| 00:34 | <Philip`> | IE's mode setting thing doesn't interact well with the Live DOM Viewer |
| 00:39 | <Philip`> | Hmm, the <svg xmlns=foo> things the second time you use it on a page |
| 00:39 | <Philip`> | Uh |
| 00:39 | <Philip`> | s/things/thing works/ |
| 00:41 | <Philip`> | ...but when I say "works", I mean that in basically the opposite sense to usual, because it breaks the parsing of everything after the second usage |
| 00:43 | <Hixie> | i really am very disappointed in their insistence on continuing this xmlns= crap in text/html |
| 00:43 | <Hixie> | the more they do with this, the more screwed we are in html5 in terms of adding namespaces |
| 00:43 | <Hixie> | though maybe we can just do ns= or something |
| 00:51 | <othermaciej> | htmlns |
| 00:51 | <Philip`> | The more annoying thing is how simply using colons (with no xmlns or anything) causes occasional XML-like parsing |
| 00:52 | <Hixie> | well that's ok |
| 00:52 | <Hixie> | don't use colons :-) |
| 00:52 | <Hixie> | we've already learnt (the hard way) that prefixes are a horrible idea |
| 00:52 | <Hixie> | so whatever we do end up doing won't have prefixes |
| 00:56 | <tantek> | Hixie, could you please document how you've learned that prefixes are a horrible idea so others won't keep repeating and reproposing that mistake? |
| 00:57 | <Hixie> | yeah i have a half-written blog post on the subject |
| 00:57 | <Hixie> | i add to it occasionally |
| 00:57 | <Hixie> | one day it'll even get finished :-) |
| 00:58 | <Dashiva> | And then someone will point out how all the problems are really just awesome features that we aren't ready to leverage yet |
| 00:58 | <Hixie> | yup |
| 00:58 | <Hixie> | but we don't have to agree with them |
| 00:59 | <Dashiva> | I'm just wondering if there are any people left who don't either know or don't want to know :) |
| 01:13 | <jwalden> | Hixie: regarding <http://lists.whatwg.org/pipermail/whatwg-whatwg.org/2008-March/014156.html>;, the paragraph starting "Fourth", was my understanding of how data: URIs work according to the spec wrong? the code to create the window at a data: location is <http://bonsai.mozilla.org/cvsblame.cgi?file=/mozilla/dom/tests/mochitest/whatwg/test_postMessage_special.xhtml&rev=1.2&mark=266-283#265>;... |
| 01:13 | <jwalden> | ...-- should events posted from that window have .origin be that of the test_postMessage_special.xhtml page? |
| 01:22 | <Hixie> | jwalden: i expect the spec to change regarding the origin of data: URIs |
| 01:24 | <jwalden> | okay |
| 01:24 | <jwalden> | I shall forget I ever had concerns about that part, for now :-) |
| 01:27 | <Hixie> | wait, the spec already seems to be updated |
| 01:27 | <Hixie> | i'm confused |
| 01:27 | <Hixie> | the spec says: |
| 01:27 | <Hixie> | "The origin of a Document or image that was generated from a data: URI found in another Document or in a script is the origin of the Document or script. |
| 01:28 | <Hixie> | " |
| 01:28 | <Hixie> | what is it firefox does? |
| 01:31 | <Hixie> | jwalden: ^ (in case you're not watching teh channel :-) ) |
| 01:32 | <jwalden> | Hixie: I could have sworn last I looked that data: URIs each had a globally unique identifier as their origin; it's possible you changed it since I last looked |
| 01:33 | <jwalden> | (and I tend to watch sporadically or flip here when I get pinged :-) ) |
| 01:33 | <Hixie> | the spec says: |
| 01:33 | <Hixie> | The origin of a Document or image that was generated from a data: URI found in another Document or in a script is the origin of the Document or script. |
| 01:33 | <Hixie> | The origin of a Document or image that was generated from a data: URI from another source is a globally unique identifier assigned when the document is created. |
| 01:33 | <Hixie> | is that what you want it to say? |
| 01:33 | <Hixie> | or do you want it to say something else? |
| 01:34 | <jwalden> | I think that's what Firefox does, which is most convenient for me (I don't have a particular desire one way or another except to avoid work on my part) :-) |
| 01:34 | <jwalden> | I'm guessing the "other source" is, e.g., stuff typed in the location bar |
| 01:35 | <Hixie> | yeah |
| 01:36 | <jwalden> | might be worth mentioning that as an example of the latter, fwiw |
| 01:39 | <Hixie> | send mail |
| 01:39 | <Hixie> | no wait |
| 01:39 | <Hixie> | i have it right here |
| 01:39 | <Hixie> | i can just do that |
| 01:43 | <jwalden> | thanks |
| 02:16 | <jwalden> | I am convinced I have at least a 50% failure rate at remembering to use my +whatwg email address for messages to the list |
| 03:00 | <Hixie> | jwalden: that's why i don't bother with different mail addresses |
| 03:00 | <Hixie> | not to mention what happens when you cross-post |
| 03:01 | <Hixie> | hey have you tested IE8 on your tests? |
| 03:01 | <jwalden> | no, not yet |
| 03:01 | <jwalden> | I need to set up vmware and windows to do so, which will be a small pain |
| 03:01 | <jwalden> | based on what they've written, I "suspect" they implemented the pre-origin API |
| 03:02 | <jwalden> | so far I've only run against WebKit+patch |
| 03:02 | <jwalden> | Opera's latest still has postMessage on document, so it's a bit pointless there |
| 04:20 | <Lachy> | FYI, for anyone using html5.lachy.id.au, my site will soon be transferred to Dreamhost (as soon as the DNS is updated) and some things on on there will be broken due to the move. That (and any other broken things) will be fixed on the weekend. |
| 04:44 | <othermaciej> | I believe their docs described the pre-origin API |
| 05:29 | <jwalden> | Hixie: with pain, I got the tests set up well enough to be able to say that IE breaks on parts of the harness used in the tests, such that the tests can't be run in it -- was getting runtime errors trying to access the |attributes| property of an element |
| 09:01 | <Hixie> | wow, rsayrer is bitter |
| 09:02 | <annevk> | no kidding |
| 09:07 | <virtuelv> | url? |
| 09:08 | <virtuelv> | or are you refering to his mail to pubic-webapi? |
| 09:08 | <annevk> | http://lists.whatwg.org/pipermail/whatwg-whatwg.org/2008-March/014168.html |
| 09:08 | <annevk> | http://lists.w3.org/Archives/Public/public-webapi/2008Mar/0014.html |
| 09:08 | <annevk> | yeah |
| 09:09 | <Hixie> | woah, glazman and plinss are replacing bert |
| 09:09 | <annevk> | CSS versioning ftw! |
| 09:10 | <zcorpan> | hmm? pointer? |
| 09:13 | <annevk> | Hixie, the quotation marks checking replaced the ending quotation mark with a colon... |
| 09:13 | <Hixie> | annevk: hm? |
| 09:13 | <hsivonen> | annevk: CSS versioning? |
| 09:13 | <zcorpan> | Hixie: "<code title="">false</code>: |
| 09:14 | <Hixie> | oops |
| 09:14 | <annevk> | hsivonen, glazou likes that |
| 09:15 | <zcorpan> | @DOCTYPE CSS PUBLIC "-//W3C//DTD CSS 2.1//EN"; |
| 09:15 | <hsivonen> | annevk: interesting. I don't like even strict levels. (setting level to 3 and getting errors for Prince extensions sucks) |
| 09:17 | <annevk> | Hixie, also "<code title="">true</code>: |
| 09:17 | <annevk> | twice |
| 09:23 | <Hixie> | fixed |
| 09:23 | <Hixie> | anyone got a tool that gets true type font metrics? |
| 09:53 | <jgraham_> | zcorpan: Oops I entirely missed your message about aria before I replied |
| 09:54 | Hixie | finally finds a ttfdump tool on microsoft's site and determines that the Ahem ex height is 0.8em as he suspected |
| 10:15 | <Hixie> | hm, i wonder if lachy is coming back |
| 10:15 | Hixie | summons Lachy |
| 10:17 | Hixie | sacrifices a chicken: |
| 10:17 | <Hixie> | . , |
| 10:17 | <Hixie> | ('>) |
| 10:17 | <Hixie> | LL ..'o |
| 10:17 | <Hixie> | eep, my second line was misindented |
| 10:17 | <Hixie> | now my summoning ritual wil go wrong! |
| 10:17 | Hixie | runs away |
| 10:18 | zcorpan | wonders what will happen instead of Lachy coming back |
| 10:27 | <Hixie> | woot, it worked after all |
| 10:27 | <Hixie> | Lachy: do you still think contentEditable="" is non-semantic? |
| 10:28 | <Lachy> | I really don't like the way its designed |
| 10:28 | <Hixie> | "designed" is putting it nicely |
| 10:28 | <annevk> | Hixie, hey, did you stop solving parsing issues? |
| 10:28 | <annevk> | :( |
| 10:28 | <annevk> | heh |
| 10:28 | <Hixie> | annevk: were there specific ones you wanted me to resolve that i didn't resolve? |
| 10:29 | <Hixie> | annevk: the parsing folders were down to just a few issues |
| 10:29 | <annevk> | forms :) |
| 10:29 | <Lachy> | s/designed/put together in an ad hoc way that focuses more on presntational editing rather than semantics |
| 10:29 | <annevk> | nested forms, to be specific |
| 10:29 | <Hixie> | annevk: oh, oops, i forgot about that one |
| 10:29 | <Hixie> | annevk: will do that after i've done this contentEditable e-mail |
| 10:30 | <annevk> | cool |
| 10:30 | <Hixie> | annevk: remind me if i don't after i've sent it :-) |
| 10:30 | <Hixie> | Lachy: yeah, no disagreement from me there (i've tried hard to work around that in html5) |
| 10:30 | <Lachy> | I will have to review the spec. It's been a while since I even looked at content editable |
| 10:31 | <Hixie> | k |
| 10:31 | <Hixie> | me tyoo |
| 10:31 | <Hixie> | too even |
| 10:45 | <hsivonen> | wow. visitmix.com suggests I might use VLC to view their stuff as an alternative to Silverlight |
| 10:49 | <Lachy> | Hixie, just reviewing it now. It's not particularly clear about this markup: <p contenteditable="true">...<span contenteditable=false>is this editable?</span>...</p> |
| 10:49 | <Hixie> | it is not |
| 10:49 | <Lachy> | in particular, this paragraph is a little confusing about that. "If an HTML element has a contenteditable attribute set to the true state, or if its nearest ancestor HTML element with the contenteditable attribute set has its attribute set to the true state, or if it has no ancestors with the contenteditable attribute set but the Document has designMode enabled, then the UA must treat the element as editable (as described below). " |
| 10:50 | <Lachy> | since that span falls into the "or if its nearest ancestor HTML element with the contenteditable attribute set has its attribute set to the true state" condition. |
| 10:50 | <annevk> | is the contenteditable section already updated to match browsers? i think my outstanding comment on that hasn't been addressed yet |
| 10:51 | <annevk> | in particular, "The false keyword maps to the false state, which is also the invalid value default. " is wrong |
| 10:52 | <annevk> | invalid values are equivalent to the attribute being missing |
| 10:54 | <Hixie> | Lachy: hm yeah, that needs fixing. |
| 10:54 | <Hixie> | annevk: haven't gotten to that e-mail yet |
| 10:55 | <annevk> | kk |
| 10:55 | <Lachy> | Hixie, should I send mail about that, or will you just remember that? Are you going to be editing that section today? |
| 11:17 | <Hixie> | i'll record it here |
| 12:07 | <hsivonen> | looks like html5lib does not have test coverage for spec rev 1320 |
| 12:07 | annevk | looks |
| 12:07 | <annevk> | hsivonen, true |
| 12:08 | <annevk> | hsivonen, actually, I did check changing the implied end tags stuff and several fragment tests started reporting different amount of errors |
| 12:09 | <hsivonen> | annevk: I'd be more interested in catching the tree change arising from the different confidion for generate implied ond tags |
| 12:10 | <hsivonen> | but for some reason, it isn't quite obvious to me what the right test case would be |
| 12:10 | <annevk> | I don't think there's a difference for that |
| 12:10 | <annevk> | I also did not look at the </form> change |
| 12:11 | <hsivonen> | oh. right. However, there does seem to be a difference with popping stuff |
| 12:12 | <hsivonen> | unless I'm missing something important |
| 12:12 | <hsivonen> | oops. |
| 12:12 | <annevk> | are you missing that <td> etc. never work in "in body"? |
| 12:12 | <hsivonen> | I was missing something important |
| 12:13 | <hsivonen> | nothing to see here |
| 12:13 | <hsivonen> | annevk: I was missing the details of me own stack search optimization |
| 12:13 | <hsivonen> | s/me/my/ |
| 12:35 | <hsivonen> | is it recommended to keep the table tainting data on the element stack or to keep a smaller separate taint stack? |
| 12:36 | <Hixie> | i would just toggle a bit on the element itself |
| 12:37 | <annevk> | i've done it on the element stock, as Hixie said |
| 12:37 | <annevk> | that seems to work good for nested tables, etc. |
| 12:37 | <Hixie> | i hate the tainted stuff |
| 12:37 | <Hixie> | but i didn't see another solution |
| 12:38 | <hsivonen> | but that causes excessive stack traversal, doesn't it? |
| 12:38 | <hsivonen> | compared to constant-time inspection of the top of a taint stack |
| 12:39 | <annevk> | your idea would require making sure the taint stack is always correct |
| 12:39 | <annevk> | which might be tricky with nested tables |
| 12:39 | <annevk> | <table>X<tr><td><table></table></td> </table> |
| 12:39 | <hsivonen> | that depends on how many different cases there are where a table comes off the stack |
| 12:40 | <annevk> | that becomes "X <table>..." iirc |
| 12:40 | <annevk> | though spaces in the inner <table> should not be moved |
| 12:40 | <Hixie> | oh you really mean how do you keep track fo the "current table" |
| 12:40 | <Hixie> | and the answer is, yeah, optimise it to a separate stack |
| 12:41 | <Hixie> | i believe you can only pop a table using <table> or </table>, but i may be wrong |
| 12:41 | <annevk> | that could work, yes |
| 12:41 | <hsivonen> | hmm. let's see if an unoptimized impl. is good enough |
| 12:41 | <annevk> | not done in our python impl |
| 12:42 | <annevk> | hmm, two implementations not doing namespaces in the Selectors API so far... |
| 12:42 | <annevk> | hixie might have been on to something :) |
| 12:43 | <Hixie> | :-) |
| 12:44 | <Hixie> | btw i think you may have been onto something with dropping multiple views |
| 12:44 | <Hixie> | if you want support for that, i'm happy to give it to you |
| 12:44 | <annevk> | i currently specced it, deferring definitions of "view" and "default view" to HTML 5 :) |
| 12:45 | <Hixie> | hah |
| 12:52 | <Hixie> | bed time |
| 12:52 | <Hixie> | nn |
| 12:53 | <annevk> | g'n |
| 14:31 | <Philip`> | Hixie: http://philip.html5.org/data/encoding-detection.svg shows the number of bytes required for the encoding detector |
| 14:33 | <annevk> | nice |
| 14:33 | <Dashiva> | What's the value at 4096? |
| 14:33 | <annevk> | the SVG looks pretty complex though |
| 14:35 | <Philip`> | Dashiva: 99.2%, I think |
| 14:38 | <Philip`> | annevk: Blame Gnuplot :-p |
| 14:39 | <Philip`> | The biggest jump seems to be around 90, which is where people are saying <html> <head> <meta http-equiv="Content-Type" content="text/html; charset=windows-1252"> etc |
| 14:40 | <Philip`> | The worst was http://www.the-movie-times.com/thrsdir/actress/actressProfiles.mv?gclose which requires 90KB |
| 14:41 | <Dashiva> | Starts with a close tag, nice |
| 14:42 | <Dashiva> | And two sets of <html><head>etc... |
| 14:44 | <Philip`> | The are some other small spikes where people are doing doctype + html + head + meta |
| 14:44 | <Philip`> | which all makes sense and is not unexpected |
| 14:46 | <hsivonen> | Philip`: I don't see any line in the .svg in Firefox 3 or in Safari 3 |
| 14:47 | <hsivonen> | it's another one of those SVG pages that makes safari autoquit on me |
| 14:47 | <hsivonen> | Opera renders it. weird |
| 14:47 | <annevk> | currentColor bug in Firefox maybe? |
| 14:48 | <annevk> | it uses color:red; stroke:currentColor; to render the red color |
| 14:48 | annevk | remembers Firefox had such a bug |
| 14:48 | <Philip`> | hsivonen: Works for me in 2008030604 Minefield/3.0b5pre |
| 14:48 | <hsivonen> | that would explain it |
| 14:48 | <Philip`> | and in 2.0.0.12 |
| 14:49 | <hsivonen> | oh. my build is out of date |
| 14:49 | <hsivonen> | I'm using beta3 |
| 14:49 | <hsivonen> | seems to be the latest beta still, though |
| 14:50 | <Philip`> | Oops, my encoding detector didn't ignore comments |
| 14:50 | Philip` | wonders if there are test cases somewhere |
| 14:51 | <Philip`> | Oh, they're in a hard to parse format so I won't bother |
| 14:52 | <Philip`> | and they don't match the current spec |
| 14:53 | <annevk> | hsivonen, you should download a nightly then i guess |
| 14:57 | <Philip`> | I see BOMs on zero pages |
| 15:01 | <annevk> | is there an easy way to get a non-HTML element in IE? |
| 15:02 | <zcorpan> | colon in tag name? |
| 15:05 | <annevk> | if anyone has a better name for boolean = window.media.query(mediaquery) please let me know |
| 15:05 | <annevk> | i'm thinking of isQueryApplicable or something like that |
| 15:06 | <annevk> | or maybe it should be boolean window.media.supportsMedia(mediaquery) |
| 15:07 | <annevk> | s/Media/Medium/ |
| 15:07 | <annevk> | let me know |
| 15:07 | <annevk> | supportsMedium seems best so far, quite consistent with CSSOM too |
| 15:08 | <Philip`> | Should HTML5 look for e.g. <?xml version="1.0" encoding="ISO-8859-2"?> when finding the encoding? |
| 15:08 | <Philip`> | That would make e.g. http://bajkoland.com.pl/ work correctly |
| 15:08 | <aroben> | annevk: window.media.matchesQuery(mediaQuery) ? |
| 15:08 | <zcorpan> | hasQuery? |
| 15:09 | <annevk> | reason for the suggestion to remove query is http://dev.w3.org/csswg/cssom/#medialist btw |
| 15:10 | <annevk> | appendMedium / deleteMedium |
| 15:10 | <annevk> | aroben, yeah, something like that sounds better |
| 15:10 | <annevk> | so either matchesQuery or matchesMedium |
| 15:11 | <aroben> | annevk: is there a difference between a medium and a query? |
| 15:12 | <annevk> | medium takes a media query |
| 15:12 | <annevk> | that would be the case for either API |
| 15:12 | <annevk> | it's just that medium is already used |
| 15:13 | <aroben> | annevk: still not sure I understand the distinction |
| 15:13 | <annevk> | there's none, it's just a question of what name would be better |
| 15:14 | <aroben> | annevk: window.media implements MediaList? |
| 15:14 | <annevk> | noo |
| 15:14 | aroben | clearly needs to do some reading |
| 15:15 | <qwert666> | Hi |
| 15:15 | <annevk> | my idea was that since MediaList.appendMedium already takes a media query it might make sense to call the x in window.media.x matchesMedium |
| 15:15 | <annevk> | so that authors get somewhat consistently named APIs |
| 15:17 | <aroben> | annevk: seems like it should be "matchMedium" since "media" is plural |
| 15:17 | <aroben> | annevk: though that sounds like of strange too |
| 15:18 | <annevk> | yeah, but it's probably better than just query() |
| 15:20 | <aroben> | annevk: agreed, especially since query is both a noun and a verb |
| 15:21 | <zcorpan> | Philip`: i wouldn't like it, but it might be interesting to know how much it fixes and how much it breaks |
| 16:04 | <annevk> | Philip`, don't really like that either and I believe that IE doesn't do it |
| 16:04 | <annevk> | otoh, if it makes more sites work... grmbl |
| 16:16 | <Philip`> | http://msdn2.microsoft.com/en-us/library/cc197026(VS.85).aspx - "Other browsers may return strings specific to themselves, for example "Opera" or "Mozilla"." |
| 16:16 | <Philip`> | At least they're not suggesting to use "Firefox" and making other Gecko users unhappy |
| 16:17 | <annevk> | i think for a while we said IE there |
| 16:17 | <Philip`> | In DocumentCompatibleInfo? |
| 16:18 | <annevk> | for userAgent |
| 16:18 | <Philip`> | That seems to be API for the new X-UA-Compatible thing |
| 16:18 | <annevk> | maybe I'm misunderstanding something |
| 16:18 | <annevk> | oh ok |
| 16:18 | <annevk> | ooh |
| 16:18 | <annevk> | X-UA-Compatible is evil |
| 16:19 | <Philip`> | (found via http://msdn2.microsoft.com/en-us/library/cc304071(VS.85).aspx ) |
| 16:20 | <annevk> | :after Pseudo-class lol |
| 16:21 | <Lachy> | http://www.wpdfd.com/issues/86/html_the_foundation_of_the_web/ |
| 16:21 | <Lachy> | that's a really good article |
| 16:24 | <Philip`> | http://msdn2.microsoft.com/en-us/library/cc288123(VS.85).aspx - "IHTMLWindow6" - they do quite like versioning |
| 16:28 | <annevk> | minimalitis ftw |
| 16:31 | <zcorpan> | yeah, minimalitis is cool ;) |
| 17:06 | <annevk> | hmm, http://intertwingly.net/blog/2008/03/07/Design-By-Attrition#c1204908912 is getting too cryptic |
| 17:06 | annevk | -> supermarket |
| 17:10 | <svl> | invite Sam to help edit? |
| 17:15 | <Philip`> | It's comparatively easy for IE to support HTML namespaces, since they don't support XHTML and so they can (and do) have a non-XML-DOM API for their namespaces, whereas other browsers want to keep HTML and XHTML as similar as possible |
| 17:59 | <annevk> | svl, edit what? |
| 18:00 | <annevk> | Philip`, the other problem is that pages depend on how <foo:bar> works |
| 18:00 | <annevk> | live.com uses IE namespaces for instance if I remember correctly and relies on other browers to do things in a different way |
| 18:04 | <Philip`> | annevk: I assume it's depending mostly on how <foo:bar> works in the DOM, rather than just on how it's parsed |
| 18:04 | <annevk> | also on how it's parsed |
| 18:04 | <annevk> | <html xmlns:web |
| 18:05 | <Philip`> | Does it rely on <ns:foo/><ns:bar/> being adjacent in IE but being parent-child in other browsers? |
| 18:05 | <annevk> | i wouldn't expect that, but I don't know |
| 18:06 | <Philip`> | How else would it depend on how it's parsed? |
| 18:06 | <annevk> | I missed "just" in your above sentence btw |
| 18:06 | <annevk> | so I guess I agree :) |
| 18:22 | annevk | tries asking one more time about where all this "namespaced" HTML content will come from |
| 18:29 | <svl> | annevk: The way I interpreted Sam's "us" and "we" remark was that even though there's strong requests coming from the field, whatwg editors (e.g. you and hixie) seem to be ignoring it. |
| 18:29 | <svl> | And even though you're not ignoring it, and merely don't see a good solution yet, you also don't have time to really spend on finding a solution, on account of all the other things you're editing, and so the final specification will (in his view) be decided by inactivity, rather than design. |
| 18:29 | <svl> | The big bottleneck is lack of editors (see also that google geolocation api proposal). So maybe that's something where he can help out. I don't know how any spec for namespaces in HTML would be integrated or split off from other specs - but that seems to me to be something which can be worked out. |
| 18:34 | <annevk> | well, you see to indicate that it's not a lack of editors, but of researches :) |
| 18:45 | <svl> | I honestly don't know; I'm just a humble outsider with not nearly enough domain knowledge. Sam seems to believe that there's been enough research to move forward, though. I mostly think that rather than seeing you two clash on this, his strong belief in the necessity of a spec here is something which could be used to actually move things forward; entice him into actually doing the research... |
| 18:45 | <svl> | ...that you believe is still needed and he doesn't see (yet?). Which is probably easiest when starting from a concrete proposal... |
| 18:47 | <annevk> | i don't get how he goes from demand for silverlight to demand for namespaced HTML |
| 18:48 | <svl> | Does that really matter, though? |
| 18:49 | <annevk> | if there's no demand, why should we add it? |
| 18:54 | <svl> | Because there's small scale demand (e.g. the experiments Sam is doing himself), and enough people seeing opportunity which they believe will almost certainly translate into demand? Because it can't hurt to have other people explore options more fully? |
| 18:55 | <svl> | I understand it won't have priority for you personally if you don't see that demand, but you can still enable forward movement... |
| 19:01 | svl | guesses he's basically trying to encourage you to say something like Jeff did in his latest comment. |
| 19:01 | <svl> | Anyway, I'm out for a bit. |
| 19:02 | <annevk> | kk |
| 19:02 | annevk | ponders |
| 19:04 | <Philip`> | Does http://philip.html5.org/misc/xmlns-dom.html get double-spaced lines in IE6/IE7 on Windows? |
| 19:04 | <Philip`> | (It does in IE8 on Windows, but not in IE6 on Linux) |
| 19:04 | <Philip`> | (and not in Firefox or Opera on Linux) |
| 19:04 | <billmason> | No, I'm seeing single space in IE6/Windows XP. |
| 19:05 | <billmason> | Double space in IE8 in IE7 emulation mode, which is the only IE7 I have at hand. |
| 19:06 | <Philip`> | Okay, thanks |
| 19:06 | <billmason> | np |
| 19:06 | <annevk> | single space on Linux |
| 19:06 | <annevk> | IE7 |
| 19:07 | <Philip`> | Does IE8's IE7 emulation actually work like IE7, or does it work the same as IE8 with the IE7 meta compatibility flag? |
| 19:07 | <annevk> | firefox is fixing da parsing bug |
| 19:07 | <annevk> | yay yay yay |
| 19:07 | <Philip`> | (where "IE8 with the IE7 meta compatibility flag" includes various bugs introduced in IE8, so it's not just a clone of the IE7 engine) |
| 19:08 | <billmason> | Not sure; haven't really hammered on it that much yet. |
| 19:08 | <annevk> | i'd expect IE8 with compat flag |
| 19:08 | <annevk> | otherwise they'd have 5 engines |
| 19:09 | <Philip`> | (Try <meta http-equiv=x-ua-compatible content=ie=7><foo xmlns=bar> to see the bug I'm thinking of) |
| 21:56 | annevk | wonders if <datagrid> is too generic |
| 22:01 | <Hixie> | how so? |
| 22:01 | <Hixie> | it's basically <xul:treegrid> |
| 22:02 | <Hixie> | Philip`: dude that graph is the awesomeness |
| 22:02 | <Hixie> | and it seems 512 was a particularly good choice |
| 22:03 | <Hixie> | but i think we'll increase it to 2048 |
| 22:03 | <Hixie> | Philip`: can you make a lin-log version of that graph? (log vertically) |
| 22:04 | <Philip`> | I suppose a useful amount to wait for is the first one or two response packets, but there's no way to tell how that corresponds to document size because the headers are variable length |
| 22:05 | <Hixie> | well in practice i doubt browsers are going to pay attention to our recommendation |
| 22:05 | <Hixie> | personally i would just use whatever data you have in the first 150ms or so |
| 22:05 | <Hixie> | maybe even less |
| 22:06 | <Hixie> | that is, i would only wait as long as the wait doesn't add real overhead to the render time |
| 22:06 | <annevk> | i don't quite get that concern because until you have data how can you render? |
| 22:07 | <annevk> | re: datagrid, it seems to be for tree and table like widgets, but maybe that's ok |
| 22:07 | <gsnedders> | Philip`, Hixie: what graph? |
| 22:07 | <Philip`> | Hixie: Something like http://philip.html5.org/data/encoding-detection-log.svg ? |
| 22:08 | <Philip`> | gsnedders: http://philip.html5.org/data/encoding-detection.svg |
| 22:08 | <Dashiva> | 0.001 is a really useful point :) |
| 22:16 | <Philip`> | Dashiva: That's still an integer number of pages :-) |
| 22:17 | <annevk> | how many sites without the right <meta> ? |
| 22:18 | <Philip`> | What is "the right <meta>"? |
| 22:19 | <annevk> | <meta content=...charset...> or <meta charset=...> |
| 22:20 | <Philip`> | 126989 total, 89541 with <meta content>, 171 with <meta charset> |
| 22:21 | <Philip`> | Sniffer apparently finds 89695 |
| 22:21 | <Philip`> | (with <meta something>) |
| 22:25 | <Philip`> | (The ones not found by the sniffer include <meta http-equiv="Content-Type"content="text/html; charset=windows-1252">, <meta http-equiv="Content-Type"content="text/html; charset=iso-8859-2">, <meta http-equiv="Content-Type"content="text/html; charset=iso-8859-1"> and <meta http-equiv="Content-Type"content="text/html; charset=iso-8859-1"> |
| 22:25 | <Philip`> | so, uh, the first four I looked at are all exactly the same problem |
| 22:25 | Philip` | smells a pattern |
| 22:29 | <Hixie> | Philip`: hmm, i guess i should have suggested log on the horizontal axis |
| 22:32 | <Philip`> | Hixie: Like http://philip.html5.org/data/encoding-detection-log.svg (reloaded)? |
| 22:33 | <Hixie> | yeah that's more like it |
| 22:33 | <Hixie> | thanks |
| 22:33 | <Dashiva> | What if you made it inverse log towards 100 |
| 22:33 | <Hixie> | 512 really was a good choice |
| 22:33 | <Hixie> | completely by chance though :-) |
| 22:34 | <Hixie> | me posted on sam's blog |
| 22:34 | <Philip`> | Dashiva: Exponential? |
| 22:35 | <Dashiva> | Yeah |
| 22:38 | <Philip`> | I'm not entirely sure how to do that sanely |
| 22:38 | <Philip`> | You can just zoom in on the graph anyway :-p |
| 23:07 | <jgraham> | couldn't you just do 100-% of sniffable pages and then make the graph log-log? |
| 23:09 | <Hixie> | sorry Philip`, didn't mean to start something :-) |
| 23:09 | <annevk> | http://intertwingly.net/blog/2008/03/07/Design-By-Attrition#c1204929970 |
| 23:12 | <jgraham> | I don't get that thread. Usually Sam seems amenable to pitching in when he wants a problem solved. Here less so. |
| 23:13 | <annevk> | all i get is snappy responses and no real answers |
| 23:13 | <annevk> | i won't bite next time, i hope |
| 23:14 | Hixie | replies |
| 23:14 | <Hixie> | i don't get it though |
| 23:14 | <Hixie> | he keeps bringing this up |
| 23:14 | <Hixie> | but never responds to requests for clarification |
| 23:15 | <Philip`> | I suppose http://philip.html5.org/data/encoding-detection-loglog.svg almost makes sense, though it hits a rendering bug in KPDF |
| 23:15 | <Philip`> | I'm not sure what that graph is useful for, though |
| 23:17 | <jgraham> | Well it tells you that for every factor of 2 you increase the number of bytes above ~256 you get an approximately equal decrease in the number of incorrectly sniffed pages |
| 23:17 | <jgraham> | So there is no obvious sweet spot |
| 23:20 | <annevk> | Hixie, do you know that if you use min(x, x) syntax in your spec you need some kind of reference? |
| 23:20 | <Hixie> | hm? |
| 23:20 | <Philip`> | min(x, x) seems a quite verbose way of writing x |
| 23:21 | <annevk> | min(x, max(y, z)) |
| 23:21 | <annevk> | is using such constructs acceptable mixed with prose or is it considered pseudo-code? |
| 23:22 | <Philip`> | Is x+y pseudo-code? |
| 23:22 | <Hixie> | annevk: oh you are asking whether i think one needs a reference, sorry, i misunderstood what you were asking |
