| 00:21 | <hyuuu2> | Hixie: what is the limit size per domain on that |
| 01:12 | <Hixie> | hyuuu2: depends on the browser, but spec says ~5MB |
| 01:13 | <olliej> | hyuuu2: In Safari it starts out at 5mb per domain, but if more is needed the users is asked if more should be allowed |
| 01:48 | <hyuuu2> | Hixie, olliej : thanks |
| 02:57 | <MikeSmith> | blooberry: you around? |
| 03:09 | <MikeSmith> | http://dev.opera.com/articles/view/mama/ |
| 03:25 | <roc> | "Approximately 85% of all of MAMA's pages would be rendered in browsers using their "Quirks" modes." |
| 03:25 | <roc> | does that mean "Quirks and Almost-Standards modes"? |
| 03:35 | <MikeSmith> | roc: does Opera have an Almost-Standards mode? |
| 03:35 | <roc> | I expect so |
| 03:35 | <roc> | those Transitional doctypes better be triggering some kind of mostly-standards-mode rendering |
| 03:38 | <MikeSmith> | roc: so about whether it includes both modes, blooberry would be the guy to ask, I guess |
| 05:35 | <blooberry> | roc: that only includes quirks mode. The rest (15%) are in almost standards or standards. |
| 05:35 | <roc> | that's strange, because a couple of days ago Hixie told me 50% of the Google index was standards or almost-standadrs |
| 05:36 | <blooberry> | roc: I used this chart to judge and built up the counts of the doctypes in the URLs encountered: http://hsivonen.iki.fi/doctype/ |
| 05:37 | <blooberry> | roc: Google's URL set will differ from the URL set I used...it definitely has more deep-URL representation |
| 05:37 | <blooberry> | that could make a difference. |
| 05:38 | <roc> | surprisingly large difference though |
| 05:38 | <blooberry> | yes. 8-} |
| 05:38 | <roc> | I don't know who's right, it's just interesting |
| 05:39 | <blooberry> | I think so too on both counts. The DMoz URL set has its known issues. I'd love to hear more about some statistics regarding google's set (do they publish such things?) |
| 05:40 | <blooberry> | I guess I should have expected people might see those articles already...those weren't supposed to "officially" be released until a few hours from now. Guess my blog post will have to wait until after my nap. |
| 05:43 | blooberry | hopes the couple of remaining issues are resolved before I wake up again |
| 05:45 | <hsivonen> | blooberry: my page doesn't document thesniffing algorithm. it just has data to show to authors |
| 05:46 | <hsivonen> | blooberry: it might be more accurate to apply the algorithm from the HTML5 spec |
| 05:46 | <hsivonen> | (which is a fusion of what Opera, Gecko and WebKit do) |
| 05:47 | <blooberry> | hsivonen: I thought it was a good guide at the time I made the comparison...I was looking for some way to judge adherence to the spirit of the whole standards/almost/quirks debate. |
| 05:47 | <MikeSmith> | blooberry: as far as the timing, the problem with dev.opera.com is that the feed often gets updated before any announcement gets posted to the dev.opera.com home page |
| 05:48 | <blooberry> | our dev.opera gatekeep was up until about 3 in the morning his time fixing a bunch of bugs that were propagated with the posting, so fixing the couple validation errors on the validation article (oops!) will have to wait until morning. ;-} |
| 05:48 | <blooberry> | s/gatekeep/gatekeeper/ |
| 05:49 | <blooberry> | mikesmith: I was surprised at the limitations of the dev.opera site, but I guess timeliness is not one of its problems. 8-D |
| 05:49 | <MikeSmith> | heh |
| 05:50 | <MikeSmith> | blooberry: anyway, congratulations on putting this stuff together.. it looks great so far, and I'm looking forward to seeing the other specific reports (e.g., the markup on) |
| 05:50 | <blooberry> | there are 16 chapters/sections to the markup topic. (hold on to your hat) |
| 05:51 | <blooberry> | I'll start small next week covering HTTP headers, and then move on to markup, I think |
| 05:52 | <MikeSmith> | wow |
| 05:52 | <blooberry> | In some ways it was a lot of fun to gather this data, because I was able to poke into a lot of areas that I hadn't seen poked before. |
| 05:52 | <blooberry> | being able to correlate and compare different factors just adds to the festivities. 8-} |
| 05:55 | blooberry | heads for the pillow |
| 06:32 | <hsivonen> | Dashiva: I don't know about the cost of bytecodes except that the 8001st bytecode in a method has a severe cost and makes the execution speed of the method drop to a tenth of what it otherwise would be |
| 06:33 | <BenMillard> | blooberry looks like someone I should talk to a lot, reading back the logs. |
| 06:33 | BenMillard | is reading the MAMA articles with breakfast. |
| 06:36 | <BenMillard> | "a structural search engine is much smaller, [...] there have been no obvious efforts to address this shortcoming ... until now." surely the studies Hixie has done at Google and the DMOZ crawls Philip` has done use software which is in this category and predate the MAMA one? |
| 06:39 | <BenMillard> | what's with first-letter capitals on CSS property names and HTML attribute names, with uppercase HTML element names? that's a very rare authoring style, in my experience...indeed, the MAMA pages just use lowercase in their HTML and CSS |
| 06:41 | <BenMillard> | "it is difficult to know if such tests truly represent how those things are actually used in the real world of the live, evolving Web." indeed! |
| 06:46 | <BenMillard> | "markup elements with the lowest MAMA representation are generally phrase elements with defined semantics" cool, this correlates with my 2008 collection :) |
| 06:49 | <BenMillard> | I like how the full numbers are present for everything, like Philip` does |
| 06:54 | <BenMillard> | the sentence preceeding each of the 2 tables in this section could be omitted (along with the style attribute for inline CSS) if the <caption> of each <table> was positioned above it: http://dev.opera.com/articles/view/mama-key-findings/#structsize |
| 06:59 | <BenMillard> | it's weird how the tables have 2 entries per row...not quite enough space to transpose the table to have 1 entry per column but they could have 1 entry per row |
| 07:00 | <BenMillard> | "of the number of sites proudly displaying "W3C validation badges", only ~50% of them actually validate" I'm surprised it's that high |
| 07:01 | <BenMillard> | bold red for positive numbers confuses me a little |
| 07:03 | <BenMillard> | ok, making this table (and the one below it) have 2 entries per row is crazy: http://dev.opera.com/articles/view/mama-key-findings/#flash |
| 07:08 | <BenMillard> | "Control of font characteristics are the dominant use of CSS" nice to see numbers on that, and encouraging to see width and height in the top 10 |
| 07:11 | <BenMillard> | it's interesting that the validation error is from omitting a </li> tag, which would be completely fine if they were using an HTML doctype: http://dev.opera.com/articles/view/mama-markup-validation-report/ |
| 07:11 | <BenMillard> | (since the documents are text/html they should be using an HTML doctype anyway) |
| 07:12 | <BenMillard> | "Many authors "speak" pidgin markup" indeed, in my 2008 collection there seem to be a strata or authoring ability for each general area of HTML |
| 07:13 | <BenMillard> | s/strata or/strata of/ |
| 07:16 | <BenMillard> | "just over 20% of the [W3C Member] companies' URL top pages pass validation" ouch |
| 07:17 | <BenMillard> | oh, using Generator to figure out the authoring tool is clever |
| 07:18 | <BenMillard> | I hadn't heard of iWeb before |
| 07:19 | <BenMillard> | "Strict Doctype flavors pass validation at much higher rates (17.5%) than Transitional (8.4%) or Frameset (7.2%)." heh, even though Strict is more difficult to pass :) |
| 07:20 | <BenMillard> | "The most frequent fatal validation error: characters are used that aren't allowed by the detected character set (8.6%)." wow, didn't expect that these days |
| 07:20 | <BenMillard> | "Required attribute X not specified" is probably <img alt>, I'd guess |
| 07:21 | <BenMillard> | "We all make mistakes, even the experts." this very much correlates with my collection |
| 07:25 | <BenMillard> | the pagination links at the bottom are better than those at the top: http://dev.opera.com/articles/view/mama-w3c-validator-research-2/ |
| 07:27 | <BenMillard> | "domains can and do die on a distressingly regular basis" indeed, the history of GTA modding is slowly being wiped out because modders' sites are dying out |
| 07:29 | <BenMillard> | the sentence opening "Sticking rigorously to a standard does not necessarily spell success" is used twice in close succession here: http://dev.opera.com/articles/view/mama-w3c-validator-research-2/#whatuse |
| 07:31 | <BenMillard> | "people are BAD at this 'HTML thing'" haha, I'll testify to that! |
| 07:34 | <BenMillard> | this heading has no id so people can't link directly to it: "Validation rates where select Web-page authoring features are also involved" |
| 07:34 | <BenMillard> | surprisingly little increase in validation among authors who use CSS |
| 07:35 | <BenMillard> | the difference in validation between IIS an Apache makes sense to me |
| 07:35 | <BenMillard> | having audited several sites using each one professionally |
| 07:40 | <BenMillard> | "Put simply, a pessimist might say that a company on [the W3C Member Company] list is just about as likely to drop out of the W3C as it is to achieve a successful validation." |
| 07:43 | <BenMillard> | "even though MAMA might be comparing apples to oranges, at least it compares a fairly equal number of apples and oranges" made me giggle, from http://dev.opera.com/articles/view/mama-w3c-validator-research-2/?page=2 |
| 07:44 | <hsivonen> | iWeb has a high validation rate but also a low rate of non-div/span elements |
| 07:45 | <MikeSmith> | hsivonen: does that imply its abusing div and span? |
| 07:49 | <BenMillard> | the results about validation on "badged" websites are repeated in several places of the MAMA series, each with slightly different text |
| 07:52 | <hsivonen> | MikeSmith: from what I recall about iWeb output, yes |
| 07:53 | <BenMillard> | why is the example of a "typical Doctype statement" an XHTML doctype when HTML was measured as being twice as common? Re: http://dev.opera.com/articles/view/mama-w3c-validator-research-2/?page=2#doctype & http://dev.opera.com/articles/view/mama-key-findings/#structsize |
| 07:55 | <hsivonen> | ooh. I have iWeb on my hard drive. Looks like my Mac came bundled with it. |
| 08:01 | <BenMillard> | "MAMA looked anywhere in the document for a Doctype, but the validator only looks near the beginning of the document. A rather large set of URLs unfortunately fit this description." Philip` found something similar, IIRC |
| 08:01 | <BenMillard> | (quoted from http://dev.opera.com/articles/view/mama-w3c-validator-research-2/?page=2#doctype) |
| 08:04 | <MikeSmith> | BenMillard: lot of good questions -- here's hoping blooberry will have some time to spend scrolling back through his IRC client buffer (or krijnh's logs) |
| 08:04 | <MikeSmith> | but you might want to e-mail him some selected questions |
| 08:06 | MikeSmith | unfortunately finds no iWeb on his Mac |
| 08:10 | <BenMillard> | MikeSmith, thanks and I hope so too! I'm continuing the commentary on MAMA which you started here: http://krijnhoetmer.nl/irc-logs/whatwg/20081015#l-56 |
| 08:10 | <BenMillard> | seems more valuable for it to be public |
| 08:10 | <MikeSmith> | BenMillard: yeah, definitely |
| 08:12 | <annevk3> | canPlayType, ugh |
| 08:12 | <annevk3> | navigator.mimeTypes and friends are all epic fail, why add more? |
| 08:13 | <BenMillard> | blooberry, the use of <code> and <strong> is inconsistent through these sections: http://dev.opera.com/articles/view/mama-w3c-validator-research-2/?page=2#charset |
| 08:13 | <BenMillard> | bloo_sleep, when you awake ^^^ |
| 08:15 | <BenMillard> | man, this charset stuff is exhausting |
| 08:17 | <BenMillard> | the "[Note: ...]" under "Number of failures" has remained inside the previous <p>: http://dev.opera.com/articles/view/mama-w3c-validator-research-2/?page=3 |
| 08:21 | <annevk3> | Hixie, you have "never\n each" at some point |
| 08:21 | <annevk3> | Hixie, that should be "never\n reach" |
| 08:22 | <BenMillard> | "What IS of concern is how a page that is less than 250KB in size generates over 26MB from the validator's SOAP output mode." I've seen hsivonen and Hixie talk about combining multiple instances of certain errors and warnings into one entry for with the number of instances. |
| 08:26 | <BenMillard> | "How can the Web-at-large strive to do better when these key companies do not seem to be trying harder?" -- http://dev.opera.com/articles/view/mama-w3c-validator-research-2/?page=3#conclusion |
| 08:29 | MikeSmith | wonders which meetings hendry might be attending at the TPAC |
| 08:32 | <BenMillard> | putting the TOC at the bottom makes it hard to navigate the MAMA series: http://dev.opera.com/articles/view/mama/ |
| 08:35 | <BenMillard> | this is worth bookmarking on its own merit: http://dev.opera.com/articles/view/mama-what-has-come-before/ |
| 08:36 | <BenMillard> | I think the studies Philip` has done should feature on there as well |
| 08:49 | <annevk3> | BenMillard, feel free to comment on the forum |
| 08:51 | <BenMillard> | annevk3, I haven't anything else to say now :P |
| 08:51 | <annevk3> | you could provide pointers ;) |
| 08:52 | <BenMillard> | MikeSmith, I watched the Standards Suck episode with you in it yesterday and thought you described the various processes very well. I have a better understanding now, so thanks. :) |
| 08:53 | <MikeSmith> | BenMillard: I'm glad it was useful to somebody. My main purpose in making that was to get video footage of Lachy laughing, so in personal terms for me it was a success :) |
| 08:53 | <BenMillard> | indeed! :D |
| 09:12 | <zcorpan> | - attribute must be set to <a href=#dom-media-loading title=dom-media-LOADING>LOADING</a>.</li> |
| 09:12 | <zcorpan> | + attribute must be set to <a href=#dom-media-network_idle title=dom-media-NETWORK_IDLE>NETWORK_IDLE</a>.</p> |
| 09:12 | <zcorpan> | Hixie: was that intentional? ^ |
| 09:15 | <Hixie> | i believe so -- isn't LOADING then set later? (around the progress event_?) |
| 09:16 | Hixie | peers skeptically at his dreamhost invoice summary |
| 09:16 | <zcorpan> | i haven't finished reading the diff :) |
| 09:16 | <zcorpan> | it just looked like an unintentional search/replace to me |
| 09:17 | <Hixie> | Previous Balance: $0.00; New Charges: $59.70; New Payments: $144.20; Balance Due: $10.84 |
| 09:17 | Hixie | wonders if dreamhost are using bistromathics now |
| 09:17 | <Hixie> | i didn't do any search/replace :-) |
| 09:17 | <Hixie> | made all the changes manually :-/ |
| 09:17 | <Hixie> | in an attempt to reduce the mistakes |
| 09:18 | <zcorpan> | ok |
| 09:28 | <hsivonen> | BenMillard: part of the size problem may be SOAP :-) |
| 09:29 | <hendry> | MikeSmith: mobile testing & observing web apps hopefully. later in the week HTML WG. |
| 09:29 | <hendry> | hsivonen: are you attending the TPAC? |
| 09:29 | <hsivonen> | hendry: yes |
| 09:29 | <hendry> | oh cool |
| 09:29 | <hendry> | Hixie: look forward to seeing there too :) |
| 09:31 | <gsnedders> | ooo… new XKCD comic! |
| 09:35 | <Philip`> | gsnedders: You say that as if it's unexpected for a new one to be released on the constant thrice-weekly schedule :-p |
| 09:35 | <gsnedders> | Philip`: Oh, sure. :P |
| 09:35 | <gsnedders> | Philip`: But I'm on holiday, and thus have no perception of time beyond the releasing of XKCD comics :) |
| 09:36 | <Philip`> | gsnedders: There are more precise web sites for keeping of track of time, like http://www.whattimeisit.com/ |
| 09:37 | <gsnedders> | Philip`: But I don't look at that normally |
| 09:38 | <BenMillard> | hsivonen, from what it said an HTML results page would've been even bigger... |
| 09:39 | <hsivonen> | BenMillard: &out=gnu FWT! :-) |
| 09:40 | <hsivonen> | FTW even |
| 09:40 | BenMillard | looks up what that would produce... |
| 09:40 | <hsivonen> | (an validator.nu, that is) |
| 09:41 | <Philip`> | Does validator.w3.org's SOAP output guarantee well-formedness yet? |
| 09:42 | <hsivonen> | s/an/on/ |
| 09:42 | <hsivonen> | typo++ |
| 09:42 | <BenMillard> | hsivonen, yeah I just tried this and it's cool: http://validator.nu/?doc=http%3A%2F%2Fwww.accessifyforum.com%2Fforum2%2F&out=gnu |
| 10:31 | <annevk3> | Hixie, what you advocate for <video> with respect to accessibility seems at odds with what YouTube is doing, but maybe I'm missing something |
| 10:31 | <Hixie> | what youtube is doing doesn't allow the video to be moved around |
| 10:31 | <Hixie> | without losing subtitles |
| 10:31 | <Hixie> | (and can be done with <video> without any UA-level support) |
| 10:32 | <annevk3> | and easily without a Flash backend? |
| 10:33 | <annevk3> | given e.g. Wikipedia, it seems much easier to have a video file and a separate subtitle file from an editing perspective than have them both in the same file requiring some backend to repackage the whole thing |
| 10:33 | <Hixie> | what's flash got to do with it? |
| 10:33 | <Hixie> | if you want to do subtitles manually with <video>, it's easy |
| 10:33 | <annevk3> | Hixie, dunno, depends on how subtitles in YouTube are implemented |
| 10:34 | <hsivonen> | Hixie: without UA support, do you mean script triggering on time-based callback? |
| 10:34 | <Hixie> | yeah |
| 10:34 | <Hixie> | with overlapped spans or whatever |
| 10:34 | <Hixie> | not native subtitles |
| 10:34 | <annevk3> | ah ok |
| 10:34 | <Hixie> | what you want is real subtitles |
| 10:34 | <Hixie> | which survive when you save the video file externally |
| 10:34 | hsivonen | has no idea how video playback integrates with the event loop |
| 10:35 | <Hixie> | hsivonen: nothing really special there |
| 10:35 | <hsivonen> | do browsers render video and audio on the UI thread? |
| 10:36 | <Hixie> | i hope not |
| 10:36 | <annevk3> | ok, that solution does work, but requires script |
| 10:36 | <hsivonen> | if not, it would seem "special" to me :-) |
| 10:36 | <Hixie> | hsivonen: ? |
| 10:36 | <Hixie> | hsivonen: i don't see what the actual playback has to do with the event loop really |
| 10:36 | <annevk3> | if that's expected to be the common case of subtitles on the Web, and that seems plausible, we might want to do better |
| 10:36 | <hsivonen> | Hixie: otherwise, painting is on the UI thread, isn't it? |
| 10:36 | <Hixie> | annevk3: the common case should be subtitles in the video file |
| 10:37 | <annevk3> | Hixie, that requires advanced editing software |
| 10:37 | <Hixie> | hsivonen: sure |
| 10:37 | <Hixie> | annevk3: so does making video |
| 10:37 | <annevk3> | Hixie, shooting a video is trivial with a phone, camera, etc. |
| 10:37 | <annevk3> | see YouTube |
| 10:38 | <Hixie> | *shrug* |
| 10:38 | <Hixie> | no reason youtube couldn't embed the subtitles in the video files as far as i can see |
| 10:42 | <hsivonen> | what does YouTube do for iPhone and AppleTV when it comes to captions/subtitles? |
| 10:42 | <Hixie> | exactly |
| 10:42 | <Hixie> | (no idea what they do, but that kind of reuse is exactly why you need to have a single video file) |
| 10:42 | <annevk3> | I don't see why that is needed for such reuse |
| 10:42 | <annevk3> | also, again, if YouTube were to do that, that would require complex software on the server, which seems out of reach if people want to upload videos to their own site, etc. |
| 10:42 | hsivonen | wonders if the 3GPP-oriented video player shipping on S60r3.1 supports 3GPP Timed Text |
| 10:43 | <Hixie> | merging subtitles into an MPEG stream on the fly is trivial as i understand it |
| 10:53 | <blooberry> | philip`: yt? |
| 10:55 | <Hixie> | ok bed time for the day. |
| 10:55 | <Hixie> | nn |
| 11:00 | blooberry | thinks BenMillard had a lot of great comments on MAMA and wishes he was still on |
| 11:01 | blooberry | is doing the same in reverse: "BenMillard: blooberry looks like someone I should talk to a lot, reading back the logs." :) |
| 11:04 | <wilhelm> | Any interesting responses to the MAMA publication yet? (c: |
| 11:05 | <annevk3> | see logs for comments from Ben |
| 11:08 | <blooberry> | wilhelm: yeah, Ben's comments were great...lots of excellent commentary and I agree with many parts and have many responses. I don't know whether to reply with him in absentia due to the logs ability or wait.8-} |
| 11:12 | <wilhelm> | Yeah, read through the backlog now. |
| 11:15 | <mpt> | Wow, is that the same blooberry that taught me HTML 4? |
| 11:18 | <blooberry> | mpt: maybe? www.blooberry.com/indexdot ? |
| 11:18 | <mpt> | indeed :-) |
| 11:18 | <mpt> | Many thanks sir |
| 11:18 | <annevk3> | gotta love http://lastweekinhtml5.blogspot.com/2008/10/tpac-what-wg-teamster.html :D |
| 11:19 | <hsivonen> | I wonder if there's a page that matches the criteria at http://dev.opera.com/articles/view/mama-the-average-web-page/ (except for mutually contradictory criteria) |
| 11:19 | <blooberry> | many welcomes 8-} *wishes I had time lately to update that site for all the latest browsers) |
| 11:20 | <blooberry> | hsivonen: I don't know. That was mostly an intellectual exercise that doesn't (of course) hold a lot of water in aggregate |
| 11:20 | <mpt> | blooberry, yes, I would have thought Google would have hired you to work on Doctype |
| 11:22 | <blooberry> | They grabbed hixie from Opera to work on html5. Good hire. ;-} |
| 11:23 | <Dashiva> | I heard it was more like Hixie threw himself at them to get into their pants (i.e. search index) ;) |
| 11:24 | <blooberry> | dashiva: Well, one can't argue with "size" envy. 8-} MAMA was much, MUCH smaller at the time and couldn't answer many of his then-pressing questions. |
| 11:24 | blooberry | daydreams of billion-page indices |
| 11:24 | <mpt> | ooo |
| 11:25 | <mpt> | So there are now at least two people here who can answer a question I've had for the past seven-ish years: |
| 11:25 | <mpt> | What is the average number of resources that need to be loaded, in total, to display any HTML page? |
| 11:25 | <blooberry> | you mean "just HOW rabid is hixie about stargate?" |
| 11:26 | <mpt> | average number of unique resources, that is |
| 11:27 | <blooberry> | We can look at some of those things in isolation first, before combining them |
| 11:29 | <blooberry> | like inline images were used in over 90% of pages. average total image usage was ~20, with unique references averaging about 12. |
| 11:30 | <blooberry> | I didn't look at image file size this time because I was worried about the performance hit, but I'll probably include that in the future |
| 11:31 | <blooberry> | but now we'd also have to look at plug-ins, frames/iframes, external scripts, external css as well. |
| 11:32 | <blooberry> | In the writeup I've tried to address each of these areas in isolation...hopefully an aggregate picture arises from that |
| 11:33 | <mpt> | The reason I ask is that it's a clue to giving browsers a page-loading progress bar that is, on average, smooth |
| 11:37 | <blooberry> | you mean perhaps giving an indication of the typical state of affairs for the "average" user encountering a web page and their browsing experience? |
| 11:39 | annevk3 | would expect that number to increase over time |
| 11:39 | <mpt> | I mean the browser giving a better guess of how much of the overall page has loaded, when it doesn't yet know the sizes of all the resources involved, and may not yet even bet at the point where it knows how many there will be. |
| 11:39 | <blooberry> | annevk3: me too. 8-/ |
| 11:40 | <mpt> | As opposed to the Internet Explorer progress bar behavior of starting out as if the page will have no images etc at all and therefore slowing to a crawl, or the Safari behavior of (as best I can tell) treating the last resource as if it's going to be much bigger than all the others. |
| 11:41 | <mpt> | bet -> be |
| 11:41 | <annevk3> | Opera isn't great at all with that either |
| 11:48 | <mpt> | afaict Opera doesn't even try to show a progress bar for non-HTML stuff |
| 11:51 | <annevk3> | it does |
| 12:00 | blooberry | is sometimes annoyed by Opera's behavior with some pages where the overall download object count in the progress bar just keeps slowly increasing. It feels like Monty Python and the Spanish Inquisition. |
| 12:00 | <BenMillard> | blooberry, hi glad they went down well. :) |
| 12:00 | <blooberry> | benmillard: hi! |
| 12:02 | <blooberry> | benmillard: I have lots of comments to your comments. 8-} |
| 12:03 | <BenMillard> | blooberry, fire away :) |
| 12:04 | <blooberry> | I've seen some of Philip`s results here, but I hadn't (especially at the time of the write up) seen anything "official"...not that that matters much. I'd be happy to list his work but I think when I asked him before he didn't have something over-arching or permanent to give. |
| 12:05 | <BenMillard> | blooberry, here's one I remember: http://canvex.lazyilluminati.com/survey/2007-07-17/analyse.cgi/index |
| 12:06 | <hsivonen> | http://www.google.fi/search?q=http://canvex.lazyilluminati.com/misc/stats/&filter=0 |
| 12:07 | <BenMillard> | (with English UI: http://www.google.com/search?q=http://canvex.lazyilluminati.com/misc/stats/&filter=0) |
| 12:07 | <blooberry> | Cool. I will ask him if he wants to be included and will definitely do so. |
| 12:11 | <blooberry> | As for pre-dating MAMA, Ian's published work is similar but was more stats-centric and not so focused on individual cases. Plus when he was with Opera I had worked with him and others in brainstorming ideas for new features. He was just able to leverage Google's set much more quickly than I was (it was a part time project for me back then) |
| 12:13 | <BenMillard> | I've seen Ian give specific stats after doing specific crawls, but these were within the past year or two |
| 12:13 | <BenMillard> | my understanding of "now" is "at the time these words were published" :) |
| 12:14 | <BenMillard> | like, searching for how often headers+id is used and how much of that isn't bogus |
| 12:14 | <blooberry> | Regarding the ELEMENT Attribute style, I wanted each to stand out from each other, especially as I'd be mentioning elements and attributes ALL over the place. 8-} and I didn't know the eventual publishing environment. This was a good move now that I know the dev.opera environment a little better (it is terrible, btw 8-/) |
| 12:15 | <BenMillard> | blooberry, FWIW I tend to write <element> and either "the foo attribute" or <element attribute> |
| 12:15 | <BenMillard> | other people working on HTML5 use attribute="" |
| 12:16 | <BenMillard> | <code>title</code> is different enough from <code><code></code> for me, but that might just be me :P |
| 12:16 | <blooberry> | I know he's done more recent incremental studies with (I think) smaller URL sets. Evidently arranging for using/analyzing a bulk of Google's set takes some finagling |
| 12:18 | <blooberry> | It was really just a convention I started using early on to distinguish elements and attributes though. No specific agenda. 8-D |
| 12:20 | <blooberry> | <BenMillard> it's weird how the tables have 2 entries per row...not quite enough space to transpose the table to have 1 entry per column but they could have 1 entry per row |
| 12:22 | <blooberry> | ok, THAT one pissed me off yesterday...I originally wrote such tables with a cheesy gutter in the middle to create a 2-column group. Mostly to present enough data in as short a vertical space as possible for larger tables. So when the dev.opera gatekeep tells me yesterday that dev.opera somehow STRIPS OUT rowspans...*shakes head* I don't get it. I guess it isn't allowed. Apparently I'm stuck. But I will pursue a solution because as i |
| 12:22 | <blooberry> | hard to read. In some cases in later documents I have *3* groups side by side. |
| 12:23 | <BenMillard> | I see... |
| 12:24 | <BenMillard> | I think squishing more data into less vertical space is unhelpful for data tables, especially those which are ranked in a specific order |
| 12:25 | <BenMillard> | one entry per row would keep the order straight up-and-down |
| 12:25 | <blooberry> | bold/red for positive numbers: again, I didn't know the environment the data would live in. I had to choose between keeping a class that actually specified some metadata on the contents...that they were numbers. Or, I could choose appearance. Another dev.opera limitation is that you can't specify your own external stylesheets. So I had to leverage on existing stock classes. Much to my chagrine. ;-} |
| 12:27 | <BenMillard> | blooberry, I guess this feedback should be passed on to those gatekeepers then, as the same editorial choices are likely to be present in other documents they maintain? |
| 12:27 | <BenMillard> | Bruce Lawson's message here makes me think changes for the better can happen: http://www.accessifyforum.com/viewtopic.php?t=10824#62695 |
| 12:28 | <BenMillard> | (I'm registered as Cerbera on that forum and was giving feedback on the Web Curriculum in that thread.) |
| 12:28 | <blooberry> | oh, and the validation errors on the articles were fixed. That was just some 3AM sloppiness on my part. 8-} On the main validation document, since day 1 I had a prominent H1 comment emblazoned at the very type "MAKE SURE THIS DOCUMENT VALIDATES!!!" heheh. |
| 12:29 | <BenMillard> | hehe, happens to everyone -- as your research found! :) |
| 12:31 | <blooberry> | Totally. The incidental validating evidence I list about reasons for failing to validate actually happened to me yesterday. |
| 12:32 | <blooberry> | "Strict Doctype flavors pass validation at much higher rates (17.5%) than Transitional (8.4%) or Frameset (7.2%)." |
| 12:33 | <blooberry> | I really like that the people who "do" Strict or "move to" strict really seem to put forth more effort in making things validate. It is like, they've gone THIS far, so they might as well take that long, last big step. |
| 12:34 | <blooberry> | ""Required attribute X not specified" is probably <img alt>, I'd guess": I actually wish I could have done more here. |
| 12:35 | <BenMillard> | blooberry, there seem to be phases of authoring ability in which the step from Transitional to Strict is like a rite of passage. :) |
| 12:35 | <blooberry> | That validation pass was really my first big experiment with the validator...I wasn't at all sure how big the result set would be. |
| 12:35 | <blooberry> | So I was ambitious yet conservative. |
| 12:36 | <blooberry> | If I had unlimited storage, I would have stored each error, with each error's argument type separately, aggregating the data by those variables. |
| 12:37 | <blooberry> | Instead of aggregating by error code, it would be by error code AND the error's variable arguments |
| 12:37 | <blooberry> | but that would blossom the error storage requirements. Plus the SOAP output doesn't make that easy. 8-} |
| 12:39 | <blooberry> | "the pagination links at the bottom are better than those at the top: http://dev.opera.com/articles/view/mama-w3c-validator-research-2/ ": I agree. This was originally 1 long article, but due to more dev.opera constraints it had to be broken up. I actually requested the text links between parts at the top to help top-down random access. So...mostly my fault. ;-} |
| 12:41 | <annevk3> | oops |
| 12:41 | <annevk3> | netsplit? |
| 12:42 | <BenMillard> | blooberry, if the top links were in the style of the bottom links, I think it would be fine |
| 12:43 | <blooberry> | "Sticking rigorously to a standard..." on http://dev.opera.com/articles/view/mama-w3c-validator-research-2/#whatuse : Hmm. You are right. I'll get that taken care of. |
| 12:43 | <BenMillard> | btw, I think it's very cool how receptive you are to feedback :) |
| 12:43 | <hsivonen> | hmm. the word "astronauts" has disappeared from the title and description of session 2 on TPAC day |
| 12:43 | <BenMillard> | I want you to know that I appreciate that it's a huge piece of work |
| 12:43 | <annevk3> | hsivonen, pointer? |
| 12:43 | <hsivonen> | annevk3: http://www.w3.org/2008/10/TPAC/TPDay-Agenda.html |
| 12:44 | <blooberry> | well, the top links were added statically, the bottom navigation thing is produced automatically, I'm told. |
| 12:45 | <BenMillard> | blooberry, producing both automatically would be best, imho, as it would improve all paginated articles |
| 12:47 | <blooberry> | thanks. 8-} I'm really glad it is finally seeing the light of day. |
| 12:49 | <blooberry> | Totally agreed. But Chris (Mills) conveyed that dev.opera just can't do it that way right now/yet. 8-{ |
| 12:49 | <hsivonen> | it would have interesting to see Hixie on the architecture panel |
| 12:49 | <blooberry> | I think my articles are stressing the boundaries of the dev.opera system too much. 8-D |
| 12:51 | <BenMillard> | blooberry, that seems to be the case for some Curriculum articles as well...the popularity of these publications should encourage improvements to the system over time? |
| 12:52 | <blooberry> | "results about validation on "badged" websites are repeated in several places of the MAMA series, each with slightly different text": I tried not to repeat or rehash too much. |
| 12:52 | <BenMillard> | blooberry, it felt like 3 different places covered it...I may have got confused and re-read earlier instances though |
| 12:52 | <blooberry> | re: curriculum articles. I hope so. Chris already seems as much an advocate for change as I am just now becoming (are the tenses in that sentence even close to being correct??) |
| 12:53 | <blooberry> | The "key findings" document was added recently, and the badges info has repeatedly stood out as being an "interesting" finding of all this and worthy of repeat. But I prefer not to repeat myself repeat myself too much |
| 12:54 | <BenMillard> | lol :D |
| 12:55 | blooberry | takes a small break to work on a blog post about MAMA that is overdue |
| 12:57 | <hsivonen> | hmm. annoying. the doctype name is an interned string in Gecko, but the spec requires a case-insenstive compare |
| 12:58 | <Dashiva> | split, split |
| 13:16 | <hsivonen> | argh. the damowmow portal doesn't work in IE8 |
| 13:21 | <hsivonen> | Hixie: the multipage version of the spec tells me it has been updated even if I click Reload |
| 13:31 | <Philip`> | blooberry: I'm here now |
| 13:32 | <Philip`> | blooberry: http://philip.html5.org/data.html is the closest thing to a summary of my stuff that's spread all over the place, so it'd probably be the best place to link |
| 13:33 | <Philip`> | though maybe I should make it more comprehensive, and/or organise it better so it's easier to find the less rubbish things |
| 13:33 | <Philip`> | mpt: If by average you mean mean, then the concept of averages doesn't really make sense - e.g. I could make one page with an infinite number of external resources, and then the average per page will be infinite. So you need a more complex view of the distribution of values, else it'll be silly :-) |
| 13:34 | <blooberry> | philip`: Yeah, I found I had to discard some values due to out of control conditions |
| 13:35 | <blooberry> | I occasionally ran into pages fed by databases that were caught in infinite loops. Those were fun and crashy before I instituted a max page size. 8-} |
| 13:35 | <Philip`> | I had that problem with a radio station, which was streaming MP3 over HTTP and was linked from dmoz.org |
| 13:36 | <Philip`> | and I couldn't quite work out how to make the HttpClient library stop downloading after a certain point, so I just had to ctrl-c it and remove that URL from the list :-/ |
| 13:37 | <blooberry> | philip`: Do you want me to wait or should I use that link then? |
| 13:39 | <Philip`> | blooberry: You might have to wait an unbounded amount of time, since I don't plan to actually update that page in the near future; so if you want to link to anything, it'd probably be best to just link to the current version of that page |
| 13:39 | <mpt> | Philip`, the mean of all reasonably-measurably-finite pages :-) |
| 13:39 | <blooberry> | I'm ok with waiting until you have something you like better/want to put out there...ah, ok. |
| 13:39 | <Philip`> | mpt: Then I'll just make a site with an infinite number of reasonably-measurable finite pages, to force the mean to whatever value I want :-) |
| 13:39 | <mpt> | Philip`, and weighting pages by popularity would make it even more accurate (but would be correspondingly more difficult) |
| 13:40 | <mpt> | Philip`, there are already sites like that, honeypots for spammers |
| 13:43 | <Philip`> | Averages destroy almost all of the information anyway, and even if you measured stuff like standard deviation then you'd e.g. be losing the interesting bumps in http://code.google.com/webstats/2005-12/charts/unique-elements-per-page.svg |
| 13:46 | <blooberry> | I had done a bunch of standard deviation calculations for the validation study in MAMA, but a co-worker persuaded me that it was mostly superfluous noise to do so in those cases. |
| 13:51 | <Philip`> | It's probably nice to indicate the difference between e.g. an error which lots of pages have several times, and an error which most pages don't have (or almost never have) but a few pages have thousands of times |
| 13:51 | <Philip`> | but I don't know how you can concisely indicate that kind of information to a reader |
| 13:53 | <Philip`> | (I suppose for most readers it won't make any difference anyway because they're just reading it for entertainment and you could have made up all the numbers; I'm not sure exactly who would care about the actual data or what they'd want to do with it, so I have no idea how to present it to them) |
| 13:53 | blooberry | is NOT an expert or even anything like an afficionado of graphing... (but would a 2 plane graph be interesting to present that sort of thing?) |
| 13:54 | <Philip`> | Not quite sure what you mean by "a 2 plane graph" |
| 13:54 | <Philip`> | I guess you could use a histogram, but they're not very concise |
| 13:55 | <blooberry> | me neither 8-} ...trying to find a visualization to describe it |
| 13:57 | <mpt> | Philip`, my original question was for progress bars, not error reports |
| 13:57 | <mpt> | oh neat, that's just what I was looking for, thanks :-) |
| 13:58 | <Philip`> | mpt: Uh, I don't think that's anything like what you said you were looking for |
| 13:58 | <blooberry> | hmm.not finding one based on my searches. What I meant was, you have an ordinary 2d graph of error type versus URL frequency. right angle to that plane you have 2d graph of error type versus quantity, with error type being the common intersection line between the planes. (have no idea if that is clear at all...or if it would even be a helpful visualization) |
| 13:58 | <Philip`> | (http://code.google.com/webstats/2005-12/pages.html is the context) |
| 13:58 | <mpt> | oh, *HTML* elements |
| 13:58 | <mpt> | bother |
| 13:59 | <Philip`> | mpt: You want to count the total number of distinct images, scripts, iframes, etc, and transitively count all the contents of iframes, etc? |
| 14:00 | <mpt> | Philip`, if by transitively you mean recursively, yes |
| 14:00 | <hsivonen> | hmm. the Adobe guy on the Open Web Podcast says Adobe would love to be a part of the HTML5 video story |
| 14:01 | <hsivonen> | also, some changes in Flash Player licensing |
| 14:01 | <Philip`> | blooberry: I think I see what you're saying, but it sounds kind of like it'd turn into a confusing mess of lines when it's rendered on a 2D screen :-( |
| 14:01 | <Philip`> | mpt: I do mean recursively |
| 14:02 | Philip` | is thinking too mathsishly |
| 14:02 | <mpt> | I know that value would be in the formula somewhere, but I don't know enough maths to know what the formula would be anyway |
| 14:04 | <Philip`> | Hmm, am I being confused? |
| 14:04 | <Philip`> | I thought the only relevant formula is like "a + b + c + ...", where a is the number of distinct images on the page or whatever, etc |
| 14:06 | <Philip`> | Anyway I expect it'd be pretty hard to count because you want to include all the ads, which are probably commonly document.written in crazy ways, so it's hard to statically determine how much stuff you'll have to load |
| 14:06 | <mpt> | hm, true |
| 14:07 | <Philip`> | (and the range would be large enough that the average would be pretty useless, and there isn't really any information you can know in advance to do a better prediction for a given page) |
| 14:08 | <mpt> | And since waiting for server response is a non-trivial part of the overall time for 3rd-party ads etc, maybe it's best measured by just throwing thousands of pages at a browser rather than throwing millions of pages at a script |
| 14:08 | <blooberry> | philip`: yeah, it would probably be pretty visually confusing. |
| 14:09 | Philip` | is happy if the progress bar just animates prettily while it's actively doing work, and stops animating if it's lost the network connection or everything has stalled, and he doesn't care much about knowing the actual progress percentage :-) |
| 14:09 | Philip` | goes away for an hour or so |
| 15:16 | <gsnedders> | hiho |
| 15:17 | <gsnedders> | http://lastweekinhtml5.blogspot.com/2008/10/tpac-what-wg-teamster.html — :D |
| 15:18 | <gsnedders> | I do like how he failed to find a horrible photo of me |
| 15:18 | <gsnedders> | Though I guess most horrid ones are on private bebo pages :P |
| 15:30 | <hsivonen> | on the TPAC panel, I have a position statement that takes 15 sentences. |
| 15:31 | <annevk3> | you're on a panel? |
| 15:31 | <hsivonen> | the only way I know how to make slides for those is to make put each sentence on a slide and change slides at sentence pace... |
| 15:31 | <hsivonen> | annevk3: yes |
| 15:32 | <annevk3> | what other people are on that panel? |
| 15:32 | <hsivonen> | annevk3: T.V. Raman, Charles Wiecha and Erik Dahlström |
| 15:33 | <annevk3> | topic? |
| 15:33 | <hsivonen> | Future of XML ecosystem in W3C client-side work |
| 15:33 | <annevk3> | that could be fun :) |
| 15:33 | <zcorpan> | Hixie: can't you have playcount='-1' to represent loop forever? |
| 15:34 | <hsivonen> | oh. John Boyer is on the panel, too. |
| 15:34 | <annevk3> | hsivonen, maybe just provide pointers in your slides or have no slides at all? |
| 15:34 | <annevk3> | nice, I hope nobody holds back :) |
| 15:35 | <Philip`> | zcorpan: That's not exactly intuitive |
| 15:35 | <Philip`> | zcorpan: and nobody reads the documentation |
| 15:35 | <Philip`> | zcorpan: so if they read a page containing playcount=-1 they won't know what it means, and if they want to write an infinitely looping video then they won't be able to guess what they have to type |
| 15:36 | annevk3 | is in favor of dropping loop attributes |
| 15:36 | <annevk3> | would still allow configuration of your UA to stop looping |
| 15:37 | Philip` | would have thought infinite looping was much more common than finite looping, so it's odd that the latter is supported while the former isn't |
| 15:37 | <gsnedders> | Shows your thought ability. |
| 15:37 | <Lachy> | what are the use cases for looping video at all? |
| 15:38 | <annevk3> | presentational effects I assume |
| 15:38 | <Philip`> | Lachy: Incredibly annoying ads |
| 15:38 | <Philip`> | which would want to loop infinitely, to be as annoying as possible |
| 15:39 | <Lachy> | but the kind of ads that are produced as videos generally don't loop infinitely. |
| 15:40 | <Lachy> | e.g. the Ads inserted by YouTube into their full length TV shows, or the Get a Mac ads published by Apple on their website |
| 15:40 | <Philip`> | Those ads aren't intended to be annoying, so they're atypical |
| 15:41 | <Lachy> | well, the YouTube ones are reportedly unskippable, so they're a little annoying |
| 15:41 | <hendry> | what has happened to clientinformation ? http://www.whatwg.org/specs/web-apps/2007-10-26/multipage/section-browser.html#clientinformation |
| 15:41 | <hendry> | it's dissapeared from the current spec or renamed? |
| 15:42 | <Philip`> | I'm thinking of the ones where you visit a web page, and a little box in the corner of the screen starts downloading and displaying ugly low-quality video and probably plays sound effects, to advertise some kind of irrelevant product that you don't care about |
| 15:43 | <Lachy> | so you think people will start using <video> as a replacement for Flash? |
| 15:43 | <annevk3> | hendry, renamed |
| 15:43 | annevk3 | looks |
| 15:43 | <gsnedders> | hendry: http://www.whatwg.org/specs/web-apps/current-work/#system-state-and-capabilities |
| 15:44 | <Philip`> | hendry: Merged into Navigator in http://lists.w3.org/Archives/Public/public-html-commits/2008Aug/0110.html |
| 15:44 | <annevk3> | ah, seems gsnedders was already looking :) |
| 15:44 | <gsnedders> | annevk3: :) |
| 15:47 | <zcorpan> | Philip`: authors read other authors' code |
| 15:48 | gsnedders | wonders what will manage to cope with blockquote:only-child *:first-child::before |
| 15:48 | <Philip`> | zcorpan: Those authors will see playcount=-1 and not know what it means |
| 15:49 | <gsnedders> | hah. |
| 15:49 | <zcorpan> | hsivonen: html5 used to case-fold doctype name before the big quirks revamp. iirc, mozilla used to never uppercase, but started to do so at some point (because html5 said to uppercase) |
| 15:49 | <gsnedders> | MSN for OS X copes! |
| 15:49 | <gsnedders> | No, it doesn't |
| 15:49 | <gsnedders> | It falls down on the ::before |
| 15:49 | <zcorpan> | Philip`: no they will se a looping video and go "hmm wonder how they did that" and then copy the <video> tag |
| 15:50 | <Philip`> | zcorpan: And then they'll change the -1 to -2 because that's like twice as good |
| 15:50 | <zcorpan> | Philip`: so the spec could say that any negative number means infinate |
| 15:50 | <zcorpan> | Philip`: compare with tabindex :) |
| 15:51 | <zcorpan> | i haven't seen tabindex=-2 |
| 15:51 | <hsivonen> | zcorpan: do you recall if there was a reason to stop case folding as part of the quirks revamp? |
| 15:51 | <zcorpan> | hsivonen: i guess Hixie looked at mozilla source code and did what mozilla did |
| 15:52 | <zcorpan> | hsivonen: but i don't know |
| 15:52 | <zcorpan> | i also don't know why mozilla started to uppercase it, i'm just guessing :) |
| 15:52 | <annevk3> | they did that because of html5lib tests |
| 15:53 | <annevk3> | sayrer implemented that |
| 15:53 | <zcorpan> | ok |
| 15:53 | <hsivonen> | I must have missed the bugzilla trail |
| 15:54 | <hendry> | are plugins and JS allow to set values on navigator? |
| 15:59 | <Philip`> | zcorpan: There's http://www.nhstateparks.org/state-parks/alphabetical-order/annett-wayside-park/ with <iframe tabindex="-32768" ...> :-) |
| 15:59 | <Philip`> | (Well, there was some months ago) |
| 15:59 | <zcorpan> | Philip`: nice :) |
| 15:59 | <Philip`> | (<iframe tabindex="-32768" src="/content/state-parks/alphabetical-order/weather/annett-wayside-park.htm" frameborder="0" height="300" scrolling="auto" width="200"></iframe>) |
| 16:00 | zcorpan | wonders whether that's conforming html5 or if just -1 is conforming |
| 16:01 | <zcorpan> | or should be... |
| 16:04 | <annevk3> | in any case, doing what hsivonen suggests makes sense |
| 16:04 | <annevk3> | it simplifies code, is more consistent, and he'll fix it for the one correct impl :) |
| 16:05 | <Philip`> | zcorpan: ...but that's the only case I can find with (?i)tabindex=.?-[^1] so I'd have to agree that setting it to -2 is quite rare :-) |
| 16:41 | <Philip`> | Does <source pixelratio> actually do anything? |
| 16:43 | Philip` | is wondering whether it would end up get rounded to integers, so 0.909090909 would be identical to 10/11 as long as your video was smaller than five billion pixels |
| 16:45 | <gsnedders> | http://gsnedders.com/about-the-author — anyone able to come up with anything good for that? |
| 16:46 | <Philip`> | gsnedders: The topic in this channel should probably be changed at some point :-) |
| 16:46 | <hsivonen> | minibios are hard to write |
| 16:47 | <gsnedders> | Philip`: I should also probably upload it at some point :) |
| 16:50 | Dashiva | wonders why Hixie changed OS to "other" to mark a bug as delayed |
| 17:04 | <annevk3> | http://twitter.com/motleyceo/statuses/960778535 |
| 17:10 | <Philip`> | For security reasons, I think the 'required' attribute should be renamed to 'required-(but-validate-this-on-the-server-side-too)' |
| 17:12 | <Philip`> | At least for the next several years it'll be obvious that you can't rely on WF2's client-side validation, since most client-sides won't implement it and so anyone writing tutorials will include a note to say that; the problems are likely to come much further in the future, when everyone supports client-side validation |
| 17:13 | <gsnedders> | Apart from 1337 hack0r browser |
| 17:15 | <Philip`> | zcorpan: Oops, I spoke too soon about tabindex |
| 17:15 | <Philip`> | http://grammar.ccc.commnet.edu/grammar/ had <a href="http://fastcounter.bcentral.com/fc-join"; target="_top" TABINDEX="-2"> |
| 17:15 | <Philip`> | and http://web.jaes.tpc.edu.tw/ had lots of <a tabIndex="-32768" href="http://epaper.edu.tw"; target="_blank"> |
| 17:20 | <Philip`> | gsnedders: You should fix about-the-author to not include </br> |
| 17:21 | <gsnedders> | Philip`: Habari's autop has far bigger bugs. meh. |
| 17:21 | <gsnedders> | Philip`: And don't distract me from my work! |
| 17:22 | <Philip`> | Autop? |
| 17:23 | <gsnedders> | automagically adds p and br tags, but does it too much |
| 17:24 | <Philip`> | Oh |
| 17:24 | <Philip`> | It messes up the layout of the page quite badly |
| 17:25 | <gsnedders> | not in Safari :P |
| 17:25 | <gsnedders> | nor in Fx 3 |
| 17:26 | <Philip`> | It does in Opera - that blockquote gets indented a lot |
| 17:27 | <gsnedders> | Hmm, that's right per HTML 5, I think |
| 17:27 | <gsnedders> | (blockquote:only-child stops it from being) |
| 17:28 | <Philip`> | Why is it different in the other browsers? |
| 17:29 | <gsnedders> | Dunno |
| 17:29 | <Philip`> | (But you should use class instead of fancy selectors :-p ) |
| 17:29 | <gsnedders> | Saf has no <br> in the DOM |
| 17:29 | <gsnedders> | so it only has one child |
| 17:29 | <gsnedders> | it just ignores the /br |
| 19:38 | hsivonen | extends his position to 18 sentences |
| 19:45 | <Philip`> | Sounds like quite a complex position |
| 19:54 | <hsivonen> | Philip`: the short version is "Let's make the text/html parser insert SVG DOM nodes." |
| 19:54 | <hsivonen> | Philip`: the rest is explaining why |
| 20:50 | <Hixie> | hsivonen: why do you think it would be interesting to see me on the arch panel? |
| 20:52 | <hsivonen> | Hixie: you clearly have opinions about architecture, and those opinions presumably differ from the opinions of TAG members |
| 20:52 | <hsivonen> | Hixie: so it would be interesting to hear how they respond to your opinions and you to theirs |
| 20:52 | <Hixie> | we know how they'll respond |
| 20:52 | <Hixie> | "sending passwords in the clear is always wrong." |
| 20:57 | <Philip`> | They don't seem to be able to agree that it's always wrong |
| 20:58 | <Hixie> | which is ridiculous, because it _is_ always wrong, it's just that correctness isn't the only constraint |
| 20:58 | <Hixie> | sometimes one has to compromise on ideals |
| 20:59 | <Hixie> | eric_carlson: do you or your colleagues have a position on whether we should keep the looping attributes on <video>? |
| 20:59 | <eric_carlson> | Hixie: I was just going to reply to the thread |
| 20:59 | <Hixie> | cool, great |
| 21:00 | <Philip`> | But if you're writing a normative document with MUSTs and SHOULDs and PROBABLY OUGHT TOs and so on then you shouldn't write down MUST requirements that might have to be compromised on, which seems like it might be their problem |
| 21:01 | <eric_carlson> | Hixie: I do think that looping can be useful, and it can be very difficult to make it seamless in script |
| 21:02 | <eric_carlson> | Hixie: "loopstart" and "loopend" aren't essential |
| 21:02 | <KrocCamen> | Javascript shouldn't be assumed, it's for joining actions together. One example of a loop might be for a complex animated background of sorts. Ancillary uses of "video". |
| 21:02 | <Hixie> | eric_carlson: there are lots of things that fall into that category, the question is are they common enough to warrant supporting in version 1, really |
| 21:03 | <KrocCamen> | Also, if you think of a DVD menu. It has looping animated video buttons / background. (S5+Video õ_o) |
| 21:04 | <Hixie> | dvd menus are pretty much the pinacle of what i want to avoid us ever doing :-P |
| 21:05 | <KrocCamen> | I'm just thinking broadly, I could cringe too :) |
| 21:05 | <Hixie> | i agree that dvd menus have looping |
| 21:06 | <Hixie> | does anyone on the web want to do looping of video in that way? |
| 21:06 | <eric_carlson> | Hixie: I think a valid use-case for looping is background audio. |
| 21:07 | <Hixie> | hmm, very good point |
| 21:07 | <KrocCamen> | With the HTML5 changes, I'm thinking of the content / design issues of my site, where I have no classes and no javascript. HTML5 not being Javascript or CSS, should ensure that my site is fully functional without replicating document-structure in Javascript/CSS. (My main complaint-example is the removal of the "type" attribute on OL/UL) |
| 21:07 | <Hixie> | so the current features we have totally fail at looping background audio |
| 21:07 | <Hixie> | seems like if that's our use case, we should replace the whole set of features with a single boolean attribute, "loop when you get to the end" |
| 21:07 | <eric_carlson> | Hixie: agreed |
| 21:08 | <Hixie> | ok that would be nice and easy to do |
| 21:08 | <Hixie> | and would get out of our way on the issue of picking the start point |
| 21:08 | <KrocCamen> | What about ads? They loop. Would beat using Flash. |
| 21:09 | <eric_carlson> | While we are simplifying things, I don't think that "loopstart" and "loopend" are common enough to keep. |
| 21:10 | <Hixie> | i would be dropping all five attributes -- start, end, loopstart, loopend, and playcount -- and replacing them with a single attribute, loop, that is either on or off |
| 21:10 | <Hixie> | at least, that's what i'd like to do |
| 21:10 | <Hixie> | leaving just two DOM attributes, one to reflect loop, and one to report the current loop number |
| 21:10 | <Hixie> | which starts at zero and increments on end |
| 21:12 | <KrocCamen> | What if they were one and the same. 0 = false, >1 = true, count. |
| 21:12 | <KrocCamen> | correction >=1 |
| 21:13 | <eric_carlson> | Hixie: "start" and "end" *can* be very useful |
| 21:14 | <Hixie> | eric_carlson: oh i agree, but i think we should use a better story for "start", and it's unclear that "end" really has enough of a solid story and enough use cases for us to really do it in version 1 |
| 21:15 | <eric_carlson> | Hixie: wasn't it you that was just arguing that addCueRange() needs the pauseOnExit parameter because it is too difficult to stop precisely from script :) |
| 21:17 | <Hixie> | eric_carlson: sure, but you can use addCueRange() to implement end="" for that case. :-) |
| 21:21 | <eric_carlson> | Hixie: true, but you won't be able to loop a portion of file. |
| 21:21 | <eric_carlson> | Hixie: or limit playback to just a portion, eg one chapter. |
| 21:22 | <KrocCamen> | What about a game |
| 21:22 | <KrocCamen> | ^ What about the SFX for a game. To reduce HTTP requests, put the soundbank in one audio file |
| 21:23 | <KrocCamen> | (I'm writing a Javascript/Canvas game, and sound is an unknown atm) |
| 21:23 | <Hixie> | eric_carlson: is that needed much? |
| 21:23 | <eric_carlson> | Hixie: good question |
| 21:24 | <Hixie> | KrocCamen: there has been talk of doing that using archive files |
| 21:24 | <eric_carlson> | Hixie: and it is possible if you also don't use the native controls |
| 21:24 | <Hixie> | KrocCamen: so you download a zip file of audio files and play them |
| 21:24 | <Hixie> | eric_carlson: yeah |
| 21:25 | <KrocCamen> | Hixie: Excellent, so good. It'd be nice to support JARs too, so you can include 10 js files all in one or something. Or a whole file structure of js/audio/images in a jar file, all inserted into a virtual file system as if the files were in the website's normal file structure. That sort of thing would be good for web apps / games. |
| 21:25 | <eric_carlson> | Hixie: OK, I agree that it is not a common usage, and it is possible to do with a custom controller. |
| 21:26 | <eric_carlson> | Hixie: still not possible to do seamless looping of a portion of a file, but I am not sure that is a common need. |
| 21:27 | <annevk3> | Hixie, also, it provides a way to get feedback from W3C Members in a way they like giving feedback |
| 21:28 | <Hixie> | KrocCamen: yeah, jar=zip |
| 21:28 | <annevk3> | Hixie, since you're around anyway losing an hour of HTML editing time doesn't seem much of a loss and there's potential gain |
| 21:28 | <Hixie> | eric_carlson: well there's always going to be things we can't do, the key is to make sure we can do the important ones, and do the others later |
| 21:29 | <eric_carlson> | Hixie: Agreed |
| 21:29 | <Hixie> | annevk3: i see no potential gain (there was no gain from the panels i was in last year, or the year before, or the year before that) and an hour is a lot of time. |
| 21:30 | <annevk3> | but it probably has benefited other people |
| 21:30 | <Hixie> | i'm unclear on that |
| 21:31 | <annevk3> | more mutual understanding helps communication going forward |
| 21:31 | <eric_carlson> | Hixie: while we are on the subject of "is that needed much", I am not sure that "pixelRatio" passes the smell test |
| 21:31 | <annevk3> | if you're wasting two weeks for that one week (as you mentioned) you might as well lose another hour imo |
| 21:31 | <annevk3> | I for one thought the panels were worth listening to |
| 21:32 | <blooberry> | eric_carlson: heheh. "passes the smell test" is one of the kindest euphemisms I've heard in quite some time. 8-} |
| 21:33 | <Hixie> | eric_carlson: pixelratio came out of feedback from the youtube guys, iirc, apparently there is a lot of misencoded video out there and having a way to fix it would really help such sites |
| 21:33 | <Hixie> | eric_carlson: note that it is intentionally done in an obtuse way so as to discourage authors from setting it (e.g. that's why height/width don't override the aspect ratio) |
| 21:37 | <eric_carlson> | Hixie: having width/height only set the canvas size is absolutely the right thing to do. |
| 21:37 | <eric_carlson> | Hixie: but aspect ratio is much harder to get right. |
| 21:37 | <Hixie> | you'd be surprised how many people beg for width/height to control the aspect ratio |
| 21:37 | <eric_carlson> | Hixie: for example, a DV NTSC movie has a Rec.601 pixel aspect ratios (10:11) |
| 21:38 | <eric_carlson> | Hixie: but you can't take the 720x480 and multiply by 10:11 |
| 21:38 | <eric_carlson> | Hixie: you need to crop to clean aperture (CLASP) to 704x480, then multiply by 10:11 and you get the proper 640x480 |
| 21:39 | <Hixie> | don't think we really want to expose the clean aperture as well |
| 21:39 | <eric_carlson> | Hixie: exactly! |
| 21:39 | <Hixie> | i mean the main use case here is crappily-encoded videos anyway |
| 21:40 | <Philip`> | YouTube has proved that people are perfectly able to put up with 320x240, so we might as well just standardise on that so we don't have to worry about sizes and aspect ratios and all that junk |
| 21:41 | <Hixie> | YouTube has shown that people will upload all kinds of anamorphic content without correctly setting the metadata in the files they upload |
| 21:41 | <Hixie> | hence youtube's desire to be able to correct it on the fly later |
| 21:42 | <annevk3> | arguably we could have two ints separated by an "x" |
| 21:43 | <annevk3> | gives the precision that's not really needed, is already required for <link sizes> and is more compatible with how aspect ratios are usually provided |
| 21:43 | <Philip`> | Why does YouTube want to fix anything on the client rather than on the server? |
| 21:44 | <Philip`> | annevk3: I don't think I've ever seen aspect ratios with an 'x' - it's always written like "16:9", as far as I'm aware |
| 21:44 | <annevk3> | Philip`, true, if the : is really needed it would be slightly more verbose |
| 21:45 | <KrocCamen> | Be wary of copy-paste web-developers too. They'll spoil all your plans. |
| 21:45 | <annevk3> | though still doable using mostly the same algorithm as <link sizes>, you'd just have the separator as variable |
| 21:45 | <annevk3> | KrocCamen, HTML5 is designed for them :) |
| 21:45 | <KrocCamen> | :D |
| 21:46 | <Hixie> | Philip`: i believe it's so that they can change things on the fly |
| 21:47 | <Hixie> | but if there is pushback on pixelratio, please e-mail the list so that i can point them to the pushback |
| 21:52 | <nessy> | how does YouTube notice the wrong pixel aspect ratio anyway? |
| 21:52 | <Hixie> | right now they don't |
| 21:53 | <roc> | surely there's a great machine learning algorithm for that |
| 21:53 | <nessy> | if they don't notice it, how can an attribute fix it? |
| 21:53 | <mpilgrim> | hixie: do you object to installing analytics on the whatwg blog? |
| 21:54 | <nessy> | are they expecting users to fix it in their embed (uhh video) tags? |
| 21:54 | <KrocCamen> | Interesting side note. The CSS @media queries use a slash—16/9. |
| 21:56 | <Hixie> | mpilgrim: go ahead |
| 21:58 | <gsnedders> | mpilgrim: Some of the Atom Autodiscovery tests are broken |
| 21:58 | <mpilgrim> | really? |
| 21:58 | <mpilgrim> | which ones? |
| 21:58 | gsnedders | checks |
| 21:58 | <gsnedders> | Some on remote domain |
| 21:59 | <annevk3> | Hixie, why not let <video> be like <img> for .height/.width? |
| 22:00 | <gsnedders> | mpilgrim: 47–50 |
| 22:00 | <Hixie> | annevk3: i don't really care either way, feel free to send feedback on it |
| 22:00 | <gsnedders> | mpilgrim: They fail on both my impl. and your impl, and it really does appear to be broken looking at it |
| 22:01 | <annevk3> | I guess I don't either, it just seems that the string behavior is less desirable |
| 22:02 | <Hixie> | i just changed it to be consistent with the things it was already consistent with |
| 22:02 | <Hixie> | i have no opinion |
| 22:03 | <eric_carlson> | annevk3: what do you mean by "be like <img> for .height/.width"? |
| 22:03 | <annevk3> | eric_carlson, see http://html5.org/tools/web-apps-tracker?from=2332&to=2333 |
| 22:04 | <eric_carlson> | annevk3: ah, thanks |
| 22:11 | <mpilgrim> | hixie: i don't have access to the design section of blog.whatwg.org |
| 22:11 | <mpilgrim> | perhaps lachy could give me access? |
| 22:12 | <Hixie> | lachy is your man, yes |
| 22:12 | <mpilgrim> | gsnedders: i'll try to look into it tonight |
| 22:12 | <annevk3> | do you need admin access? |
| 22:12 | <annevk3> | or something else? |
| 22:12 | <gsnedders> | mpilgrim: Much thanks |
| 22:12 | <mpilgrim> | i need access to the design section so i can change the template to include the tracking code |
| 22:13 | <mpilgrim> | i don't know much about the different levels of wordpress privileges |
| 22:13 | <mpilgrim> | but i need more than i have right now :) |
| 22:13 | <annevk3> | you're an admin now |
| 22:13 | <annevk3> | feel free to abuse your new power |
| 22:13 | <annevk3> | oh, wait! |
| 22:14 | <mpilgrim> | mwahahahaha |
| 22:14 | <mpilgrim> | that did it, thanks |
| 22:14 | <Hixie> | jesus christ, i got about 50 e-mails on video overnight |
| 22:18 | <annevk3> | seems that "issues" have gained some ground again http://www.whatwg.org/issues/data.html |
| 22:18 | <annevk3> | guess that was mostly due to WF2 |
| 22:19 | <Hixie> | the uptick was when i took the huge mess of notes at the bottom of the html5 spec as well as the list of notes at the bottom of the wf2 spec and merged them into one clean list of notes with XXX markers for each issue |
| 22:20 | <annevk3> | yeah, I meant e-mail issues though |
| 22:20 | <annevk3> | the green line |
| 22:21 | <Hixie> | oh |
| 22:21 | <Hixie> | yeah |
| 22:21 | <Hixie> | six weeks without replying to mail will do that |
| 22:24 | <annevk3> | sigh |
| 22:24 | <annevk3> | http://feeds.feedburner.com/~r/ajaxian/~3/421961005/fancyupload-for-flash-10 |
| 22:24 | <annevk3> | we need HTMLInputElement.fileList and XHR.send(FileList) soonish |
| 22:31 | <KrocCamen> | Gtg. Have a nice evening all. |
