watching Waiting for Godot or something |
| 21:50 | <jgraham> | Not Mama Mia! |
| 21:50 | <gsnedders> | jgraham: I was thinking more Avenue Q |
| 21:50 | <jgraham> | But then watching Mama Mia! is a bd idea in general |
| 21:50 | <gsnedders> | Hixie blatantly has far too much influence over me :P |
| 21:51 | <jgraham> | gsnedders: Well I guess that isn't as sutiable for watching in the singular as Beckett but I think it will be better than not seeing it at all |
| 21:52 | <jgraham> | (Not that I have seen it or anything) |
| 21:52 | <gsnedders> | Unless, of course, anyone is in London in the week-beginning 22nd June… |
| 21:52 | <jgraham> | gsnedders: Sorry |
| 21:53 | <jgraham> | Do you know where you are staying in Linkoping yet? |
| 21:53 | <gsnedders> | I got email about one apartment giving me some details about that, yet to get any reply to my reply |
| 21:54 | <gsnedders> | (I emailed her on Monday) |
| 21:55 | <jgraham> | You spoke to Bibbi then? The mysteries of Swedish letting are still, well, mysterious to me |
| 21:55 | <jgraham> | Since our first apartment was found for us and the second one we found very luckily |
| 21:55 | <gsnedders> | Bibbi was finding one for me |
| 21:56 | <jgraham> | Yeah, that makes sense |
| 21:56 | <gsnedders> | Sent me some information about one with photos, and I basically replied saying that it'd be all right, and yet to get a response from her |
| 21:57 | <jgraham> | She has been away a lot this week. And I guess since you said it will be OK it doesn't need any action |
| 21:57 | <gsnedders> | I probably also better email her to make sure contract arrived |
| 23:59 | <Hixie> | man, i didn't notice the whatwg list had gotten so popular |