| 00:09 | <jruderman> | use firefox? |
| 00:09 | <anne-mac> | i guess |
| 00:10 | <Philip`> | Or IE? |
| 00:10 | <anne-mac> | i was trying to get away with using just Opera on my Mac but Google is against me |
| 00:11 | <anne-mac> | (on my other laptop I'm trying to get away with not having Flash, which works reasonably so far) |
| 00:11 | <anne-mac> | (but this one had it by default, so I guess I'm cheating) |
| 00:12 | <Philip`> | (You should get another one and set the firewall to block port 80, to complete your attempts at disabling significant portions of the web :-) ) |
| 00:13 | <anne-mac> | I'm not convinced that's quite the same |
| 00:13 | <anne-mac> | :) |
| 00:14 | Philip` | doesn't do much unusual except disabling referer, and has only noticed that breaking about one site in the past several months |
| 00:16 | <anne-mac> | "Nokia and Apple have successfully pressured the WC3 board into dropping Ogg support from the HTML5 spec. Sheesh." |
| 00:16 | <Hixie> | i'm on the WC3 board? sweet! |
| 00:16 | <Hixie> | what's the WC3 board? |
| 00:16 | <Philip`> | A dartboard? |
| 00:17 | <Philip`> | I'm not sure what Warcraft 3 has to do with this, though |
| 00:17 | <anne-mac> | holly crap |
| 00:18 | <anne-mac> | Krzysztof e-mails way too much |
| 00:18 | <Hixie> | i've seen worse |
| 00:18 | <Hixie> | but yeah |
| 00:19 | <Philip`> | Fortunately I'm protected from the onslaught by a 2.5 hour email lag again |
| 00:19 | <Hixie> | heh |
| 00:21 | <anne-mac> | hmm, WCAG 2.0 in LC |
| 00:37 | <MikeSmith> | Is there any kind of whatwg timeline/history at the whatwg wiki? |
| 00:42 | <Hixie> | yeah |
| 00:42 | <Hixie> | look on the faq |
| 00:42 | <Hixie> | it's one of the questions |
| 00:43 | <Hixie> | also, http://lists.w3.org/Archives/Public/www-archive/2006Nov/0045.html |
| 00:46 | <MikeSmith> | Hixie - thanks |
| 00:48 | <MikeSmith> | about timeline, what I was looking for was something that showed when whawg first published Web Forms 2.0, when first Web Apps was published |
| 00:48 | <MikeSmith> | that kind of stuf |
| 00:49 | <Hixie> | oh |
| 00:49 | <MikeSmith> | I been thinking about adding it to the timeline on the HTML WG home page |
| 00:49 | <Hixie> | first draft of webforms 2, then known as xforms basic, came out in october/november 2003, just after the xforms PR went to vote |
| 00:49 | <MikeSmith> | or at least linking to something |
| 00:49 | MikeSmith | nods |
| 00:50 | <MikeSmith> | how about first Web Apps? |
| 00:50 | <Hixie> | whatwg itself (along with first draft of wa1, now known as html5) came out the week after the adobe meeting |
| 00:50 | <Hixie> | march 2004? |
| 00:50 | <Hixie> | about then |
| 00:50 | <MikeSmith> | OK |
| 00:50 | <Hixie> | my blog commented on it all, if you want exact dates |
| 00:50 | <Lachy> | MikeSmith, some of that info is listed here http://www.whatwg.org/specs/ |
| 00:50 | <MikeSmith> | Lachy - thanks |
| 00:51 | <MikeSmith> | Hixie - yeah, I want to go back and read through your blog archive, dbaron's, some others |
| 00:51 | <anne-mac> | Web Forms 2.0 has all the dated links |
| 00:51 | <anne-mac> | in the draft |
| 00:51 | <MikeSmith> | anne-mac - thanks |
| 00:52 | <anne-mac> | first (W3C Member-only): http://lists.w3.org/Archives/Member/w3c-archive/2003Sep/att-0014/hfp.html |
| 00:52 | <MikeSmith> | I remember dbaron once posted a summary about his perspective on the adobe meeting and the results of it |
| 00:53 | <Hixie> | wow, september |
| 00:53 | <Hixie> | that's even earlier than i remember |
| 00:53 | <dbaron> | This one? http://dbaron.org/log/2004-06#e20040609a |
| 00:54 | <dbaron> | I revised my opinions a bit later on... http://dbaron.org/log/2006-08#e20060818a |
| 00:54 | <MikeSmith> | dbaron - yep, thanks much |
| 00:54 | <Hixie> | my early blog entries on the matter are quite amusing in retrospect |
| 00:54 | <Hixie> | since they mention the whatwg before it was announced |
| 00:54 | <Hixie> | but without saying so :_) |
| 00:54 | <MikeSmith> | dbaron - thanks, I remember reading that one too |
| 00:55 | <dbaron> | Hixie, but you did that intentionally... |
| 00:55 | <Hixie> | yes |
| 00:56 | <Hixie> | i am easily amused :-P |
| 00:58 | <MikeSmith> | I'm hoping to have some time later today to write up draft timeline of the events from the adobe workshop up until the announcement of the current HTML WG |
| 00:58 | <Lachy> | what was the adobe workshop? |
| 00:59 | <Hixie> | the web apps workship |
| 00:59 | <Hixie> | shop |
| 00:59 | <MikeSmith> | Lachy - http://www.w3.org/2004/04/webapps-cdf-ws/ |
| 01:00 | <Lachy> | ah, is that the one after which the whatwg formed? |
| 01:00 | <Hixie> | it's the one after which the whatwg was announced |
| 01:00 | <Hixie> | we actually had the whatwg ready to go at least a month earlier but we wanted to give the w3c a chance |
| 01:01 | <MikeSmith> | Lachy - dbaron blogged just after it at http://dbaron.org/log/2004-06#e20040607a |
| 01:01 | <MikeSmith> | Brendan Eich at http://weblogs.mozillazine.org/roadmap/archives/005632.html |
| 01:01 | <Hixie> | i also blogged many detailed blogs on it both during and after |
| 01:02 | <dbaron> | about a month and a half before, looks like |
| 01:02 | <Hixie> | based on registration date? |
| 01:02 | <Hixie> | of the domain? |
| 01:02 | <Hixie> | sounds about right |
| 01:03 | <MikeSmith> | I guess the following is the minutes of the workshop |
| 01:03 | <MikeSmith> | http://lists.w3.org/Archives/Public/public-webapps-cdf-discuss/2004Jun/att-0000/2004jun01.html |
| 01:03 | <dbaron> | Hixie, no, based on email threads I have |
| 01:04 | <MikeSmith> | or maybe above is just position statements, I dunno |
| 01:04 | <MikeSmith> | hmm, no, I guess that's the minutes |
| 01:05 | <Hixie> | dbaron: ah |
| 01:05 | <MikeSmith> | anyway, I will add whatever I come up with to the http://esw.w3.org/topic/HTML/ wiki |
| 01:06 | <MikeSmith> | but if somebody else has time and interest in doing it before I get to it, feel free |
| 01:06 | <MikeSmith> | either there or at the whatwg wiki |
| 01:07 | <MikeSmith> | I guess if we really wanted to be ambitious, it could go back to events since publication of HTML 4.0 or 4.01 rec |
| 01:08 | <Hixie> | there aren't many between that and wf2 :-P |
| 01:08 | <Hixie> | that was the problem :-P |
| 01:11 | <MikeSmith> | I guess showing that there weren't HTML milestones during those years could be another part of the usefulness of it |
| 01:11 | <MikeSmith> | or showing what else was going on instead |
| 01:17 | <Hixie> | heh |
| 01:18 | <Hixie> | wow, globalStorage is already on more sites than <t:video> |
| 01:19 | <Hixie> | including on http://maps.live.com/localsearch/ |
| 01:19 | <Hixie> | !! |
| 01:19 | <Hixie> | microsoft use html5! |
| 01:23 | <gavin> | "<i>What working group</i> is going to work on extending HTML..." :) |
| 01:25 | <gavin> | (from http://ln.hixie.ch/?start=1086387609&count=1 ) |
| 01:26 | <Hixie> | hey sweet, there's a book that mentions globalStorage: http://safari.adobepress.com/9780132242066 |
| 01:37 | <roc> | wow!!! |
| 01:37 | <othermaciej_> | Hixie: that's funny |
| 01:38 | <Hixie> | so it seems that we will break in the region of 200 sites by simplifying globalStorage |
| 01:39 | <Hixie> | we can work around one of the problems by just making globalStorage[domain] === globalStorage |
| 01:39 | <Hixie> | (just in firefox) |
| 01:39 | <Hixie> | but it seems that there is a bigger problem |
| 01:39 | <Hixie> | which is that people are using the .value accessor |
| 01:39 | <Hixie> | which i wasn't expecting |
| 02:15 | <Hixie> | <cite><audio>...</audio></cite> could be valid i guess... right...? |
| 08:05 | <hsivonen> | Hixie: what was "the Adobe meeting"? |
| 08:06 | <hsivonen> | canceling the question. noticed Lachy asked it too |
| 08:15 | <Hixie> | heh |
| 08:17 | <Hixie> | can anyone think of a use case for <address> <blockquote> ... ? |
| 08:18 | <hsivonen> | Hixie: indented contact info |
| 08:19 | <Hixie> | -_- |
| 08:20 | <Hixie> | <dialog> <dt> Fred <dd> <audio> makes sense |
| 08:20 | <Hixie> | but what about <dialog> <dt> <audio> ? |
| 08:22 | <Hixie> | i guess we should allow <sup><audio/></sup> though i'm hard pressed to come up with any valid use cases for that either |
| 08:23 | <hsivonen> | Hixie: what's the worse case scenario if we do allow those? |
| 08:23 | <Hixie> | people make mistakes that aren't caught by the validators, i think |
| 08:24 | <Hixie> | i really need a third option, "yes", "no", and "dumb" |
| 08:26 | Hixie | removes ins and del from the table since really they're orthogonal to the block/inline thing |
| 08:27 | <hsivonen> | what's the right way to abstract away createElement vs. createElementNS these days? |
| 08:28 | hsivonen | is reviewing and integrating zcorpan's message collapsing script |
| 08:32 | <Hixie> | abstract? |
| 08:38 | <hsivonen> | Hixie: zcorpan wrote a script for the validator.nu UI that collapses duplicate instances of the same error message |
| 08:38 | <hsivonen> | I want UI scripts to work with &out=xhtml as well |
| 08:38 | <hsivonen> | or if that's too hard, I want to turn them off properly instead of letting them fail in an ugly way |
| 08:43 | <Hixie> | ah |
| 08:43 | <Hixie> | the html5 way is to just use createElement or to use createElementNS with the xhtml ns |
| 08:43 | <Hixie> | the way that actually works... |
| 08:43 | <Hixie> | you need to use createElement in text/html, for IE |
| 08:43 | <Hixie> | but createElementNS() with the XHTML ns will work everywhere, i believe |
| 08:44 | <Hixie> | everywhere else, i mean |
| 08:44 | <Hixie> | even text/html (though you'll be creating the "xhtml" version of the nodes) |
| 08:44 | <hsivonen> | Hixie: and presumably, IE doesn't have createElementNS? |
| 08:44 | <Hixie> | (or some such weirdness) |
| 08:44 | <Hixie> | correct |
| 08:44 | <hsivonen> | so I could do if (!document.createElementNS) { |
| 08:45 | <hsivonen> | document.createElementNS = function() {... |
| 08:45 | <Hixie> | that might work |
| 08:46 | <hsivonen> | of course, I don't have IE to test with, so backporting to IE after Firefox/Opera/Safari work is at the bottom of the list... |
| 08:47 | <hsivonen> | perhaps I live in a reality distortion field, but having easy access only to Mac/Ubuntu is not unique to me |
| 08:48 | <Hixie> | it's becoming much more common |
| 08:48 | <Hixie> | i wonder what <datagrid>'s content model should be |
| 08:48 | <hsivonen> | when I gave the talk about HTML5 a week and a half ago, the audience seemed to be an all Mac/Ubuntu crowd |
| 08:49 | <hsivonen> | and one person commented that he doesn't test in IE because it doesn't run on his systems |
| 08:49 | <Hixie> | hehe |
| 08:49 | <Hixie> | it's a good trend. i just hope that it stops at 20% |
| 08:49 | <Hixie> | i wouldn't want IE's monopoly to be replaced by another monopoly |
| 08:49 | <Hixie> | we should have four browsers at 20% each and a mirad of other filling the remaining 20% |
| 08:49 | <Hixie> | four browser engines, rather |
| 08:50 | <Hixie> | myriad, even |
| 08:51 | <Hixie> | you know right now <canvas> is inline |
| 08:51 | <Hixie> | but it really should be either inline or inline or block |
| 08:51 | <Hixie> | in legacy terms |
| 08:51 | <Hixie> | er, either only block, or inline/block |
| 08:52 | <Hixie> | same with video/audio |
| 08:53 | <hsivonen> | Hixie: while you are at it, please consider the <figure> content model, too |
| 08:54 | <hsivonen> | that one has a clear bug now |
| 08:54 | <hsivonen> | and being able to showcase a block of text would make sense, too |
| 08:55 | <hsivonen> | the bug being the telescoping thing that you have to terminate <video> with <img> or <embed> |
| 08:56 | <hsivonen> | hmm. is iterating over childNodes in a for loop faster than iterating over the .nextSibling linked list? |
| 08:56 | <Hixie> | well i expect we'll just allow figure to contain anything |
| 08:56 | <Hixie> | as in, <figure> <legend> Listing 1 </legend> <pre> <code> ... </code> </pre> </figure> |
| 08:56 | <hsivonen> | Hixie: right |
| 08:57 | <Hixie> | so it'll solve itself |
| 08:57 | <Hixie> | i'm seriously considering allowing <figure> to basically contain anything |
| 08:57 | <Hixie> | even like <article> |
| 08:57 | <Hixie> | anything media or what we now think of as block-level |
| 08:57 | <hsivonen> | perhaps I shouldn't care about DOM perf right now |
| 08:58 | <Hixie> | don't pre-optimise :-) |
| 08:58 | <hsivonen> | the DOM is weird, because the internal impl can make different access patterns have counterintuitive perf characteristics if you consider the API |
| 08:59 | <hsivonen> | IIRC, Gecko uses a child node array internally instead of neighbor links like Xerces |
| 08:59 | jgraham_ | isn't sure what <figure><article> would do but is generally in favour of less restrictive content models |
| 08:59 | <jgraham_> | s/do/mean/ maybe |
| 09:00 | hsivonen | likes neighbor link tree access |
| 09:00 | <Hixie> | jgraham_: well it'd be a figure of an entire article, e.g. maybe <figure> <legend> Sample blog comment </legend> <article> <p>v1agr4!!!1</p> </article> </figure> or something |
| 09:00 | <Hixie> | (not a quoted blog comment) |
| 09:01 | <Hixie> | the more i look at this the more i find that we have a whole bunch of subtly different element types |
| 09:02 | <jgraham_> | Yeah, that seems quite tenuous to be. But I don't see that as a reason to not allow it :) |
| 09:03 | <Hixie> | no disagreement there |
| 09:07 | <anne-mac> | Hixie, in order for all those simple elements to get implemented the rendering section really needs to be written |
| 09:07 | <hsivonen> | zcorpan: is the className redundancy with irrelevant in the collapsing script for IE? |
| 09:07 | <anne-mac> | especially for section, etc. |
| 09:08 | <hsivonen> | Was there a way to trigger attribute selectors dynamically in IE7? |
| 09:08 | <hsivonen> | should I support IE6? |
| 09:08 | <Hixie> | anne-mac: well most of them are just display:block |
| 09:08 | <Hixie> | anne-mac: but actually, i was hoping for implementation feedback :-) |
| 09:08 | <Hixie> | IE6 is around as big as IE7 iirc |
| 09:09 | <hsivonen> | Hixie: it's qualitatively different, though :-) |
| 09:10 | <anne-mac> | Hixie, I mean that we probably need something like :heading(n) |
| 09:10 | <anne-mac> | Hixie, implementation feedback for rendering has failed for the last 10 years or so except for trivial issues |
| 09:10 | <anne-mac> | see forms |
| 09:11 | Philip` | wishes people didn't make sites that only work with IPv6 |
| 09:11 | <Hixie> | anne-mac: yeah well |
| 09:11 | <Hixie> | anne-mac: it's worked as well as for other things |
| 09:11 | <hsivonen> | Philip`: wow. what site? |
| 09:12 | <Hixie> | anne-mac: i agree that pseudo-classes would be helpful; not sure it's critical to getting html implemented though |
| 09:13 | <hsivonen> | Hixie: I think the HTML5 outline algorithm is mostly pointless without making sure that authors *see* the effects of the outline from day 1 |
| 09:13 | <hsivonen> | Hixie: that is, h1-h6 should have traditional style rules in the UA style sheet |
| 09:13 | <anne-mac> | I side with hsivonen although I'm willing to be convinced we should simply add them as display:block |
| 09:13 | <hsivonen> | when there's no section parent |
| 09:13 | <hsivonen> | ancestor, rather |
| 09:13 | <Philip`> | hsivonen: streaming.ist-ring.eu, though maybe that's just because they haven't quite got around to adding an IPv4 address yet |
| 09:14 | <Hixie> | hsivonen: well, h1 should auto-style, i agree. (not h2-h6, for compat reasons) |
| 09:14 | <hsivonen> | Hixie: what's the compat problem with |
| 09:14 | <hsivonen> | section h2:depth(n) |
| 09:14 | <hsivonen> | ? |
| 09:15 | <hsivonen> | Hixie: that is, limiting the autostyling to cases where there is at least one section ancestor? |
| 09:16 | <hsivonen> | Hixie: does the outline depth of a heading ever depend on stuff the follows the heading in document order? |
| 09:16 | <anne-mac> | <h1>xxx</h1>heading 1<h3>xxx</h3>heading 2 iirc |
| 09:17 | <Hixie> | hsivonen: the idea is to allow authors to get the right rendering in legacy UAs |
| 09:17 | <hsivonen> | anne-mac: in that case, as far a I can tell, the outline depth of a given node depends on stuff before it in the document order |
| 09:18 | <Hixie> | hsivonen: but then we want compatible rendering across browsers |
| 09:18 | <hsivonen> | Hixie: well, section doesn't play nice in IE/Firefox, so... |
| 09:18 | <Hixie> | hsivonen: having it autostyle in newer browsers and not in the others would just lead to very confused authors and users with content that doens't do what the author intended |
| 09:19 | <hsivonen> | Hixie: leaving the styling to authors would lead to mostly academic applicability of the outline algorithm |
| 09:19 | jgraham_ | agrees with hsivonen |
| 09:19 | <Hixie> | hsivonen: the intention of hte algorithm was mostly just to define what it was, so that if people _do_ try to implement one, they at least do the same one |
| 09:20 | <hsivonen> | Hixie: btw, I think it would help a great deal if the outline algorithm were restated from the point of view of a node whose outline depth is being computed |
| 09:20 | <zcorpan> | hsivonen: ie6, yes (re className) |
| 09:20 | <Hixie> | hsivonen: you mean http://www.whatwg.org/specs/web-apps/current-work/#associatedSection ? |
| 09:20 | <hsivonen> | Hixie: such that it would be straight-forward to evaluate on a per-node basis |
| 09:20 | <zcorpan> | hsivonen: className += '' (or something similar) is needed to trigger a reflow in ie7 |
| 09:21 | <hsivonen> | zcorpan: thanks. can reflow be forced in IE6 similarly? or does IE6 even support attribute selectors? I forget |
| 09:22 | <anne-mac> | Hixie, you want a clear algorithm to compute the heading level so implementors can say something about the impact of supporting :heading(n) |
| 09:22 | <zcorpan> | hsivonen: ie6 doens't support attr selectors |
| 09:22 | <hsivonen> | Hixie: not only walking back to find heading but walking back to find outline depth |
| 09:22 | <hsivonen> | zcorpan: ok. thanks |
| 09:23 | <hsivonen> | Hixie: I think the depth of a node in the outline should be computable on a per-node basis by walking the document order backwards |
| 09:23 | <Hixie> | hm |
| 09:23 | <jgraham_> | Yeah, the current algorithm is a) black magic and b) quite far away from what many types of UA are likely to want to implement |
| 09:24 | <hsivonen> | Hixie: and, conversely, on a batch basis by walking from the root forward without backtracking while keeping some variables |
| 09:24 | <Hixie> | anne-mac: i never actually intended for <h1> styling to be based on implied sections and so forth |
| 09:24 | <hsivonen> | Hixie: i.e. the isolated selector re-evaluation case and the SAX/batch selector case |
| 09:24 | <Hixie> | anne-mac: my original intent was for <h1> styling to be based exclusively on the number of <section> ancestors |
| 09:24 | <Hixie> | hsivonen: i guess that must be possible |
| 09:25 | <Hixie> | hsivonen: i haven't really thought about it |
| 09:25 | <Hixie> | hsivonen: the algorithms in the spec are intended to be optimised when implemented |
| 09:25 | <Hixie> | hsivonen: but i'm happy to change the spec's algorithm at some point if we find a better way of phrasing it |
| 09:25 | <anne-mac> | it's not just <section>, it's also <article>, etc. |
| 09:25 | <zcorpan> | hsivonen: function createHtmlElement(tagName) { return document.createElementNS ? document.createElementNS("http://www.w3.org/1999/xhtml";, tagName) : document.createElement(tagName); } |
| 09:25 | <Hixie> | anne-mac: my original intent was for <h1> styling to be based exclusively on the number of <section> ancestors |
| 09:25 | <anne-mac> | but yeah, implied headings too |
| 09:26 | <anne-mac> | ok |
| 09:26 | <hsivonen> | zcorpan: thanks |
| 09:26 | <Hixie> | anne-mac: of course we later added other elements... |
| 09:26 | <Hixie> | anne-mac: which may make that no longer really the best solution |
| 09:26 | <othermaciej> | the OggStorm seems to have passed |
| 09:28 | <hsivonen> | Hixie: it seems to me that stating the spec in terms of algorithms is helpful when the general point of view of the algorithm matches the way implementors will need to look at the issue |
| 09:29 | <hsivonen> | Hixie: however, requiring each implementor to work out the invariats / declarative statement of what the algorithm does and recasting it to another algorithm from a different POV doesn't seem useful |
| 09:29 | <hsivonen> | Hixie: if the implementor is expected to find a very different algorithm, a declarative statement of the identities/constraints would be more useful |
| 09:30 | <othermaciej> | are y'all talking about the sectioning algorithm? |
| 09:31 | <Hixie> | if we can find a way to phrase it that way that works too |
| 09:31 | <hsivonen> | othermaciej: yes |
| 09:31 | <othermaciej> | I must admit as written it's not directly useful to a UA that wants to efficiently style headings by level |
| 09:31 | <othermaciej> | but I'm not actually sure it even defines an effective heading level |
| 09:32 | <othermaciej> | (should both the rank and level of section nesting affect the effective heading level? only level of section nesting?) |
| 09:32 | <jgraham_> | othermaciej: Both, in different ways |
| 09:33 | <anne-mac> | i'm fine with only doing h1 auto-styling btw, although it might encourage people to create pages that are not backwards compatible |
| 09:33 | <anne-mac> | you'd do h1:heading(n) in the browser basically |
| 09:34 | <hsivonen> | zcorpan: is the purpose of addValueAttrs() to prevent the UA from renumbering the list items when some are hidden? |
| 09:34 | <othermaciej> | jgraham_: can you be more specific? |
| 09:35 | <othermaciej> | the sectioning algorithm does specifically allow having lower rank headers than the first inside a section |
| 09:35 | <hsivonen> | Hixie: thinking about it more, I think the we need two things: |
| 09:35 | <othermaciej> | without creating a section break |
| 09:35 | <hsivonen> | 1) a forward-only sweep that keeps variables as it goes and assigns depth to nodes |
| 09:36 | <hsivonen> | 2) A way to walk back from a tainted node to an untainted node and then do a partial sweep from there onwards |
| 09:37 | <hsivonen> | it would be a bonus if the partial sweep could end at the last sibling of the tainted node and wouldn't require resweeping the whole doc until the end |
| 09:38 | <Hixie> | send mail |
| 09:38 | <Hixie> | i'm far from working on that right now |
| 09:38 | <hsivonen> | Disclaimer: I have no idea if it is even footprint-wise acceptable to keep a level variable and taint flag on nodes |
| 09:38 | <Hixie> | it's content models all the way til the new year for me, i think |
| 09:38 | <hsivonen> | Hixie: ok |
| 09:38 | <othermaciej> | if the styling is solely based on rank and number of section ancestors that might not be too hard |
| 09:39 | <anne-mac> | i think it should also be based on previous siblings |
| 09:39 | <othermaciej> | keeping flags on HTMLHeaderElement nodes only is likely acceptable |
| 09:39 | <Hixie> | hsivonen: but in general i agree with your suggestions and requests |
| 09:39 | <anne-mac> | <h1>1</h1>xx<h3>2</h3> |
| 09:40 | <anne-mac> | <h1>1</h1>xx<div><h3>2</h3></div> makes this more complicated I guess |
| 09:40 | <Hixie> | (also <h3>1</h3>xx<h2>1</h2> iirc) |
| 09:40 | <krijnh> | I don't get 'backwards compatibility' with h1-h6 combined with section/article, when section/article isn't even supported now.. |
| 09:41 | <krijnh> | When people start using <section> <hx> surely they don't care about backwards compatibility, right? |
| 09:42 | <krijnh> | (Sorry, didn't mean to ruin the conversation) |
| 09:43 | <hsivonen> | krijnh: as far as I can tell, <section> is backwards-compat poison, yes |
| 09:43 | <krijnh> | hsivonen: I know |
| 09:43 | <zcorpan> | Hixie: it seems wf2 doesn't allow <form target>. is that on your radar or should i send email? |
| 09:43 | <Hixie> | |
| 09:43 | <zcorpan> | ok |
| 09:43 | <Hixie> | but wf2 is blocked until the forms tf comes back |
| 09:44 | <hsivonen> | zcorpan: I speculatively treated the non-allowance as a spec bug last night |
| 09:45 | <othermaciej> | oh great, that means the forms tf needs to do something |
| 09:45 | <hsivonen> | othermaciej: no, they don't. target isn't architectural consistency. (or so I'd hope) |
| 09:46 | <hsivonen> | Hixie: whatever happened with WF2 call for impls from the WHATWG POV? |
| 09:47 | <Hixie> | still waiting :-) |
| 09:47 | <Hixie> | afk |
| 09:48 | <othermaciej> | hsivonen: the Forms TF needs to do something at some point to remove the blocker to forms progress |
| 09:51 | <anne-mac> | there's a deadline |
| 09:55 | <Philip`> | What happens if the deadline is exceeded? |
| 09:56 | <Hixie> | anne-mac: yeah because timelines have worked really well so far in this working group |
| 09:57 | <Hixie> | also, as far as i can tell there isn't a deadline per so, just a desired timeline :-) |
| 09:58 | <anne-mac> | well, if there isn't any progress by then I think there's enough evidence that it doesn't work |
| 10:00 | <anne-mac> | zcorpan, where does WF2 state it has removed a feature from HTML4? |
| 10:00 | <hsivonen> | Hixie: what's the deal with HTML5 zapping the HTMLHeaderElement interface? |
| 10:02 | <Hixie> | there wasn't anything left in it, since i haven't put the align="" attributes, etc in the idl |
| 10:05 | <hsivonen> | Hixie: but browsers need to support HTMLHeaderElement until the end of the world anyway |
| 10:06 | <Hixie> | yeah, there's a bunch of stuff i need to re-add, probably when i do the rendering section |
| 10:07 | <anne-mac> | let me know when you change your mind about IDL interfaces being for authors |
| 10:07 | <anne-mac> | so I can complain about document.write() again |
| 10:07 | <anne-mac> | :) |
| 10:08 | <hsivonen> | Hixie: email sent about the outline algorithm. I hope the email captures the main points. |
| 10:09 | <othermaciej> | I don't think it makes sense to leave "bad" things out of the IDL interfaces |
| 10:10 | <othermaciej> | it makes life harder for implementors (since we have to reconstruct the real IDL interfaces) and conformance checking script use of DOM APIs seems pointless and infeasible |
| 10:10 | <Hixie> | anne-mac: almost all idl changes are on hold until bindings for dom is done |
| 10:10 | <Hixie> | hsivonen: thanks |
| 10:10 | <Hixie> | othermaciej: agreed |
| 10:11 | <Hixie> | othermaciej: i just haven't done any of the presentational attributes yet, dom or not |
| 10:12 | <othermaciej> | sounds reasonable |
| 10:13 | anne-mac | suggested to heycam to name it Web IDL instead |
| 10:14 | <hsivonen> | do the Samsung products support Theora, too, or only Ogg and Vorbis? |
| 10:14 | <Lachy> | woah! Microsoft have shipped ogg vorbis in Halo for PC |
| 10:15 | <othermaciej> | hsivonen: I'd be surprised if any products support Theora already given that the bitstream format wasn't frozen until a few months ago |
| 10:15 | <othermaciej> | (not impossible though) |
| 10:15 | <hsivonen> | Lachy: yeah, *Vorbis* has shipped in a lot of high-$$$ games |
| 10:16 | <aphid> | metavid is theora based, we have a few thousand hours of congressional video |
| 10:17 | <othermaciej> | it's unfortunate that the Xiph press release conflates Ogg, Vorbis and Theora |
| 10:17 | <hsivonen> | othermaciej: yeah. |
| 10:21 | <zcorpan> | anne-mac: <form target> is not allowed in html4 strict |
| 10:22 | <anne-mac> | sure |
| 10:23 | <maikmerten> | the Ogg family is considered to be a set of codecs which is simply called "Ogg". Granted, the release could have been more specific. |
| 10:23 | <zcorpan> | anne-mac: i had presumed that wf2 built on top of html4 strict, but it actually just says html4, so i don't know |
| 10:24 | <anne-mac> | it is true that the XHTML Module definition at the end doesn't mention <form target>, <input target> or <button target> |
| 10:24 | <maikmerten> | anyway, the press release at least totally applies to Ogg Vorbis. Albeit most discussion is centered on video the audio part shouldn't be neglected either |
| 10:24 | <Hixie> | maybe i should just allow any nesting, except that each element adds to a set of elements that are banned as descendants |
| 10:26 | <Hixie> | e.g. <header> can't contain <footer> but can contain anything else |
| 10:26 | <Hixie> | oh also <section>, <article>, <aside> wouldn't be allowed |
| 10:26 | <othermaciej> | maik|afk: it seems to say things that mostly apply to Ogg Vorbis, but then talks about the <video> element and video in general |
| 10:26 | <Hixie> | but <nav> maybe would? maybe not. |
| 10:26 | <Hixie> | gah |
| 10:27 | <anne-mac> | where is the <a> element on the v axis? |
| 10:27 | <Hixie> | oops, maybe i dropped it |
| 10:27 | <Hixie> | added it |
| 10:28 | <othermaciej> | oops, I was wrong about when the Theora bitstream format was frozen |
| 10:28 | <anne-mac> | oh, <dt> should probably be similar to <h1>-<h6> too |
| 10:28 | <othermaciej> | mea culpa |
| 10:29 | <maikmerten> | othermaciej, while this is true the itent is also to show xiph.org is knowing what it is doing |
| 10:30 | <maikmerten> | and of course the millions of deployments *do* apply to Theora |
| 10:30 | <Hixie> | are there any deployments of theora that are done by companies with big targets on their backs? because that's the only reason we're not requiring it at the moment (the submarine patent issue) |
| 10:31 | <hsivonen> | maikmerten: it seems to me that Ogg proponents arguing about Theora risks is like a Green party making statements about pollution: if it isn't 100% correct, the incorrect part gets all the attention |
| 10:31 | <maikmerten> | I consider Novell rather big |
| 10:31 | <maikmerten> | hsivonen, same for Nokia position papers, unfortunately ;) |
| 10:32 | <maikmerten> | (but that is another cup of tea) |
| 10:32 | <Hixie> | novell is pretty small these days, heck they're below $3bn market cap and living mostly on chicken feed from microsoft |
| 10:33 | <hsivonen> | is IBM shipping Theora in any of their Linux things? |
| 10:33 | <maikmerten> | but at least it's clear that at least some of the Ogg codecs are rather widely deployed, well tested and adopted by big players, still e.g. word has been that "Ogg" is not an option |
| 10:33 | <maikmerten> | what about Ogg Vorbis? What about Ogg Speex? (Xbox live) |
| 10:33 | <Hixie> | as far as i know the only argument is about ogg theora |
| 10:34 | <Hixie> | i don't know that anyone has made any particular claims about vorbis |
| 10:34 | <hsivonen> | maikmerten: almost all of the naysaying in the HTML5 context has been about Theora. Xiph using Vorbis references is not helpful there |
| 10:34 | <roc> | IBM doesn't ship Linux |
| 10:34 | <maikmerten> | hsivonen, if IBM is shipping a full blown linux desktop: I'd think so |
| 10:35 | <Philip`> | Should the <audio> baseline be the same as the audio component of <video>? |
| 10:35 | <roc> | they aren't |
| 10:35 | <anne-mac> | hsivonen, Vorbis was in the spec before though |
| 10:35 | <Philip`> | (*...of the <video> baseline) |
| 10:35 | <Hixie> | audio isn't really anywhere near as important as video |
| 10:36 | <othermaciej> | Philip`: possibly, although simple lossless formats might also be worth supporting for things like sound effects |
| 10:36 | <hdh> | ... outside myspace |
| 10:37 | <hsivonen> | although it is well-known in Maemo circles that Nokia specifically avoids shipping Vorbis, too |
| 10:37 | <maikmerten> | hsivonen, albeit this is correct most of the press release states that albeit major proponents of MPEG give the public image that MPEG is somehow safe (they know otherwise and won't deny when being asked in detail) in fact it's not quite that easy |
| 10:37 | <stijntje> | hdh: doesn't myspace use flash for audio anyway, presumably to make copying it harder (which would be dead easy with <audio src="...">? |
| 10:39 | <othermaciej> | it's not that hard to grab the audio from myspace |
| 10:39 | <othermaciej> | (at least with Safari's activity window) |
| 10:39 | <Philip`> | stijntje: That would require people to 'view source' and work their way through masses of horrid HTML to find the URI, which sounds harder than just copying Flash's cache of the file from its temp directory |
| 10:40 | <Philip`> | (unless their browser adds a 'save this file' button to the <audio> UI) |
| 10:40 | <stijntje> | considering myspace's use of HTML, that would make sense ;) |
| 10:40 | <roc> | I bet there's a Firefox extension for this |
| 10:40 | <hdh> | then there's commons.wikimedia, librivox, gutenberg(?), streaming sites like jamendo |
| 10:41 | <stijntje> | othermaciej: it is not hard indeed, in the way it is not hard to copy a "protected" (transparent .gif on top) photo on flickr too, but it does stop the "masses" from doing it |
| 10:41 | <stijntje> | of course I'm not working at MySpace so I wouldn't know whether that's really the reason to use flash, I can imagine interoperability and usability would play a significant role too :P |
| 10:41 | <othermaciej> | hah, I didn't know flickr did silly things like that |
| 10:42 | <hdh> | myspace example was an uneducated guess from xkcd #134 anyway |
| 10:42 | <Hixie> | maikmerten: at least in the html5 context, i don't think anyone has claimed that mpeg is safe |
| 10:43 | <roc> | http://lifehacker.com/software/downloads/download-of-the-day--video-downloader-firefox-extension-170659.php |
| 10:43 | <Hixie> | maikmerten: i can't comment on the "public image" aspect, but i assure you that has no bearing on the technical decisions of the spec |
| 10:43 | <hdh> | doesn't a royalty-free requirement exclude all things under MPEG-LA? |
| 10:44 | <Hixie> | unless MPEG-LA change their minds, yes |
| 10:44 | <Hixie> | well, there are some obsolete MPEG codecs we could use |
| 10:44 | <Hixie> | but they're obsolete. |
| 10:45 | <stijntje> | are there any alternatives to theora (patent issues-wise) that are not? |
| 10:46 | <Philip`> | Dirac is the only similar project I've heard of |
| 10:46 | <Philip`> | and that seems somewhat too experimental at the moment |
| 10:46 | <stijntje> | yeah, me too, but afaik it's still in its early stages |
| 10:46 | <roc> | Dirac is not done and has a lot less patent analysis than Theora |
| 10:46 | <roc> | AFAIK |
| 10:47 | <maikmerten> | Hixie, right |
| 10:47 | <maikmerten> | Hixie, this release is bringing nothing new to the table |
| 10:47 | <hsivonen> | Dirac uses arithmetic coding. When are the IBM patents on that going to hit the 20 year mark? |
| 10:47 | <maikmerten> | Hixie, but it was necessary to e.g. contain the damage Nokia was doing to us in the public image |
| 10:48 | <hsivonen> | The Dirac FAQ is rather vague on that pointt |
| 10:48 | <maikmerten> | Hixie, like to fight the impression xiph.org would be new/geeky/not trustworthy and generally not used at all |
| 10:48 | <hdh> | there's a #dirac here |
| 10:49 | <maikmerten> | Hixie, the background is that xiph.org is being flooded right now by press people - so we can deal with the "FAQ" in a release |
| 10:49 | <Hixie> | maikmerten: sure, i wasn't commenting on the press release, i understand the reasons for it |
| 10:50 | <Hixie> | cool, being flooded with press people is a good problem to have :-) |
| 10:50 | <maikmerten> | didn't I yesterday state that press releases are boring ;) |
| 10:50 | <maikmerten> | s/release/press release, obviously |
| 10:50 | <othermaciej> | as some famous person once said, "there's no such thing as bad publicity" |
| 10:52 | <Philip`> | othermaciej: Is there any evidence to back up that statement? :-) |
| 10:52 | <Hixie> | bed time |
| 10:52 | <Hixie> | nn |
| 10:53 | <stijntje> | gn |
| 10:58 | <heycam> | anne-mac, yeah sorry about the delay with that. over the christmas break, once the next batik release is out, i'll have some time for it. |
| 11:01 | <Lachy> | this is interesting. Not sure what I think of it yet (although I'm not really allowed to say much about it). http://www.opera.com/pressreleases/en/2007/12/13/ |
| 11:03 | <hsivonen> | wow |
| 11:03 | <stijntje> | "Opera requests the Commission to implement two remedies to Microsoft�s abusive actions. First, it requests the Commission to obligate Microsoft to unbundle Internet Explorer from Windows" <- wouldn't that just lead to another Windows N no one will buy? |
| 11:10 | <hsivonen> | hmm. Opera Mobile is not linked from the press release like desktop and mini |
| 11:10 | <hsivonen> | and the Opera 9 coming soon for S60 banner is gone |
| 11:10 | <othermaciej> | stijntje: I think Microsoft can achieve that all by itself :-) |
| 11:15 | <hasather> | hsivonen: Opera 9 Mobile was announced for Windows Mobile and UIQ some days ago, but no word on the S60 version yet |
| 11:15 | <othermaciej> | now seems like an odd time to file an antitrust complaint about Microsoft bundling the browser |
| 11:15 | <hsivonen> | hasather: oh. interesting |
| 11:17 | <hasather> | hsivonen: see here, I asked Daniel about the S60 version in the comments: http://operawatch.com/news/2007/12/here-comes-opera-mobile-9.html |
| 11:19 | <hsivonen> | hasather: thanks |
| 12:07 | <stijntje> | mp3: Extince - Le Fanclub |
| 12:07 | <stijntje> | woops |
| 12:07 | <stijntje> | apologies, wrong channel :( |
| 12:09 | gsnedders | notices the whole zero conversations that stijntje disturbed :) |
| 12:10 | <stijntje> | :D |
| 12:14 | <othermaciej> | if I want to vertically center-align text in two different font sizes using CSS, can I do that? |
| 12:14 | <othermaciej> | (sorry for slightly off-topic question, figured people here might know) |
| 12:16 | <hsivonen> | now that I'm working on front end UI JS, are there other Validator.nu front-end JS requests besides duplicate message collapsing and swapping in a <textfield> or a file upload control? |
| 12:17 | othermaciej | is terrified to discover that you can relative-position a float |
| 12:18 | <Philip`> | othermaciej: You should use <table valign=middle> |
| 12:19 | <othermaciej> | Philip`: isn't it a sin against all that is righteous to use layout tables? |
| 12:19 | <Philip`> | othermaciej: It's not really layout, it's just presentation |
| 12:19 | <othermaciej> | (I got what I wanted to working but the result is a little ugly) |
| 12:20 | <stijntje> | would display: table-cell + vertical-align: middle work? notwithstanding that it doesn't work in IE7 |
| 12:20 | <othermaciej> | I'm not sure if that would properly vertically align two pieces of text in two different fonts |
| 12:21 | <othermaciej> | I think I would need them to be in two different table cells |
| 12:21 | <Philip`> | <div style="display:table"><div style="display:table-row"><div style="display:table-cell;vertical-align:middle">... |
| 12:21 | <Philip`> | That's not got evil <table> therefore it's good |
| 12:21 | <stijntje> | not saying so, just wondering whether it'd work |
| 12:22 | <gsnedders> | stijntje: well, yeah, it'd work |
| 12:22 | <Philip`> | Does CSS define that the outer table and table-row get inferred somehow? |
| 12:22 | <othermaciej> | what I have is: |
| 12:22 | <othermaciej> | <h2><span style="float: left; position: relative; bottom: 0.25em; font-size: 2em;">☀ </span> My Heading Stuff</h2> |
| 12:23 | <othermaciej> | the need for the nbsp is unfortunate |
| 12:23 | <othermaciej> | (I will move the style rules into the stylesheet once I decide how this should work) |
| 12:26 | <Philip`> | <div style="display:inline-block;vertical-align:middle"> works in Opera at least |
| 12:26 | <gsnedders> | Philip`: how do you do bar charts in HTML? |
| 12:26 | <Philip`> | gsnedders: Do you mean the ones like http://www.cl.cam.ac.uk/~pjt47/misc/attributes.html ? |
| 12:26 | <stijntje> | firefox doesn't support inline-block afaik |
| 12:26 | <gsnedders> | Philip`: yeah |
| 12:27 | <othermaciej> | Philip`: I ruled out inline-block due to lack of Firefox/IE support |
| 12:27 | <Philip`> | gsnedders: View source :-) |
| 12:27 | <Philip`> | -moz-inline-block sometimes works in Firefox |
| 12:27 | <gsnedders> | Philip`: so the JS and CSS? |
| 12:27 | <Philip`> | gsnedders: Yes, and the HTML |
| 12:28 | <Philip`> | Ignore the incorrectness of the word "histogram" as used in the code |
| 12:33 | <gsnedders> | wow. Excel 11 Mac's HTML output isn't that bad. |
| 12:46 | gsnedders | wonders why the code is failing on him |
| 12:50 | <gsnedders> | Philip`: the entire first column gets changed to NaN here :\ |
| 12:52 | <gsnedders> | ah. whitespace between <tr> and <td> |
| 12:59 | <hsivonen> | zcorpan: if the script is expected to prevent li renumbering in Firefox, it isn't working |
| 13:02 | <hsivonen> | zcorpan: script deployed |
| 13:02 | <hsivonen> | zcorpan: thank you |
| 13:05 | <hsivonen> | hmm. now expanding a grouped message is counter-intuitive when expanding |
| 13:05 | <hsivonen> | because you don't notice where it expands |
| 13:08 | <stijntje> | http://img520.imageshack.us/img520/9563/negerted2.png |
| 13:08 | <stijntje> | nvm |
| 13:17 | <annevk> | othermaciej, Firefox 3 does inline-block... |
| 13:18 | <hsivonen> | hmm. I guess the message grouping UI could be better, if the collapsed messages were moved into a sublist in <details> |
| 13:18 | <hsivonen> | but <details> lacks impls. |
| 13:19 | <hsivonen> | and if people start scripting <details> impls now, it'll poison the possibility of introducing native impls without breaking the Web. |
| 13:19 | <hsivonen> | hmm. |
| 13:20 | <hsivonen> | Quick poll: if I change the "collapse all" feature to "group by message", do I need to provide an undo button? |
| 13:21 | <hsivonen> | for ungrouping? |
| 13:25 | <zcorpan> | hsivonen: http://simon.html5.org/temp/validator-nu-collapse.html has ungrouping |
| 13:25 | <zcorpan> | and some bugs fixed, in case you were working from an older copy |
| 13:25 | <hsivonen> | zcorpan: yeah. looks like I accidentally broke ungrouping |
| 13:26 | <hsivonen> | zcorpan: I downloaded yesterday evening version |
| 13:27 | <zcorpan> | hsivonen: ok, then it should be the latest version |
| 13:28 | <zcorpan> | hsivonen: it doesn't renumber in opera |
| 13:28 | <hsivonen> | zcorpan: fixed |
| 13:28 | <zcorpan> | er, at least not on my sample |
| 13:29 | <hsivonen> | zcorpan: what's the purpose of addValueAttrs then? |
| 13:30 | <zcorpan> | hsivonen: to prevent renumbering. it works on my copy but not on yours |
| 13:30 | <hsivonen> | weird |
| 13:31 | <hsivonen> | I'm tempted to reimplement it as O(n) and with grouping the various locations in a sublist |
| 13:31 | <zcorpan> | ok, that might result in a better ui |
| 13:37 | zcorpan | is very confused as to why collapsing renumbers in v.nu but not in my copy |
| 13:50 | <Philip`> | gsnedders: It doesn't do anything clever about handling whitespace - it just assumes the HTML is written precisely like how I wrote it, without doing anything useful like documenting that fact :-) |
| 13:50 | <gsnedders> | Philip`: :) |
| 14:09 | <hsivonen> | what's the difference of textContent and innerText? |
| 14:11 | <hsivonen> | is it just a gratuitous naming difference? |
| 14:32 | <gsnedders> | http://http-parsing.gsnedders.com/ |
| 14:34 | <hsivonen> | gsnedders: hmm. the x-pingback frequency suggests that dmoz doesn't link to blogs much |
| 14:34 | gsnedders | doesn't know how Philip` chose the sample |
| 14:36 | <gsnedders> | Philip`: what's interesting is we have one header name which is blank, and one that is 0. Looks like the parsing isn't perfect :) |
| 14:40 | <hsivonen> | set-cookie2 looks like a miserable failure |
| 14:46 | <Philip`> | gsnedders: It's a random selection from the ~4.5M URIs in a dmoz.org dump from a few months ago (after removing duplicates) |
| 14:46 | <Philip`> | (where "random" means "sort -R | head") |
| 14:50 | <Philip`> | gsnedders: The one with an apparently blank header is a pretty broken server |
| 14:50 | <Philip`> | (http://www.nightwing.com.au/index.htm) |
| 14:52 | <Philip`> | It sends Content-Type twice, and "Client-Junk: :", etc, though I've got no idea why it's parsed like it is |
| 14:53 | <Philip`> | Oh, wait |
| 14:53 | <Philip`> | It does different things if I use curl instead of GET |
| 14:53 | <Philip`> | It has a header line ":" |
| 14:59 | <Philip`> | http://www.4silverhillspares.co.uk/cgi-bin/home.pl says "Pragma: no-cache" "0: no-cache" |
| 15:06 | <Philip`> | http://wm01.allmusic.com/cg/amg.dll says "Content-Type: text/html" "Content-Length: 31155" "Content:" - I'm not sure if they were intentionally adding that header last to identify the message content |
| 15:07 | <Philip`> | Where does Page-Completion-Status come from? |
| 15:08 | <Philip`> | Oh, ColdFusion |
| 15:11 | <zcorpan> | is there a reference somewhere for how much of the web is text/html? |
| 15:12 | <zcorpan> | i've seen the number 98% but i don't know where it came from |
| 15:12 | <Philip`> | 98% of what? |
| 15:12 | <Philip`> | (e.g. does that include image files and scripts and stylesheets and everything else?) |
| 15:13 | <zcorpan> | (no) |
| 15:13 | <zcorpan> | what <a href> points to |
| 15:13 | <zcorpan> | i guess |
| 15:14 | <Philip`> | How would you count sites with a million pages or sites with an infinite number of pages? |
| 15:16 | <zcorpan> | will you do a crawl and then tell me how many text/html results you got if i suggest how to count sites? :) |
| 15:17 | zcorpan | does s/98%/The majority/ |
| 15:17 | <zcorpan> | s/majority/vast majority/ |
| 15:18 | <Philip`> | I can find that 99.7% of dmoz.org is text/html with the particular User-Agent and Accept headers that I have |
| 15:18 | <Philip`> | If you can suggest what "a crawl" is, I could probably do something like that :-) |
| 15:22 | <Philip`> | gsnedders: Maybe you should indicate whether e.g. a site that sends two Set-Cookies is counted as one or two |
| 15:30 | <gsnedders> | wow. almost all the Date values are valid. |
| 15:32 | <Philip`> | I assume people writing CGI/PHP/etc don't bother outputting Date at all, so the Dates are generated by Apache/IIS/etc and are therefore more competently designed and tested |
| 15:32 | <Philip`> | *so the Dates which exist are ... |
| 15:35 | gsnedders | realises he actually has a bug in his code so even more are valid :P |
| 15:35 | <Philip`> | Try checking cookies for validity |
| 15:36 | <gsnedders> | OK, one more date is valid. :P |
| 15:36 | <Philip`> | (At least HttpClient complained a lot about cookies, so I guess there's interesting problems there) |
| 15:40 | <gsnedders> | http://http-parsing.gsnedders.com/ now has Date values |
| 15:42 | <Philip`> | "Everything done. Thank you for downloading a media file containing proprietary and patented technology. Core dumped ;)" - mplayer is a bit peculiar when it reaches the end of a stream |
| 15:42 | <Philip`> | gsnedders: That makes it sound like people actually say "Date: RFC822" |
| 15:43 | <gsnedders> | I couldn't think of any better way to phrase it though, quickly, Philip` |
| 15:43 | <gsnedders> | (and I'm not having each and every RFC 822 date separately :)) |
| 15:43 | <Philip`> | "Most Common Date Formats"? |
| 15:45 | <Philip`> | and probably show the asctime-formatted date, rather than saying "asctime()" |
| 15:45 | <gsnedders> | Philip`: it's valid, though |
| 15:46 | <Philip`> | gsnedders: It's confusing since it makes me wonder whether someone literally sent "Date: asctime()" |
| 15:46 | <hsivonen> | IIRC The httpclient docs say you need netscape emulation instea of rfc stuff |
| 15:47 | <Philip`> | hsivonen: It says that for cookies (http://jakarta.apache.org/httpcomponents/httpclient-3.x/cookies.html) - do you mean it says the same for Date headers? |
| 15:51 | zcorpan | finds that http://dev.w3.org/csswg/cssom-view/ has a different "the body element" than html5 |
| 15:51 | zcorpan | thinks cssom-view's definition is better |
| 15:52 | <Philip`> | There's quite a bit of variation in cookie expiry dates - "01-Jan-2038", "02 Jan 1970", "04-Dec-37", lots of "Friday" vs "Fri", a "12:40:25 AM GMT", some missing spaces/commas, etc |
| 15:52 | <hsivonen> | Philip`: no, just cookies |
| 15:53 | <gsnedders> | Philip`: is it clearer now? |
| 15:54 | <Philip`> | hsivonen: Okay |
| 15:54 | <Philip`> | gsnedders: It is |
| 15:54 | <gsnedders> | Philip`: now, should I include dates from other headers there? |
| 15:55 | <Philip`> | Server: Microsoft-IIS/6.0 |
| 15:55 | <Philip`> | File 'c: \mysql\share\charsets\?.conf' not found (Errcode: 2) |
| 15:55 | <Philip`> | Character set '#33' is not a compiled character set and is not specified in the 'c: \mysql\share\charsets\Index' file |
| 15:55 | <Philip`> | X-Powered-By: PHP/4.4.7 |
| 15:56 | <Philip`> | (from http://www.ordinepsicologi.piemonte.it/) doesn't look quite right |
| 15:57 | <Philip`> | gsnedders: Depends on what your data is meant to be used for and whether that'll be useful, I guess |
| 15:57 | <gsnedders> | how can you merge multiple lists? :P |
| 15:57 | <Philip`> | In Python? |
| 15:57 | <gsnedders> | just use .append() multiple times? |
| 15:57 | <gsnedders> | yeah |
| 15:57 | <Philip`> | [1,2] + [3,4] |
| 15:58 | <Philip`> | or I guess there's some concatenate method |
| 15:58 | <Philip`> | http://docs.python.org/lib/typesseq-mutable.html - list1.extend(list2) |
| 15:59 | gsnedders | doesn't really understand their equivalents |
| 16:00 | <gsnedders> | Expires is the only other header allowed in responses |
| 16:00 | <gsnedders> | and Last-Modified |
| 16:00 | <Philip`> | gsnedders: Do you mean the s[len(s):len(s)] = x equivalent? |
| 16:00 | <gsnedders> | yeah |
| 16:02 | <Philip`> | s[a:b] refers to the part of the list from index a up to index b-1, so s[len(s):len(s)] refers to the zero-length part just after the final item in the list, and if you assign something there then it gets spliced into s |
| 16:07 | <gsnedders> | Odd. The second most common date format (after valid RFC 822) is "-1". |
| 16:08 | <gsnedders> | Philip`: ah. slightly odd way to do it logically. |
| 16:08 | <Philip`> | gsnedders: That's why they have .append() and .extend() instead :-) |
| 16:09 | <gsnedders> | most of the invalid dates are relatively consistent in their format, which is nice. |
| 16:11 | <gsnedders> | defining date parsing isn't nice, though |
| 16:11 | gsnedders | points at HTML 5 |
| 16:22 | <gsnedders> | Defining date formats (including invalid ones) as ABNF seems simpler |
| 16:23 | <gsnedders> | Do any elements close open formatting elements? |
| 16:23 | <Philip`> | HTML5 isn't necessarily a good example of the neatest way to write algorithms :-) |
| 16:27 | <gsnedders> | There are a finite number of ways to do so that leave nothing that's unclear, though |
| 16:27 | <Philip`> | Experimentation suggests that nothing closes open formatting elements |
| 16:28 | <Philip`> | at least when you do <b> ... <x></x> for all tag names x ... </b> |
| 16:37 | <Philip`> | Hixie: status-documentation.html says "s quite stable" and probably needs more vowels |
| 16:53 | zcorpan | looks at the matrix |
| 16:54 | <zcorpan> | <footer><address> seems like it should be allowed |
| 16:54 | <gsnedders> | wow. 1547 Content-Location headers. |
| 16:57 | Philip` | wonders what Content-Location is meant to be used for |
| 16:59 | <gsnedders> | Philip`: Base URI to resolve relative URIs to |
| 17:00 | <Philip`> | gsnedders: Ah |
| 17:00 | <gsnedders> | IIS in some cases sends an bad URI by default |
| 17:01 | <gsnedders> | Philip`: http://support.microsoft.com/kb/218180 |
| 17:02 | <Philip`> | I guess http://page.freett.com/cahfm/ is a bad one since its Content-Location points at :8080 which isn't listening to connections |
| 17:02 | <gsnedders> | the norm is an internal network IP |
| 17:03 | <Philip`> | Are you allowed relative URIs, like "Content-Location: rfc1596.html"? |
| 17:04 | Philip` | assumes so, since ietf.org is doing that |
| 17:05 | <gsnedders> | off the top of my head, no |
| 17:05 | <gsnedders> | you are, actually |
| 17:05 | <gsnedders> | Location must be absolute, though |
| 17:06 | <gsnedders> | caused all kinds of issues in older versions of Saf that enforced that |
| 17:11 | gsnedders | wonders what this page-completion-status header is |
| 17:12 | <gsnedders> | ColdFusion apparently |
| 17:24 | <Philip`> | gsnedders: http://krijnhoetmer.nl/irc-logs/whatwg/20071213#l-771 |
| 17:24 | <gsnedders> | heh. |
| 17:24 | <gsnedders> | I missed that. |
| 17:25 | gsnedders | wonders how he should deal with Server values |
| 17:25 | <Philip`> | Use a pie chart! |
| 17:25 | <gsnedders> | mmm… π… |
| 17:26 | <gsnedders> | But do I keep the whole string, the server/version, or just the server? |
| 17:27 | <Philip`> | Do a pie chart with one slice per server, then cut the slices by version, then cut the slice cuts by string |
| 17:27 | <Philip`> | Or do three separate bar charts |
| 17:28 | <gsnedders> | And what about invalid values? :P |
| 17:29 | <Philip`> | Create an "unknown" category, since they're not really invalid since there's no syntax constraints |
| 17:30 | <gsnedders> | No, it must be a token :P |
| 17:32 | gsnedders | finds bug in RFC2616 |
| 17:32 | <Philip`> | That says it must be >= 1 token, I think |
| 17:32 | <Philip`> | but its definition of the meaning of the tokens is incompatible with reality |
| 17:33 | <gsnedders> | actually, it's unclear only |
| 17:33 | <gsnedders> | with 1*(product-token | comment) are you allowed implied *LWS? |
| 17:33 | <gsnedders> | I guess they count as two adjacent words |
| 17:33 | <Philip`> | I thought implied LWS was everywhere where it's not explicitly disallowed |
| 17:34 | <gsnedders> | yeah |
| 17:34 | <gsnedders> | it's anywhere between two adjacent words. |
| 17:34 | <gsnedders> | but is a repetition two words? |
| 17:34 | gsnedders | shrugs |
| 17:34 | <Philip`> | Parsing it as tokens is silly since "(compatible;" isn't a useful token |
| 17:35 | <gsnedders> | that's a comment! |
| 17:35 | <Philip`> | Oh, it is? |
| 17:35 | <gsnedders> | (comment) |
| 17:35 | <Philip`> | Oh, okay |
| 17:36 | <gsnedders> | that's perfectly valid |
| 17:36 | <gsnedders> | and it's why ( and ) aren't allowed in token |
| 17:37 | <Philip`> | It makes more sense when I actually read what it says |
| 17:39 | <gsnedders> | That's normal. |
| 17:48 | <gsnedders> | Philip`: http://http-parsing.gsnedders.com/#first-server-product-tokens |
| 17:48 | <gsnedders> | <http://http-parsing.gsnedders.com/#first-server-product-token>; even |
| 17:50 | <Philip`> | gsnedders: The scroll position gets messed up after it creates the graphics :-( |
| 17:50 | <gsnedders> | in Saf3/Mac it's fine :P |
| 17:51 | <Philip`> | Are people intentionally sticking with Apache/1.3.37 for as long as possible? |
| 17:52 | <gsnedders> | we have two responses from Apache/1.2.6. quite complaining. |
| 17:52 | <gsnedders> | *quit] |
| 17:52 | <gsnedders> | **quit |
| 17:55 | <gsnedders> | why on earth would anyone send ;charset=none |
| 19:23 | <zcorpan> | http://tinyurl.com/2fa6ko - html5 content model sketch |
| 19:23 | <zcorpan> | Hixie: ^ |
| 19:38 | <gsnedders> | zcorpan: could you possibly use data URIs for anything more crazy? :) |
| 19:38 | <zcorpan> | gsnedders: hmm, how about video? :) |
| 19:39 | <gsnedders> | zcorpan: I'll take that as a yes :) |
| 19:40 | <hsivonen> | hmm. DOM replaceChild has weird argument order |
| 19:41 | <gsnedders> | empty header values are oddly popular. |
| 19:41 | <hsivonen> | (and it's in general weird compared to child.replace(replacement)) |
| 19:42 | <hsivonen> | the tiny url doesn't work in Firefox 2 |
| 19:42 | <hsivonen> | security? |
| 19:42 | <zcorpan> | hsivonen: yeah, i never remember what the right order is of the arguments (re replaceChild) |
| 19:42 | <zcorpan> | hsivonen: wfm in firefox 3 |
| 19:43 | <gsnedders> | Saf3/Mac too |
| 19:44 | <kingryan> | doesn't work in camino |
| 19:44 | <hsivonen> | doh. I'm missing collapsing-relevant styles on validator.n |
| 19:44 | <hsivonen> | u |
| 19:45 | <gsnedders> | Philip`: <http://www.nightwing.com.au/>; — that has a header to outdo all other headers. |
| 19:46 | <kingryan> | gsnedders: the empty one? |
| 19:46 | <gsnedders> | yeah |
| 19:46 | <gsnedders> | kingryan: we were talking about headers without values earlier |
| 19:46 | <zcorpan> | hsivonen: now renumbering works correctly |
| 19:47 | <zcorpan> | hsivonen: aha, was that why renumbering didn't work before? |
| 19:48 | <hsivonen> | zcorpan: I don't know |
| 19:48 | <zcorpan> | hsivonen: that would make sense |
| 19:48 | <hsivonen> | zcorpan: I didn't fix anything else yet in deployment |
| 19:48 | <Philip`> | gsnedders: http://krijnhoetmer.nl/irc-logs/whatwg/20071213#l-764 - old news :-p |
| 19:48 | <gsnedders> | Philip`: you found that one too. damn. |
| 19:49 | <hsivonen> | my sandbox copy has not O(n) rewritten message grouping but no collapsing |
| 19:49 | <hsivonen> | s/not/now/ |
| 19:49 | <hsivonen> | there should probably be collapsing of the locations, too |
| 19:50 | <zcorpan> | hsivonen: it would be cool (and easy to impl :) ) if the sub lists had the same numbers as they would have when not collapsed, if you're going to do sub lists |
| 19:51 | <hsivonen> | zcorpan: hmm. I hadn't thought of that but OK |
| 19:51 | <hsivonen> | If I collapse automatically after grouping, I need to provide an "expand" all feature as well as "ungroup"... |
| 19:52 | <zcorpan> | hsivonen: it would be a side effect of just moving the <li>s into the sub list. the number also gives a hint about where the problem is in the source |
| 19:52 | <hsivonen> | zcorpan: ok. I let the outer list be renumbered |
| 19:52 | <hsivonen> | zcorpan: so the outer list counts the number of different messages |
| 19:52 | <hsivonen> | zcorpan: and I'll make the inner list items have the ungrouped numbers |
| 19:53 | <zcorpan> | hsivonen: ok |
| 19:53 | <hsivonen> | zcorpan: btw, I use innerHTML as the comparison key to avoid innerText/textContent differences |
| 19:54 | <zcorpan> | hsivonen: i don't know, perhaps having 1,2,3 on the inner lists are better anyway |
| 19:54 | <zcorpan> | hsivonen: ok |
| 19:55 | <hsivonen> | hmm. is the innerHTML getter interoperable in application/xhtml+xml? |
| 19:55 | <hsivonen> | wrong question |
| 19:56 | <hsivonen> | does innerHTML in application/xhtml+xml return something stable in each of Firefox/Safari/Opera? |
| 19:56 | <zcorpan> | stable as in won't change in the future? |
| 19:58 | <hsivonen> | stable as in in the same browser instance return the same string for the same-shaped tree fragment and different strings for different fragments |
| 19:58 | <zcorpan> | yes |
| 19:59 | <zcorpan> | (unless there's a bug i don't know about) |
| 19:59 | <hsivonen> | zcorpan: I implemented list number keeping. it looks a bit weird. I'm not sure if it is useful or just weird |
| 19:59 | <hsivonen> | zcorpan: ok. thanks |
| 19:59 | zcorpan | looks |
| 19:59 | <zcorpan> | deployed? |
| 20:00 | <hsivonen> | zcorpan: not yet |
| 20:00 | <zcorpan> | ok |
| 20:00 | <zcorpan> | i can imagine that it would be weird. feel free to remove it :) |
| 20:01 | <zcorpan> | i had in mind that a given error would have the same number regardless of view, but perhaps it's not that useful |
| 20:01 | <zcorpan> | so that authors could refer to them. "I don't understand error 16! help!" |
| 20:02 | <hsivonen> | zcorpan: I think I'm leaving it in for a while and see what comments I get on IRC |
| 20:03 | <zcorpan> | hsivonen: having the outer list not keeping the numbers probably makes the idea collapse, because then you couldn't refer to a specific error that was in the outer list |
| 20:08 | <hsivonen> | zcorpan: no problem :-) the location of the first error instance is moved to the inner list as well |
| 20:09 | <zcorpan> | hsivonen: ok. but what about errors that are "alone" and then renumbered (in the outer list)? |
| 20:10 | <zcorpan> | or do they also get an inner list? |
| 20:13 | <hsivonen> | zcorpan: they get a single-item inner list |
| 20:13 | <zcorpan> | hsivonen: aha. then it's all good :) |
| 20:14 | <hsivonen> | were the WebKit animatable css properties documented somewhere? in particular, will trasitions to and from display: none; be animatable |
| 20:14 | <hsivonen> | ? |
| 20:15 | <roc_> | there's a thread in www-style with the proposal |
| 20:16 | <zcorpan> | display:none animation would be cool |
| 20:16 | <hsivonen> | roc_: Hyatt's first email doesn't seem to have a list of animatable properties |
| 20:16 | <roc_> | oh |
| 20:17 | <roc_> | I hope 'display' is not animatable |
| 20:17 | <roc_> | that sounds both useless and hard to implement |
| 20:17 | <zcorpan> | some js libraries support display:none animation, sort of |
| 20:17 | <hsivonen> | oh, there's a value-based definition of what's animatable |
| 20:17 | <zcorpan> | roc_: collapsing things is often animated |
| 20:18 | <roc_> | yeah but you can't exactly have a smooth transition from display:block to display:inline |
| 20:18 | <hsivonen> | roc_: when something changes from display: none; to display: block;, it would be cool to have the newly displayed block slide from under the block above |
| 20:18 | <zcorpan> | perhaps collapsing annimation can be faked with 'height' animation (with overflow:hidden) |
| 20:19 | <hsivonen> | but it kinda sucks to have to use overflow: hidden; height: 0; as a surrogate for display:none |
| 20:19 | <zcorpan> | i guess |
| 20:19 | <roc_> | not really |
| 20:19 | <zcorpan> | a lot of things in css suck :) |
| 20:19 | <roc_> | animating to display:none is under-constrained |
| 20:19 | <roc_> | which way should the element slide? |
| 20:20 | <hsivonen> | I guess whether animating height sucks depends on how well legacy browsers handle overflow: hidden; |
| 20:20 | <roc_> | or should it perhaps zoom out from zero size? |
| 20:20 | <roc_> | etc |
| 20:20 | <hsivonen> | roc_: I see the problem. but wanting to animate collapsing will be a common case |
| 20:21 | <hsivonen> | roc_: and currently, display:none is the way things are collapsed |
| 20:23 | <Philip`> | If I remember correctly, Apple's proposal was for 'animation of CSS properties', which is distinct from 'animation of HTML documents using CSS' |
| 20:24 | <hsivonen> | oh great. more ogg email |
| 20:25 | <Philip`> | and so it can't animate nicely between display:block and display:none because there's no intermediate state to animate the property through |
| 20:25 | <Philip`> | as a fundamental part of the feature's design |
| 20:26 | <Philip`> | (and so you'd need something totally different to get the desired effect) |
| 20:36 | <hsivonen> | interesting. in Firefox HTML DOM mode, elements created with createElementNS don't get the DOM uppercasing tagName treatment |
| 21:53 | <Hixie> | well last night's ogg flame wars sure were more polite than the previous day's |
| 21:53 | <Dashiva> | Is that RoggO guy still around? |
| 21:55 | <Hixie> | yeah, a little |
| 22:04 | <hsivonen> | now another commentator has inadvertently revealed not havin read the video part of the spec |
| 22:05 | <hsivonen> | I wonder what would be the right UI placement for a "group messages" button |
| 22:05 | <hsivonen> | clearly, the "Collapse All" button is now in the wrong place |
| 23:12 | <Hixie> | zcorpan: i think your proposal is pretty good |
| 23:12 | <Hixie> | i think i'm going to take this opportunity to discard the "block" and "inline" terms though |
| 23:12 | <Hixie> | in favour of something like "paragraph-level" and "phrasing-level" or something |
| 23:14 | <Hixie> | i don't know that we want video to be strictly speaking inline/phrasing-level... after all it has to contain blocks (i agree it's transparent)... consider this: |
| 23:15 | <Hixie> | <section> The screen should now look like <object ...> <pre> <samp> ... </samp> </pre> </object> which... |
| 23:15 | <Hixie> | i suppose it's ok for paragraphing to be loosely defined |
| 23:16 | <Hixie> | in fact i guess we can throw all of that to the winds and just say that you apply css rules to determine "blocks" for the purposes of rendering, even if you don't support css itself |
| 23:18 | <Hixie> | *ponder* *ponder* |
| 23:31 | <jruderman> | Hixie: neat, so "block in inline" will unambiguously refer to the CSS layout issue rather than the HTML parsing issue |
| 23:58 | <Hixie> | jruderman: yeah |
