| 05:40 | <Hixie> | annevk: the methods on HTMLFormElement are different, that's one reason i got rid of the others |
| 07:58 | Hixie | does some cleanup in the comment section at the bottom of the spec |
| 07:58 | <Hixie> | it was getting out of hands |
| 07:58 | <Hixie> | there's 2000 lines of nots there |
| 07:58 | <Hixie> | notes, even |
| 08:42 | <hendry> | what is the right content type for javascript? i've seen application/x-javascript (apache) application/javascript (somewhere) text/javascript (most common) |
| 08:44 | <hsivonen> | hendry: application/x-javascript on the HTTP layer and text/javascript on the HTML layer |
| 08:44 | <Hixie> | use text/javascript |
| 08:45 | <Hixie> | hsivonen: bjoern registered the types, so there are actual non x- types that one can use now |
| 08:45 | <annevk> | Hixie, oops, guess I should actually check WF2 first before making assumptions |
| 08:45 | <hsivonen> | Hixie: do they work without problems? |
| 08:45 | <Hixie> | text/javascript is the only type i know that reliably works without problems |
| 08:46 | <annevk> | except in SVG |
| 08:46 | <annevk> | (server side types are ignored for <script>, fwiw) |
| 08:47 | <hsivonen> | annevk: does SVG work without an explicit type |
| 08:48 | <hsivonen> | on the HTML side, the type is useless |
| 08:48 | <annevk> | I think it does on the SVG side, but I'm not completely sure |
| 08:49 | <annevk> | lol: http://twitter.com/diveintomark/statuses/921518773 |
| 08:51 | <annevk> | just in: http://www.webfoundation.org/ |
| 08:51 | <annevk> | (from the W3C) |
| 08:53 | <hsivonen> | does this mean the funding structure now yields different incentives? |
| 08:53 | <hsivonen> | what does this mean to WS-*? |
| 08:54 | <hsivonen> | the mobile webs? |
| 08:54 | <hsivonen> | the Web? |
| 08:54 | <Hixie> | christ, the mission is even vaguer than the w3c's |
| 08:59 | <Hixie> | i'm amused that the wf's faq's answers are so vague and talking-point-y, given that they could pick the questions |
| 09:00 | <hsivonen> | so is the W3C transforming into this foundation or is this something that will be separate from the W3C? |
| 09:01 | hsivonen | tries to continue to read the FAQ |
| 09:01 | <annevk> | I believe it is a separate entity |
| 09:02 | <annevk> | but the FAQ indeed doesn't say much |
| 09:02 | <hsivonen> | OK. I got the wrong idea from the personnel overlap then |
| 09:02 | <othermaciej> | it's unclear what the purpose is |
| 09:02 | <othermaciej> | besides to collect donations |
| 09:03 | <othermaciej> | http://www.webfoundation.org/about/concept2008 |
| 09:03 | <othermaciej> | there used to be a WebFoundation.framework in the early days of WebKit development |
| 09:04 | hsivonen | finally finds "It was preferable not to disturb the W3C ecosystem for building consensus around Web standards. " |
| 09:05 | <hsivonen> | it might have been a good idea to put the question about the relationship with the W3C at the top... |
| 09:05 | <hsivonen> | I'm now less excited than I was 10 minutes ago |
| 09:07 | <othermaciej> | Tim is becoming quite the nonprofit entrepreneur |
| 09:31 | <hsivonen> | Hixie: are you not annotating changes to WF2 compared to what Opera implemented? |
| 09:31 | <Hixie> | hsivonen: i haven't tested what opera does very carefully |
| 09:32 | <Hixie> | but mostly the only changes are cutting stuff out |
| 09:32 | <hsivonen> | ok |
| 09:32 | <annevk> | it would be nice if you kept a small file somewhere that lists the changes |
| 09:32 | <annevk> | I can figure them out for myself, but it would make things easier |
| 09:35 | <Hixie> | does anyone have a list of the changes so far? |
| 09:35 | <annevk> | no XML form submission, repetition templates, form="" takes a single ID |
| 09:35 | <annevk> | you might have dropped external files for <datalist> and <select> |
| 09:36 | <Hixie> | (i'm basically just treating wf2 as part of html4 and using the same approach i do normally, btw, which is why i haven't really considered "changes" from wf2) |
| 09:36 | <hsivonen> | are all html5lib users on this channel using lxml.etree as the tree impl? |
| 09:36 | <annevk> | (for a few features it's not really clear) |
| 09:37 | <annevk> | Hixie, I guess that's fair enough, WF2 was pretty big |
| 09:37 | <annevk> | (in terms of features) |
| 09:38 | <jgraham> | hsivonen: I do, usually |
| 09:39 | <Hixie> | annevk: http://www.whatwg.org/specs/web-apps/current-work/notes has a brief list |
| 09:40 | <Hixie> | annevk: i'll try to keep it updated, but remind me occasionally |
| 09:40 | <annevk> | thx |
| 09:40 | <hsivonen> | I'm wondering if I should distinguish methods by name and not only by signature |
| 09:40 | <hsivonen> | that would be ugly in Java |
| 09:40 | <hsivonen> | but it seems it would facilitate subclassing in Jython |
| 09:42 | <Philip`> | hsivonen: I am |
| 09:47 | <hsivonen> | Philip`: what did you do about tree builders under Jython? would you have wanted an elementtree tree builder/ |
| 09:47 | <hsivonen> | ? |
| 09:49 | <hsivonen> | Re: 2022: Wine took 15 years to go to 1.0 |
| 09:50 | <Philip`> | hsivonen: My Jython code has no chance of ever running under CPython, so I don't care about having compatible interfaces; but I haven't written very much at all, and have only used SAX so far |
| 09:51 | <Philip`> | (And I have no desire to run my CPython code under Jython) |
| 09:53 | <hsivonen> | annevk: re: your blog entry with comments disabled: I've been thinking about blogging about the new concept of a REC not being particularly useful |
| 09:54 | <hsivonen> | annevk: what's the point of getting the particular feature set of CSS 2.1 to REC as one document when by the time it is a REC, both authors and implementations will be doing CSS 2.1 plus some CSS3 bits |
| 09:54 | <annevk> | hsivonen, yeah, agreed, let me know when you post it and I'll add a link |
| 09:54 | <hsivonen> | it seems to me that maturity should be tracked per-feature or something |
| 09:55 | <annevk> | that's what the HTML5 annotation system does |
| 09:55 | <hsivonen> | (having a lot of tiny specs doesn't really work with the intertwingliness) |
| 09:55 | <hsivonen> | annevk: right |
| 10:10 | <Hixie> | haha: |
| 10:10 | <Hixie> | “I’m really looking forward to the work we’re starting to do to ramp up building a test suite in the HTML Working Group,” says Microsoft Internet Explorer platform architect and WHAT WG co-chair Chris Wilson in an e-mail. |
| 10:11 | <Hixie> | -- http://www.webmonkey.com/blog/How_HTML_5_Is_Already_Changing_the_Web |
| 10:11 | <Hixie> | (laughing at "WHAT WG co-chair") |
| 10:12 | <roc_> | be nice |
| 10:12 | <Hixie> | to whom? |
| 10:13 | <Hixie> | i wasn't laughing at chris, i was laughing at the confusion we've managed to cause around the html5 process |
| 10:13 | <roc_> | oh righg |
| 10:13 | <roc_> | that passed me by |
| 10:31 | <takkaria> | that article is pretty impressively misleading :) |
| 10:33 | <Philip`> | Their comment-truncating system is really very irritating |
| 10:41 | <takkaria> | http://news.bbc.co.uk/1/hi/technology/7613201.stm -- hmm |
| 10:44 | <hsivonen> | so what kind of certificate will foxnews.com get? |
| 10:48 | <roc_> | yeah, that way is perilous |
| 10:59 | <Philip`> | It doesn't sound like you'd get a certificate in that way - '"I'm not a fan of giving a website a simple number like an IQ rating because like people they can vary in all kinds of different ways," he said. "So I'd be interested in different organisations labelling websites in different ways".' |
| 11:00 | <roc_> | isn't that PICS all over again? |
| 11:00 | <Philip`> | Maybe you would choose a friend's RDF feed of site-trustworthiness scores and it would transitively compute the trustworthiness of the whole web on your own computer, so you can choose whose prejudices to be exposed to |
| 11:01 | <Philip`> | roc_: Hmm, I assumed it meant organisations labelling other people's websites, not their own |
| 11:01 | <roc_> | I thought PICS let organisations label other people's websites |
| 11:01 | <Philip`> | Oh, okay |
| 11:02 | <othermaciej> | I wish Tim could be proud of the beautiful chaos of the human-readable Web he created |
| 11:02 | <othermaciej> | and stop chasing pipe dreams of a non-useless machine-readable Web |
| 11:04 | <Philip`> | Hmph, HTML5 doesn't acknowledge the cabal any more :-( |
| 11:04 | <Hixie> | wow, you guys look at the diffs closely |
| 11:05 | <Hixie> | i was trying to smuggle that change in :-P |
| 11:05 | <annevk> | Hixie, "XXX * find object at (x,y)" is in http://dev.w3.org/csswg/cssom-view/ |
| 11:05 | <othermaciej> | will any cabal members publicly reveal themselves enough to object? |
| 11:05 | <annevk> | Hixie, Pillar and Hedral are your cats? |
| 11:05 | <Hixie> | (the cabal it used to acknowledge wasn't the one that people thought it was, and has been disbanded for years, so i figured i should drop it) |
| 11:06 | <Hixie> | annevk: what's the status of cssom-view? |
| 11:06 | <annevk> | Hixie, closer to REC than HTML5 :p |
| 11:06 | <Hixie> | (yes, Miss Cat Adorable Pillar and Sir Cat Hedral) |
| 11:06 | <Hixie> | annevk: no new features? :-) |
| 11:06 | <Hixie> | i didn't realise this stuff was that interoperable |
| 11:07 | <annevk> | it has a few new features that are not shipping yet |
| 11:08 | <othermaciej> | you knighted your cat? |
| 11:08 | <othermaciej> | did you give him the accolate with a catnip chew toy? |
| 11:08 | <othermaciej> | *accolade |
| 11:08 | <othermaciej> | (dammit) |
| 11:12 | <Hixie> | othermaciej: q.v. ts elliot |
| 11:12 | <Hixie> | annevk: so what's your trick for being nearer to REC? :-) |
| 11:13 | <annevk> | less features ;) |
| 11:13 | <annevk> | we'll see I guess :) |
| 11:16 | <Hixie> | :-P |
| 11:18 | Philip` | discovers that HTML5 allows <ul></ul>, whereas HTML4 didn't (and requires at least one <li>) |
| 11:18 | <Philip`> | which solves a problem I once had when doing templated HTML generation with possibly-zero-length lists, which is nice |
| 11:19 | <virtuelv> | Philip`: How do UA's apply default style to the empty ul? |
| 11:19 | <virtuelv> | s/'s/s/ |
| 11:22 | <Philip`> | virtuelv: No idea |
| 11:23 | <Hixie> | virtuelv: probably just 1em top-and-bottom collapsed margin |
| 11:24 | <Hixie> | i love how much press we've been getting just because of the 2022 date |
| 11:24 | <Hixie> | it's awesome |
| 11:25 | <Hixie> | i think this is only the second most amount of press we've gotten, after the checkin where i changed the requirement of ogg support for a note that said we needed a common codec |
| 11:25 | <othermaciej> | HTML5 won't support Ogg until 2022! |
| 11:25 | <Hixie> | by 2022 it'll probably require h.264! |
| 11:26 | <hsivonen> | Hixie: not so awesome if it makes people want HTML4+foo validation instead of HTML5 validation |
| 11:26 | <Hixie> | why would it make people want foo? |
| 11:26 | <othermaciej> | I like the counter-article of "HTML5 is improving the Web already" |
| 11:27 | <Hixie> | the comments on most of these articles is pretty balanced |
| 11:27 | <Hixie> | and the body of most of them, despite the sensationalist titles, is pretty fair too |
| 11:27 | <hsivonen> | Hixie: assume foo has the value ARIA |
| 11:28 | <hsivonen> | Hixie: then assume you are WordPress with Transitional templates and lots of talk about standards compliance |
| 11:28 | <annevk> | Hixie, fwiw, the XHTML2 and Froms WGs consider XHTML2 + XForms to be suitable for Web applications |
| 11:28 | <hsivonen> | Hixie: then you want to add some ARIA stuff |
| 11:28 | <hsivonen> | Hixie: what do you do about validation |
| 11:28 | <Hixie> | annevk: i'm sure they do |
| 11:28 | <annevk> | Hixie, I'm not sure if the HTML 5 draft should make judgements about that |
| 11:28 | <hsivonen> | Hixie: should authors have to choose between either having ARIA and <video> or having border=0 |
| 11:28 | <othermaciej> | "Valid XHTML 1.0!" |
| 11:28 | <hsivonen> | ? |
| 11:29 | <othermaciej> | is what WordPress seems to want |
| 11:29 | <Hixie> | annevk: how me a real time strategy game written in xhtml2+xforms whose author says that xhtml2+xforms was a language that was satisfactorily designed for that purpose and i'll change the text |
| 11:29 | <hsivonen> | othermaciej: well, XHTML 1.0 and HTML 4.01 are pretty much the same thing for my purposes |
| 11:30 | <Hixie> | annevk: (or, if the xhtml2 and xforms wgs are willing to provide alternative text for those sections that is still accurate, i'll change it to that -- we're still waiting for them to come back with text for the xhtml 1.1 section) |
| 11:30 | <othermaciej> | I am just saying, their idea of standards compliance is already broken |
| 11:30 | <Hixie> | hsivonen: is this hypothetical or has it happened? |
| 11:30 | <Hixie> | hsivonen: i haven't seen people asking for aria validation on any of these articles |
| 11:30 | <Hixie> | hsivonen: and wordpress doesn't seem complicated enough to warrant needing aria |
| 11:31 | <hsivonen> | Hixie: I have had a question about validating ARIA in the context of WP specifically |
| 11:31 | <hsivonen> | Hixie: I have had an inquiry about HTML4+ARIA validation |
| 11:32 | <hsivonen> | Hixie: I get bug reports about HTML4 support |
| 11:32 | <hsivonen> | Hixie: and I've gotten a non-ARIA inquiry about maintenance/development of the HTML 4 side |
| 11:33 | <hsivonen> | Hixie: I wish the practical answer I could give was "Use HTML5 validation" |
| 11:33 | <Hixie> | why isn't that a practical answer? i mean, other than the fact that html5 is still in flux |
| 11:34 | <hsivonen> | Hixie: border=0, cellspacing=0, headers=foo, language=JavaScript |
| 11:34 | <Hixie> | why isn't my proposal for those satifactory? |
| 11:34 | <Hixie> | satisfactory even |
| 11:35 | <Hixie> | (also, i thought headers=foo was resolved, what about it is a problem?) |
| 11:35 | <othermaciej> | hsivonen: that seems more about specific migration tax issues (possibly fixable) than about HTML5 not being done for a while |
| 11:35 | <hsivonen> | Hixie: do you mean saying that the document is non-conforming but collapsing unimportant errors? |
| 11:35 | <Hixie> | yes |
| 11:35 | <hsivonen> | If stuff is unimportant, why say anything? |
| 11:36 | <hsivonen> | why develop more UI for stuff that is less important? |
| 11:36 | <annevk> | because people editing the document later might think it is significant |
| 11:36 | <Hixie> | because it still wastes bytes, and because if you're not transitioning from a legacy, it's valuable to not have it as it eases maintenance |
| 11:36 | <annevk> | and might try tweaking it, etc. |
| 11:36 | <Hixie> | i.e. it _is_ important |
| 11:36 | <Hixie> | just not interesting in certain cases (namely, editing legacy documents) |
| 11:37 | <othermaciej> | might this be something that is better classed as a warning rather than a conformance failure? |
| 11:37 | <othermaciej> | what is the philosophy behind what kind of arguable errors make a document non-conforming? |
| 11:37 | <Hixie> | othermaciej: not sure what that would mean. html5 either allows things or doesn't. |
| 11:37 | <Philip`> | I don't know about RTSs, but HTML5 is not satisfactorily designed for writing FPSs :-( |
| 11:37 | <hsivonen> | othermaciej: perhaps. but in principle, I can't do that without forking the conformance definition |
| 11:38 | <othermaciej> | I mean you get a diagnostic without a conformance failure |
| 11:38 | <Philip`> | (Well, I know about RTSs, but not about ones written in HTML) |
| 11:38 | <Hixie> | othermaciej: i don't think it is a good idea to make effect-free attributes conforming. |
| 11:38 | <othermaciej> | in other words, if the primary reason to disallow "border=0" is wasted bytes in non-legacy documents but it makes validating legacy documents needlessly painful, then disallowing it seems on net unhelpful |
| 11:39 | <Hixie> | Philip`: yup, there are lots of things that html5 can't do yet :-) |
| 11:39 | <othermaciej> | I would point out that /> is effect-free in HTML |
| 11:39 | <hsivonen> | Hixie: you already make effect-free stuff conforming |
| 11:39 | <othermaciej> | as is the xhtml namespace talisman |
| 11:39 | Hixie | doesn't like those either |
| 11:39 | <hsivonen> | type=text/css |
| 11:39 | <Hixie> | we need type="" |
| 11:40 | <othermaciej> | I would say effect-free things should be a conformance failure if they are more likely to be an error on the part of the author than to be legacy migration stuff or otherwise present for reasonable reasons |
| 11:40 | <Hixie> | making type="text/css" invalid would make maintenance more complicated |
| 11:40 | <othermaciej> | (granting arguendo that XML/HTML chameleon document is a reasonable reason) |
| 11:40 | <Hixie> | othermaciej: if i write a new document, i shouldn't be using border=0 and shouldn't be allowed to do so |
| 11:41 | <Hixie> | othermaciej: and thus it should be an error |
| 11:41 | <hsivonen> | Hixie: how realistic is it that there will ever be a style language other than CSS? |
| 11:41 | <othermaciej> | but if I have an existing HTML4 document that uses it, should I have to rewrite it if I want to move to HTML5? |
| 11:41 | <Hixie> | othermaciej: having all these attributes around is a net loss, because there are far more new documents than old |
| 11:41 | <othermaciej> | isn't that basically the same as the XHTML migration story, though not as severe? |
| 11:41 | <othermaciej> | a warning that doesn't break conformance would serve the purpose of discouragement just as well |
| 11:41 | <hsivonen> | Hixie: during the transition period, there are definitely more old documents |
| 11:41 | <Hixie> | othermaciej: if you have an HTML4 Transitional document, you should have Transitioned already |
| 11:42 | <Hixie> | othermaciej: after all, HTML4 Transitional has been known as Transitional for over 10 years |
| 11:42 | <hsivonen> | Hixie: it's not about 10 years |
| 11:42 | <Hixie> | hsivonen: we'll never get out of the transition period if we don't stop allowing it |
| 11:42 | <othermaciej> | perhaps it should have been called HTML4 Non-Fascist |
| 11:42 | <hsivonen> | Hixie: IE is still here |
| 11:43 | <othermaciej> | making things a conformance error to encourage authors to save a few bytes seems like attacking a flea with a nuclear warhead |
| 11:43 | <Hixie> | i don't understand why you think it should be an error if the user types <p asdls=""> but not if the u |
| 11:43 | <othermaciej> | I mean, yes, it is better to save a few bytes by avoiding things that have no effect |
| 11:43 | <Hixie> | er |
| 11:44 | <hsivonen> | I should point out that I also have a serious feature request for alerting people if they are saving bytes by omitting tags |
| 11:44 | <othermaciej> | but it's not the same as a syntax error |
| 11:44 | <Hixie> | i don't understand why you think it should be an error if the author types <p asdls=""> but not if the author types <img border=0> |
| 11:44 | <hsivonen> | Hixie: because asdls hasn't existed historically but border has |
| 11:44 | <annevk> | because asdis might mean something in the future, border=0 will always do the same due to legacy reasons |
| 11:44 | <othermaciej> | asdls="" is unlikely to exist in many existing documents, and certainly in none that care about validation |
| 11:44 | <Hixie> | border=0 has been deprecated for over a decade |
| 11:44 | <hsivonen> | Hixie: and border is still needed for IE and Gecko |
| 11:45 | <Hixie> | and isn't needed anywhere |
| 11:45 | <othermaciej> | and asdls="" is more likely to be a typo or other mistake than intentional |
| 11:45 | <hsivonen> | Hixie: the decade doesn't matter as long as IE and Gecko do what they do |
| 11:45 | <othermaciej> | while border=0 is far more likely to be legacy than error |
| 11:45 | <annevk> | img { border:none } solves that though |
| 11:45 | <Hixie> | img { border: none; } has been supported in IE since at least IE4 |
| 11:45 | <hsivonen> | Hixie: you can start counting once both IE and Gecko have purged the default border |
| 11:45 | <Hixie> | and has been supported in Gecko since around M6 if not before |
| 11:45 | <hsivonen> | Hixie: not copy/paste friendly |
| 11:46 | <Hixie> | oh please |
| 11:46 | <Philip`> | <img style="border:none"> |
| 11:46 | <hsivonen> | Hixie: copy/paste is a real use case |
| 11:46 | <hsivonen> | Hixie: c.f. random badges |
| 11:46 | <hsivonen> | Philip`: that's along the lines of what CC does |
| 11:46 | <hsivonen> | Philip`: not really an improvement |
| 11:46 | <Hixie> | as Philip` points out, if copy/paste is THAT important to these authors but they can't add one line to their CSS, they can use style="" |
| 11:47 | <Hixie> | but even that is dumb |
| 11:47 | <hsivonen> | Hixie: is that an improvement? |
| 11:47 | <hsivonen> | Hixie: of the order of magnitude that I should bother countless of people about it? |
| 11:47 | <Hixie> | what they SHOULD be doing is making the badges have whatever semantics they want and then style the markup using CSS to get whatever look they want |
| 11:48 | <Philip`> | <style> isn't reader-friendly either - you can see an <img> when reading the page source, but you have no idea what's causing it to have no border, and it's pretty hard to work backwards to find the relevant selector |
| 11:48 | <othermaciej> | is replacing border=0 with style="border:none" a useful way to spend one's time when upgrading to HTML5? |
| 11:48 | <Hixie> | html4 deprecated these misfeatures literally a decade ago, we're not doing anyone a favour by coddling them another 20 years |
| 11:49 | Philip` | wonders when someone will tie 'view source' into a Firebug-like tool |
| 11:49 | <Philip`> | (assuming they haven't already) |
| 11:49 | <Hixie> | othermaciej: no, which is why i think that all these error messages should be collapsed into one message of the ilk of "well you aren't compliant, but the only problems you have left are with old deprecated markup. More details..." |
| 11:49 | <othermaciej> | your desire to make these conformance errors seems to be based on a moralistic impulse against them rather than practical benefit for authors or users |
| 11:50 | <othermaciej> | Hixie: you know as well as I that "you aren't compliant" is not a state that anyone who cares about validation in the first place will want to stop at |
| 11:50 | <othermaciej> | if that's the output, they will waste a bunch of time or stop caring about the validator |
| 11:50 | <othermaciej> | (or both) |
| 11:50 | <Hixie> | the practical benefits are improved maintainability, improved authoring skill and understanding, and cleaner markup going forward |
| 11:50 | <hsivonen> | othermaciej: right, which is why it's pointless to develop UI for that |
| 11:50 | <othermaciej> | Hixie: isn't that the same arguments used for removing <b> and <i>? |
| 11:50 | <othermaciej> | "cleaner markup going forward" |
| 11:51 | <othermaciej> | "we have to fix all this broken stuff on the Web" |
| 11:51 | <annevk> | othermaciej, <b> and <i> are needed because they don't have replacements |
| 11:51 | <hsivonen> | Hixie: is it the validator's job to tell people how they can make their stuff more maintainable before they know they want it? |
| 11:51 | <Hixie> | i have removed <b> and <i>, in the form they had in html4 |
| 11:51 | <hsivonen> | annevk: <span style='font-style: italic;'> |
| 11:51 | <othermaciej> | annevk: <span style="font-weight: bold"> |
| 11:51 | <Philip`> | annevk: <strong>, <em> |
| 11:52 | <hsivonen> | annevk: if <img style='border: 0'> is a replacement |
| 11:52 | <annevk> | Philip`, ouch |
| 11:52 | <othermaciej> | but of course, that cure is worse than the alleged disease |
| 11:52 | <annevk> | <i> can be used for ship names and things typically rendered italic |
| 11:52 | <Hixie> | hsivonen: sure |
| 11:52 | <Hixie> | hsivonen: that's pretty much all a validator does |
| 11:52 | <Hixie> | hsivonen: that, and point out errors the author did know about but missed when editing the document |
| 11:53 | <hsivonen> | Hixie: I thought the job of a validator was closer to a spell checker than to a prescriptive grammarian |
| 11:53 | <Hixie> | hsivonen: word processors have both |
| 11:54 | <othermaciej> | "(machine-checkably) nonconforming" should be a question of syntax, not style |
| 11:54 | Philip` | would like it if the validator acted like a spell-checker and gave you a list of suggested replacements and you could click on them to apply that replacement |
| 11:54 | <annevk> | I think I agree that finally removing a bunch of these attributes is a good thing |
| 11:54 | <othermaciej> | style is a matter of taste and therefore many tools could have different approaches |
| 11:54 | <annevk> | hsivonen, did people complain about them? |
| 11:54 | <annevk> | hsivonen, or are you just worried by survey figures? |
| 11:55 | <annevk> | s/just// |
| 11:55 | <hsivonen> | annevk: people are asking me to put more effort into the HTML4 code so that authors wouldn't need to make the jump to validating as HTML5 |
| 11:55 | <Hixie> | i'm all for helping with this transition period, but i won't help with the transition period at the cost of not transitioning at all |
| 11:56 | <Hixie> | this means we can't do what html4 did (we know it doesn't work) |
| 11:56 | <annevk> | hsivonen, do you have more detailed information than that? |
| 11:56 | <hsivonen> | Hixie: I'm OK with killing <font face>, axis='' and <basefont> |
| 11:56 | <hsivonen> | annevk: no |
| 11:57 | <Hixie> | hsivonen: what happens if you only support HTML4 Strict? |
| 11:57 | <hsivonen> | annevk: this is like the usual WHATWG requirement problem: people want stuff without specifying exactly why |
| 11:57 | <Hixie> | what would happen, even |
| 11:57 | <hsivonen> | Hixie: I don't know. |
| 11:58 | <Hixie> | might be worth trying to implement HTML4 Transitional as just saying "Your document is not valid HTML4 Strict, because it uses features only available in the deprecated Transitional syntax." or some such |
| 11:58 | <Hixie> | instead of "Valid HTML4 Transitional" |
| 11:59 | <Hixie> | anyway, i'm open to ideas other than making these things fully conforming |
| 11:59 | <hsivonen> | Hixie: the purpose of the HTML 4 features of Validator.nu is to act as a gateway drug for the service--not to tell people to go back to DTD-based validator |
| 12:00 | <hsivonen> | Hixie: one possibility is to make them conditionally non-conforming with the condition being non-machine-checkable :-) |
| 12:00 | <hsivonen> | Hixie: as with layout tables :-) |
| 12:01 | <hsivonen> | And no, I don't have a good suggestion. |
| 12:01 | <hsivonen> | but I don't like the way things are now drafted |
| 12:02 | <Hixie> | there are lots of things i don't like in html5 |
| 12:02 | <annevk> | hsivonen, but maybe we can figure it out from the documents they want to use HTML4 validation for? |
| 12:02 | <Hixie> | the /> talisman for instance |
| 12:02 | <hsivonen> | annevk: do you mean logging everything and trying to guess stuff? |
| 12:02 | <annevk> | hsivonen, I was hoping they would give examples of things that don't work |
| 12:03 | <annevk> | hsivonen, and if that's the case, doing what Hixie suggested for HTML4 Transitional might not be so bad |
| 12:03 | <annevk> | (and XHTML1 Transitional for that matter) |
| 12:03 | <hsivonen> | annevk: one problem is that people stick to HTML 4 because HTML5 is not 'ready' |
| 12:03 | <Hixie> | to be honest i'm not convinced that most of the pages that use these attributes only use them in their default state |
| 12:03 | <hsivonen> | annevk: I'm myself telling people not to upgrade to HTML5 doctype just yet |
| 12:04 | <annevk> | hsivonen, right, that seems ok |
| 12:04 | <Hixie> | and if people use these attributes in non-default states, they'll have to do some edits anyway |
| 12:04 | <Hixie> | at which point, doing the rest isn't that much of a cost |
| 12:04 | <annevk> | hsivonen, I do encourage people to play with it and say that the HTML5 DOCTYPE just works, but overall waiting a little bit longer is safer |
| 12:04 | <hsivonen> | but once Firefox 3.1 ships with <video>, we need a better story than "don't validate because HTML5 is not ready" |
| 12:05 | <hsivonen> | and I don't know how to do that and avoid an Atom 0.3-like fiasco |
| 12:05 | <Hixie> | (and it's a cost i'd be willing to risk forcing people to take, given the benefits long-term) |
| 12:05 | <hsivonen> | Hixie: the whole point of border=0 is using it in a non-default state |
| 12:06 | <annevk> | hsivonen, Atom 0.3 fiasco won't happen given how we take into account implementations |
| 12:06 | <hsivonen> | Hixie: same for cellspacing and cellpadding |
| 12:06 | <annevk> | cellpading is not necessary |
| 12:06 | <hsivonen> | recommending HTML5 would be so much safer if there were two browsers out there with the parsing algorithm implemented |
| 12:07 | <annevk> | only cellspacing lacks the CSS equivalent in IE, iirc |
| 12:07 | <othermaciej> | or even just one |
| 12:07 | <Hixie> | hsivonen: when would you ever want to specify cellspacing and cellpadding on a data table? |
| 12:08 | <annevk> | hsivonen, the conforming syntax does not depend much on a conforming parser... |
| 12:08 | <hsivonen> | Hixie: you want the cell borders to touch each other but not text |
| 12:09 | <hsivonen> | Hixie: anyway, layout tables are a cowpath if there ever was one |
| 12:09 | <Hixie> | use border-collapse: collapse, you'll get much better control |
| 12:09 | <Hixie> | we don't want to pave cowpaths that go off a cliff |
| 12:09 | <hsivonen> | Hixie: does it work in IE? |
| 12:09 | Hixie | loads wmware |
| 12:09 | <Hixie> | vmware |
| 12:09 | <Hixie> | my typing tonight sucks, i should go to bed soon :-) |
| 12:10 | <hsivonen> | Hixie: so far, evidence suggests that layout tables are further away from the cliff than e.g. CSS positioning |
| 12:10 | <Hixie> | yeah well the csswg needs to get its act together, but that's a separate story |
| 12:10 | <Hixie> | css layout tables are fine, and even work in IE now |
| 12:10 | <hsivonen> | Hixie: there's no point in poking authors about it until the CSS WG and IE have their "act together" |
| 12:11 | <Hixie> | IE8 will be out before html5 is ready for authors to use in any serious way |
| 12:11 | <hsivonen> | Hixie: do they work in the oldest IE that Gmail still supports? |
| 12:11 | <Hixie> | (and yes, IE supports border-collapse) |
| 12:12 | <Hixie> | hsivonen: gmail only supported one version of IE until very recently where we added IE6 support, so gmail might not be the best example |
| 12:12 | <hsivonen> | Hixie: my point is that there's demand for ARIA validation today and for <video> validation RSN |
| 12:13 | <hsivonen> | Hixie: my point is that IE6 isn't properly dead if Gmail supports it |
| 12:14 | <annevk> | I have a hard time believing those features will be less successful because border=0 does not validate |
| 12:14 | <annevk> | <canvas> was successful and border=0 did not validate either |
| 12:14 | <Hixie> | if there really is demand, we shoudl leverage that demand to get people to move off border=0 and co |
| 12:14 | <hsivonen> | annevk: Validator.nu is at risk of being less successful |
| 12:15 | <hsivonen> | annevk: of course people will pick <video> over validation |
| 12:15 | <Hixie> | but frankly, i doubt <video> will be any more used than <canvas> in the near (12-18 months) term |
| 12:15 | <hsivonen> | annevk: but they shouldn't have to abandon QA tools in order to use <video> or <canvas> or ARIA |
| 12:15 | <Hixie> | if people are using the validator as a QA tool, then a collapsed set of error messages for deprecated attributes would be fine |
| 12:16 | <Hixie> | i thought we were assuming that people using the validator were people who wanted to be perfectly valid always |
| 12:16 | <hsivonen> | Hixie: no, I'm assuming they are people who want to find typos without the typos getting lost in noise |
| 12:17 | <Hixie> | collapsing all the messages into one removes all the noise and places it separately from the typos |
| 12:18 | <annevk> | Is it really hard to collapse the messages? It might at least be worth trying |
| 12:19 | <hsivonen> | annevk: it's not trivial. I'd rather do only non-trivial stuff I believe in |
| 12:19 | <roc_> | What are we doing wrong with table borders? |
| 12:20 | <hsivonen> | roc_: image borders |
| 12:20 | <hsivonen> | roc_: there's a default border on linked images |
| 12:20 | <roc_> | oh |
| 12:21 | <roc_> | there's a bug filed, I assume? |
| 12:21 | <hsivonen> | yes |
| 12:21 | <hsivonen> | looking up the # now |
| 12:21 | <hsivonen> | https://bugzilla.mozilla.org/show_bug.cgi?id=452915 |
| 12:21 | <annevk> | hsivonen, wouldn't it be as simple as grouping all attribute error messages for when the attribute name is in legacy-set? |
| 12:22 | <roc_> | thanks |
| 12:22 | <annevk> | hsivonen, http://hsivonen.iki.fi/test/moz/alt.html should test image maps |
| 12:23 | <hsivonen> | annevk: UI is non-trivial |
| 12:23 | <hsivonen> | annevk: the current grouping feature is non-trivial |
| 12:25 | <hsivonen> | annevk: OTOH, if the user isn't expected to be albe to actually toggle stuff in the UI, I might as well send the errors to the bit bucket right away without UI |
| 12:26 | <annevk> | if I had a legacy page I would probably want to know that I have them and where they are located |
| 12:40 | <aaronlev> | what's the latest thinking about math in html5? |
| 12:41 | <Hixie> | the spec says the latest thinking |
| 12:41 | <annevk> | that is, MathML is supported as part of the HTML syntax and as part of the HTML parser |
| 12:41 | <Hixie> | basically, there is a syntax for mathml in text/html defined, and people are expected to use mathml |
| 12:41 | <annevk> | hah |
| 12:45 | <Hixie> | ok bed time |
| 12:45 | <Hixie> | nn |
| 12:53 | <Philip`> | The spec also includes <img>, which you can use for maths |
| 12:55 | <annevk> | very accessible |
| 12:56 | <Philip`> | The spec also includes <img alt> |
| 12:56 | <annevk> | yes, but no text syntax for math |
| 12:57 | <Philip`> | So use LaTeX :-) |
| 12:57 | <Philip`> | (inside the <img alt>) |
| 12:57 | <Philip`> | (since that's what you'll be generating the image from anyway) |
| 12:58 | <hsivonen> | Philip`: which LaTeX packages to assume? |
| 12:59 | <Philip`> | hsivonen: Whichever ones your LaTeX-subset to image converter provides |
| 13:01 | <Philip`> | s/LaTeX-subset/(La)TeX-inspired language like Itex/ |
| 13:02 | <Philip`> | I don't know :-( |
| 13:03 | <hsivonen> | Philip`: so you don't expect a speech generator to grok LaTeX? |
| 13:07 | <Philip`> | hsivonen: It could understand enough of the syntax to assist the user in making sense of the code, e.g. by understanding {} for grouping, and then the human can work out what the symbol and command names mean |
| 13:08 | <Philip`> | That's probably substantially easier than making a speech generator grok MathML |
| 13:08 | <Philip`> | and MathML has the disadvantage that sighted readers can't understand the textual source code either |
| 13:09 | <hsivonen> | Philip`: MathPlayer's speech rendering seems to work |
| 13:40 | <Lachy> | for determining the encoding algorithm, it whitespace around the value of the charset attribute supposed to be ignored? e.g. charset=" ISO=8859=2 " |
| 13:41 | <Lachy> | the spec only says "If the attribute's name is "charset", let charset be the attribute's value, interpreted as a character encoding." |
| 13:41 | <Lachy> | Hixie, yt? |
| 13:42 | <takkaria> | http://www.whatwg.org/specs/web-apps/current-work/multipage/parsing.html#character1 |
| 13:42 | <takkaria> | suggests that you ignore space characters |
| 13:43 | <takkaria> | but maybe that's a sightly liberal interpretation of the spec |
| 13:43 | <Lachy> | thanks |
| 13:45 | <annevk> | it suggests to ignore more than just space characters, fwiw |
| 14:21 | <hsivonen> | Re: markp's "my precious" tweet: is the Chrome icon Free as in Freedom for use in non-Google builds of Chromium? |
| 14:30 | <Philip`> | hsivonen: Their SVN repository only contains chrome/app/theme/chromium, which has the blue Chromium logo (instead of the multicoloured Chrome logo), and it doesn't have the chrome/app/theme/google_chrome directory which is used if you compile with GOOGLE_CHROME_BUILD |
| 14:30 | <annevk> | did "this week in html5" die? |
| 14:30 | <Philip`> | so I assume the idea is that only Chrome should use the Chrome icon, and everyone else should use the Chromium icon |
| 14:31 | <Lachy> | Philip`, is it possible to browse the SVN online somewhere? I can't find it |
| 14:31 | <annevk> | sounds exactly the same as in Firefox |
| 14:31 | <annevk> | Lachy, http://src.chromium.org/ |
| 14:54 | <hsivonen> | Philip`: thanks |
| 14:58 | <hsivonen> | one might argue that losing color and having to use "ium" is less drastic than losing the fox and calling it a weasel :-) |
| 15:00 | <Philip`> | I expect the people who might care about this don't currently care about this, since Chrom* is nowhere near being able to run on Linux |
| 15:02 | <gavin> | there are non-weasel generic icons available to people who want to distribute non-branded firefox builds |
| 15:03 | <hsivonen> | gavin: that's what I meant by losing the fox [graphic] compared to merely losing color |
| 15:03 | <gavin> | oh |
| 15:04 | <gavin> | I haven't actually seen the chromium icon |
| 15:04 | Philip` | wonders why Chrome/Chromium bother changing the 'client' string when making requests to pages like http://safebrowsing.clients.google.com/safebrowsing/diagnostic?site=http://www.google.com&client=googlechrome |
| 15:04 | <gavin> | is colour really the only difference? |
| 15:04 | <hsivonen> | gavin: http://src.chromium.org/viewvc/chrome/trunk/src/chrome/app/theme/chromium/product_logo.png?view=markup |
| 15:04 | <gavin> | that seems like it might be a bit risky from a TM protection point of view |
| 15:04 | <hsivonen> | gavin: seems to be only a color difference |
| 15:04 | <gavin> | but IANAL! |
| 15:05 | <Philip`> | (Compare to http://www.google.com/chrome/intl/en/images/logo_sm.jpg) |
| 15:06 | hsivonen | notes that Safari and WebKit have a color difference |
| 15:07 | <hsivonen> | but I don't know if the WebKit version of the icon is Free |
| 15:09 | <hsivonen> | what does it mean that according to Google, www.google.com has functioned "as an intermediary for the infection of 2 site(s)"? |
| 15:12 | <gavin> | heh |
| 15:16 | <Philip`> | It's nice to see <blockquote> being used for indentation again |
| 15:17 | <Philip`> | Also nice is <img ... height=32width=78> |
| 15:31 | Philip` | sees that Chrome has unfair favouritism towards Google, by prefetching the DNS for www.google.com when you start the browser |
| 15:32 | <hsivonen> | Philip`: is that an explicit call or a side effect of something? |
| 15:32 | <gavin> | doesn't google chrome always ping google for something at startup? |
| 15:32 | <gavin> | update or omnibox or something like that |
| 15:32 | gavin | doesn't remember the details |
| 15:33 | <Philip`> | hsivonen: It's at the bottom of http://src.chromium.org/svn/trunk/src/chrome/browser/net/dns_global.cc |
| 15:34 | <Philip`> | gavin: I suppose it does that too, though I think you can switch off the send-everything-you-type-in-the-address-bar-to-Google feature if you want |
| 15:34 | <gavin> | yeah,I remember hearing that there was something that wasn'y yet disableable |
| 15:34 | <gavin> | but that they were working on making disableable |
| 15:34 | <gavin> | I just don't remember what it is |
| 15:35 | <Philip`> | There's the Safe Browsing thing too |
| 15:36 | <hsivonen> | Philip`: at least it looks like it's a fallback if there is no overriding config file |
| 16:05 | <hsivonen> | Hixie: your exact examples in http://bugzilla.validator.nu/show_bug.cgi?id=258 are now WFM, but I can reproduce the problem with an unclosed <i> |
| 16:05 | <hsivonen> | Hixie: Do you have a suggestion how to fix? |
| 16:06 | <Lachy> | in python, if I have an list of strings like ["a", "b", "c"], is there a method that will concatenate them all into a single string? |
| 16:06 | <Lachy> | with a separator in between |
| 16:07 | <hsivonen> | Hixie: I don't know how to fix without either making it a fatal error to reconstruct formatting elements or to be non-conforming and violate the parsing algorithm |
| 16:07 | <Philip`> | Lachy: "separator".join(["a", "b", "c"]) |
| 16:07 | <Lachy> | Philip`, thanks |
| 16:08 | <hsivonen> | Philip`: is http://bugzilla.validator.nu/show_bug.cgi?id=294 now fixed to your satisfaction, or should I vary the message somehow when the tokenizer is in an attribute that takes a URL? |
| 16:10 | <hsivonen> | I'd rather have a good catch-all message and not vary based on attribute identity |
| 16:14 | <Philip`> | hsivonen: It doesn't seem to be fixed at all - if I write e.g. <a href="foo.cgi?src=bar©=baz"> then it still says just "Error: Entity reference was not terminated by a semicolon." which isn't very friendly |
| 16:16 | <hsivonen> | hmm. I haven't fixed it all states |
| 17:08 | <hsivonen> | Philip`: is the bug *now* fixed to your satisfaction? |
| 17:15 | <Philip`> | hsivonen: No - for "<!DOCTYPE html>This page is © 2022" it should tell you to write © not &copy |
| 17:15 | <Philip`> | (though for <a href="...©=..."> it should suggest the other way around) |
| 17:17 | <Philip`> | (Well, technically, the bug is only talking about URL attributes, so the brokenness of plain-text "... © ..." is not relevant, but I still think it's a brokenness :-p ) |
| 17:21 | <Philip`> | Also it seems misleading for it to sound so sure about "Error: & should have been escaped as &." (not saying anything vague like "Probable cause") when I write e.g. "... &lamba; ..." and the real error is not the failure to escape & |
| 17:30 | <hsivonen> | Philip`: what about now? |
| 17:40 | <Philip`> | hsivonen: On &lamba;, "& did not start a character reference. (& should have been escaped as &.)" should still probably say "probably should" |
| 17:42 | <Philip`> | The phrasing of "Named character reference was not terminated by a semicolon. (Or & should have been escaped as &.)" sounds odd, but otherwise I think everything seems possibly reasonable :-) |
| 17:42 | <hsivonen> | Philip`: what would you suggest |
| 17:49 | <Philip`> | hsivonen: Would it be hard to make it say something like "Character reference did not end with semicolon. (Should have been written as `©` (©) or `&copy`)"? |
| 17:49 | <Philip`> | (where it extracts the character reference name from the source) |
| 17:49 | <hsivonen> | extracting it from the source would suck |
| 17:50 | <hsivonen> | but extracting it from the name table might work |
| 17:50 | <Philip`> | This is only in cases where it's successfully parsed, so I suppose it can come from the parse representation rather than the source |
| 18:03 | <Philip`> | (Please feel free to ignore me if I'm trying to make this far too complicated :-) ) |
| 19:58 | <BenMillard> | Anyone feel like reviewing another follow-up from me about the headers+id thread? |
| 20:03 | <BenMillard> | I've e-mailed it to a bunch of you, now I'm off to watch Fifth Gear. :) |
| 22:05 | gsnedders | yawns |
| 22:05 | <gsnedders> | I don't like writing a personal statement |
| 22:06 | <gsnedders> | I suck at sounding motivated to do physics |
| 22:08 | <Philip`> | Are you motivated to do physics? |
| 22:10 | <gsnedders> | Yeah. |
| 22:10 | <Philip`> | Good - that should make it somewhat easier :-) |
| 22:10 | <gsnedders> | :) |
| 22:10 | <gsnedders> | I still suck at knowing what to write |
| 22:11 | <gsnedders> | I can write stuff that makes me look awesome at comp.sci., but when 1/5th of the places your applying to are comp.sci., that isn't overly relevant |
| 22:11 | <gsnedders> | (the other 4/5th are phys.) |
| 22:12 | <Hixie> | hsivonen: yeah i realised a few weeks ago that making the parser rules normative for a validator might be counter-productive |
| 22:12 | <Hixie> | hsivonen: it might be that we should change the spec to allow validators to do whatever they like to recover once they hit one error in the parser |
| 22:26 | <Philip`> | http://www.codeweavers.com/services/ports/chromium/ |
| 22:27 | Philip` | sees that they're using the multicoloured Chrome logo |
| 22:31 | <hendry> | Philip`: i can't figure out what to run after installing it. |
| 22:32 | <Philip`> | hendry: I haven't even installed it, so I can't help :-( |
| 22:35 | <Lachy> | hendry, on which platform? |
| 22:37 | <Lachy> | woah, the icon they've used for the OSX icon is terrible. The resolution is far too low. |
| 22:38 | <Lachy> | looks like they've just stretched the 32px windows icon |
| 22:38 | <hendry> | Lachy: debian/linux |
| 22:39 | <Lachy> | there must be an executable somewhere. It's probably called "chromium" or something like that |
| 22:40 | <Lachy> | maybe you have to start it from the command line if there's nothing you can double click to launch |
| 22:41 | <hendry> | Lachy: there is http://www.flickr.com/photos/hendry/2860126641/ but I get "cannot execute binary file". I'll try in the morning after I've sobered up. |
| 22:42 | <Lachy> | hendry, what's the relevance of the flickr photo? |
| 22:43 | jgraham | strongly suspects a msipaste |
| 22:44 | <Lachy> | the OSX version doesn't integrate well with the system |
| 22:44 | <Hixie> | oh dear, svg 1.2 tiny is back in last call |
| 22:45 | <Lachy> | Hixie, did it drop back from CR, or progress from WD? |
| 22:45 | <Dashiva> | So there's tiny and basic, who implements full? |
| 22:46 | <hendry> | Lachy: oh crap, wrong link. i meant: /opt/cxchromium/support/chromium/drive_c/chromium/chrome.exe /me goes to sleep now |
| 22:48 | <Lachy> | well, that would be the windows executable. I'm guessing you need to use Wine to launch it somehow, though I've no idea how that works |
| 22:49 | <Hixie> | Lachy: progressed from LC, i think |
| 22:49 | <jgraham> | Running it under crossover seems like a bad way to get a fair first impression. I guess the animation and scrolling suck less on a native build |
| 22:51 | <Lachy> | it's definitely not good to run on the Mac. It's bad enough being like a windows up, but the integration problems make it largely unusable |
| 22:52 | <Lachy> | s/windows up/windows app/ |
| 22:52 | <jgraham> | Which integration problems in particular? |
| 22:52 | <Lachy> | Expose doesn't work |
| 22:53 | <Lachy> | scrolling is bad |
| 22:54 | <jgraham> | Yeah the scrolling is abysmal. Oh well, I guess that's a wine bug. |
| 22:55 | <Lachy> | when you drag tabs off the tab bar, it's quite slow and the transparency is buggered up |
| 22:56 | <Lachy> | if I didn't have Windows in a Virtual machine to run it in, it would be ok, but it's not really worth it, especially since I already have webkit available |
| 23:23 | <annevk> | yay http://blog.whatwg.org/this-week-in-html-5-episode-5 |
| 23:31 | <annevk> | http://twitter.com/brunsvold/statuses/922129928 |
