| 01:03 | <bga_> | :) |
| 01:04 | <bga_> | night guys |
| 01:35 | <itissid> | I want to understand a trend here... Is it a given that most of google gears will/already has moved into HTML5 and google gears will become a legacy project... |
| 02:00 | <MikeSmith> | itissid: Gears has already become a legacy project, quite a while back |
| 02:00 | <MikeSmith> | all the work is going instead into trying to put features into the core Web platform itself |
| 02:00 | <MikeSmith> | to get them implemented natively across multiple browsers |
| 02:01 | <itissid> | Is there a page showing the cross browser adoption of the features specific to gears that were moved to HTML5 |
| 02:01 | <MikeSmith> | not that i know of |
| 02:01 | <MikeSmith> | most people working on this stuff tend to be pretty busy and pages like that don't get put together so often |
| 02:02 | <MikeSmith> | not updated once they have been put together |
| 02:02 | <MikeSmith> | do to lack of time to get them written up and maintained |
| 02:02 | <MikeSmith> | in part because things are just changing very quickly |
| 02:03 | <itissid> | hmm |
| 02:03 | <MikeSmith> | we are in the midsts of what is by far the most rapid pace of significant new features going into the Web platform that there has ever been in the entire history of the Web |
| 02:05 | <itissid> | Yeah I noticed... I had not worked in gears at all... Then i made an Chrome Extension which used things like local storage and Webworkers and then read that this was all a gears idea initially... |
| 02:08 | <MikeSmith> | yup |
| 02:12 | <Aleoss> | Hello everyone. |
| 09:50 | <matjas> | http://twitter.com/WHATWG hasn’t been posting updates since Sept 1st — intentional? |
| 09:50 | <matjas> | Not switched to oAuth yet? |
| 09:51 | <zcorpan> | matjas: iirc nobody noticed when it broke so Hixie figured that it was pointless |
| 09:51 | <matjas> | zcorpan: Real shame, I loved that stream. |
| 09:52 | <jgraham> | matjas: You are exactly the second person to complain |
| 09:52 | <slartsa> | oh there was something in twitter |
| 09:53 | <matjas> | jgraham: Damn. I’d noticed it before, just never realized it was because of Twitter switching to oAuth. |
| 09:54 | <matjas> | How does everyone keep track of spec changes nowadays (besides this channel, that is)? There doesn’t seem to be a syndication feed either. |
| 09:55 | <hsivonen> | matjas: subscribing to bugmail and spec change mail |
| 09:55 | <hsivonen> | matjas: and keeping open an insane number of bug / spec tracker tabs for later inspection |
| 09:55 | <matjas> | hsivonen: I was afraid so :) |
| 09:55 | <jgraham> | matjas: http://html5.org/tools/web-apps-tracker |
| 09:56 | <jgraham> | You could probably convince annevk to add a feed |
| 09:56 | <matjas> | jgraham: I know; sadly refreshing that page every once in a while isn’t really handy. A feed would be nice. |
| 09:56 | <matjas> | Consider this my +1 on fixing the @WHATWG bot. It was perfect. |
| 09:57 | <hsivonen> | matjas: also, I expect someone to land tests for spec changes that I've missed myself |
| 09:57 | <hsivonen> | matjas: so I'll find out when the html5lib test suite no longer passes |
| 09:57 | jgraham | honestly uses hsivonen's appraoch a lot |
| 09:57 | <jgraham> | That is "if I miss something it will become obviosu over time" |
| 09:57 | <hsivonen> | jgraham: that's scary. I'm relying on you noticing stuff! |
| 09:57 | <jgraham> | hsivonen: Well I do notice stuff |
| 09:58 | <matjas> | Surely there must be a better way… |
| 09:58 | <jgraham> | But not everything |
| 09:58 | <jgraham> | You notice some stuff I miss, for example |
| 09:58 | <jgraham> | (but generally I keep more up to date with parser changes than other areas of the spec) |
| 09:59 | <matjas> | annevk, Hixie: Would you mind adding a syndication feed to http://html5.org/tools/web-apps-tracker AND/OR adding oAuth support to the @WHATWG Twitter bot? I really liked that guy. |
| 10:00 | <annevk> | matjas, can you fix the WHATWG bot? |
| 10:00 | <annevk> | matjas, I think all it takes is for someone to write the OAuth part |
| 10:00 | <matjas> | annevk: I would certainly like to give it a try. What’s it written in? |
| 10:02 | <annevk> | I would assume Perl, but I think any service would work for Hixie |
| 10:02 | <annevk> | he just needs a simple service that he can hand the message that puts it on twitter |
| 10:04 | <matjas> | In that case, adding a syndication feed and passing it through something like TwitterFeed sounds like the easiest solution. |
| 10:05 | <matjas> | But I’ll see if I can write a simple service where he can POST to. Would be much better. |
| 10:05 | <annevk> | I'm afraid for syndication feed load with html5.org |
| 10:05 | <annevk> | though I suppose I could use one of those services that only fetches it every 5 minutes and takes over the load |
| 10:06 | <annevk> | Google bought one of those right? |
| 10:09 | <jgraham> | Fetching every five minutes seems way to often. ANyway, I think you might be thinking of FeedBurner |
| 10:10 | <annevk> | I'll think it over; I'd prefer just fixing the twitter API |
| 10:10 | <hsivonen> | annevk: you could have a cron job dumping the data into a plain file instead of hitting svn for each GET |
| 10:11 | <annevk> | I could do lots of things to complicate my life, but do I want to? |
| 10:12 | <jgraham> | MikeSmith: Did you finish the GitHub pull of the spec? |
| 10:13 | <MikeSmith> | jgraham: no, it timed out without completing :( |
| 10:13 | <jgraham> | MikeSmith: Not that surprising; I did it locally and ended yo with a 5Gb repo |
| 10:13 | <MikeSmith> | good god |
| 10:13 | <jgraham> | After a few hours |
| 10:13 | <MikeSmith> | epic |
| 10:16 | <MikeSmith> | matjas: I wonder if CIA.vc or somewhere might have some way of generating a feed from svn |
| 10:16 | MikeSmith | looks |
| 10:16 | <MikeSmith> | or oloh |
| 10:17 | <MikeSmith> | ohloh |
| 10:17 | <annevk> | looking for http://www.youtube.com/watch?v=sdShzYpa6i8 ? |
| 10:19 | <MikeSmith> | that's sick |
| 10:20 | <jgraham> | MikeSmith: So if we put the Git repo on html5.org and synced it using cron, one could write out an atom file using a post-recieve hook. The only problem is that people cloning the repo would burn through bandwidth |
| 10:21 | <MikeSmith> | hmm |
| 10:21 | <jgraham> | But the feed would just be a static file and so play nicely with web arch |
| 10:22 | <MikeSmith> | I agree with annevk sentiment about complicating life |
| 10:22 | <jgraham> | :) |
| 10:22 | <MikeSmith> | CIA.vc does feeds |
| 10:22 | <MikeSmith> | feed://cia.vc/stats/project/Chromium/.rss |
| 10:23 | <MikeSmith> | feed://cia.vc/stats/project/V8/.rss |
| 10:23 | <MikeSmith> | etc. |
| 10:23 | <MikeSmith> | so I can will add the svn rep to CIA.vc now |
| 10:23 | <MikeSmith> | then we get IRC integration too |
| 10:24 | <MikeSmith> | to #html-wg or here or wherever |
| 10:24 | <MikeSmith> | well, not integration |
| 10:24 | <MikeSmith> | but notifications at least |
| 10:24 | <jgraham> | I think I would like notifications here, but maybe others would disagree |
| 10:26 | <MikeSmith> | project name: html5-spec ? |
| 10:26 | <MikeSmith> | html-spec ? |
| 10:26 | <jgraham> | html is better |
| 10:26 | <MikeSmith> | k |
| 10:26 | <jgraham> | webapps is perhaps more accurate |
| 10:26 | <jgraham> | well sort of |
| 10:27 | Philip` | would probably prefer not to get floods of automated notifications into here, lest this becomes like #html-wg |
| 10:28 | <jgraham> | Would it be floods? What's the median/max number of checkins per day? |
| 10:29 | <Philip`> | If it's only SVN checkins then it might not be bad |
| 10:29 | <jgraham> | Yeah, I wouldn't want bugspam here |
| 10:31 | <MikeSmith> | feed://cia.vc/stats/project/html-spec/.rss |
| 10:31 | <zcorpan> | i'd prefer if notifications were left for #html-wg and #whatwg is just discussion :) |
| 10:37 | <matjas> | My Perl skills are lame, but this looks useful AFAICT: http://ashbentiel.mine.nu/tweet.pl |
| 10:38 | <annevk> | now all we need is wake Hixie |
| 10:38 | <hsivonen> | I'm rather impressed that the Windows 7 Action Center manages to link crashes of 3rd-party software to relevant documentation on the site of the 3rd-party software provider |
| 10:39 | <hsivonen> | maybe Hixie is on a Thanksgiving vacation |
| 10:40 | <zcorpan> | http://www.xkcd.com/786/ |
| 10:41 | <jgraham> | You think that Hixie will get left off the web rollercoaster if he takes vacation? |
| 10:43 | <zcorpan> | i'm just worried we'll have implemented all his specs while he's away... |
| 10:44 | <jgraham> | And built nuke-riding city ships just to alleviate the boredom? |
| 10:44 | <zcorpan> | yeah |
| 10:44 | <jgraham> | Good thing he's in America then. Vacation is basically against the law, so he'll be back soon |
| 10:45 | <jgraham> | Although imagine how much more ionteresting standards flame wars would be if the opposing factions had nuke-riding city ships under their command |
| 10:48 | <matjas> | “No img/@longdesc in HTML5? BOOM!” |
| 10:49 | <zcorpan> | well then you leave us no choice but to blow up the earth |
| 10:50 | jgraham | now has some sort of standards/Girl Genius mashup going on in his head Adventure! Romance! Mad Standards! |
| 10:55 | <jgraham> | Does anyone remember that page with a huge number of nested <strong> elements? |
| 11:06 | <matjas> | On http://html5.org/tools/web-apps-tracker, isn’t this wrong? function readCookie(name) { return localStorage["tracker-" + ""] } |
| 11:07 | <matjas> | Why would you want to + "" if not for implicit toString coercion? |
| 11:07 | <matjas> | Should probably be something like function readCookie(name) { return localStorage["tracker-" + name] }, right? |
| 11:08 | <zcorpan> | yeah |
| 11:09 | <zcorpan> | feel free to fix, it's in http://code.google.com/p/html5/source/checkout |
| 11:10 | <matjas> | Cool, I did not know that! |
| 11:44 | <MikeSmith> | I assume from hsivonen 's latest tweet that he must be working on a sanitzer |
| 11:48 | <MikeSmith> | ah sweet, I got another patch landed in V8 source |
| 11:48 | <MikeSmith> | http://code.google.com/p/v8/source/detail?r=5893 |
| 11:51 | <hsivonen> | MikeSmith: just planning at this point |
| 11:52 | <MikeSmith> | ok |
| 11:52 | <hsivonen> | MikeSmith: since the sanitizer isn't targeted to Firefox 4, all Firefox 4 blockers get service first |
| 11:52 | <MikeSmith> | yeah, I can imagine |
| 11:53 | <jgraham> | hsivonen: Have you looked at what the html5lib sanitizer does? It has some (crazy, regexp based) CSS filtering |
| 11:53 | <hsivonen> | jgraham: I haven't |
| 11:53 | <hsivonen> | I was planning on providing an interface for pluggable CSS sanitizers |
| 11:53 | <jgraham> | Which you need in order to remove scripts |
| 11:53 | <hsivonen> | so that the Gecko version could use the existing Gecko CSS sanitizer and the Java version could use something else |
| 11:54 | <jgraham> | At least if you target browsers that can run scripts from CSS |
| 11:54 | <hsivonen> | yeah, using a CSS sanitizer is part of the requirements |
| 11:55 | <hsivonen> | at some point, I think I will need to learn which SVG attributes are defined in terms of CSS in such a way that they can carry dangerous payloads |
| 12:00 | <hsivonen> | jgraham: whoa. html5lib is much more liberal with allowed protocols than I had expected |
| 12:01 | <hsivonen> | otoh, there are fewer acceptable CSS properties than I expected |
| 12:01 | <hsivonen> | also, Semantics MathML isn't allowed |
| 12:01 | <jgraham> | hsivonen: (I have not been much involved with this part of the library) |
| 12:02 | <jgraham> | In particular the list of allowed stuff mainly came from Sam and Jacques |
| 12:02 | <hsivonen> | hmm. so it sanitizer between the tokenizer and the tree builder |
| 12:02 | <hsivonen> | instead of sanitizing after the tree builder when it's known if something is an HTML, SVG or MathML element |
| 12:02 | <jgraham> | I think it predates foreignContent |
| 12:03 | <jgraham> | |Foreign Content" |
| 12:03 | <jgraham> | s/|/"/ |
| 12:06 | <jgraham> | (that is, althought there is support for SVG elements, that was based on Sam's experimental liberal XHTML parser. Although presumably that assigned namespaceness in treebuilder too) |
| 12:09 | <hsivonen> | for blog comments and such, it might be useful to have a feature for normalizing all the "semantic" aliases of <i> to <i> for simpler styling |
| 12:17 | <hsivonen> | hmm. does http://www.whatwg.org/specs/web-apps/current-work/multipage/parsing.html#determining-the-character-encoding ever care about /mode/ being "charset"? |
| 12:18 | <hsivonen> | it seems to me it only cares about it being "pragma" vs not |
| 12:20 | <hsivonen> | oh. it does care |
| 12:21 | <hsivonen> | so <meta charset="UTF-8" content="text/html; charset=UTF-8"> doesn't sniff to UTF-8 |
| 12:21 | <hsivonen> | but <meta content="text/html; charset=UTF-8" charset="UTF-8"> does |
| 12:21 | <hsivonen> | looks like spec bug to me... |
| 12:22 | <hsivonen> | I hate it when I find spec bugs when I just want to implement and be done |
| 12:29 | <zcorpan> | hsivonen: how does it not sniff to utf-8? "If the attribute's name is "charset": If charset is still set to null, let charset be the encoding corresponding to the attribute's value, and set mode to "charset"." |
| 12:32 | <hsivonen> | zcorpan: the content attribute sets mode again, this time to pragma |
| 12:32 | <hsivonen> | zcorpan: when content="" comes after charset="" |
| 12:33 | <zcorpan> | hsivonen: oh, i missed step 9 |
| 12:34 | <zcorpan> | hsivonen: yeah, spec bug |
| 12:34 | jgraham | wonders is document.written meta charset has an effect |
| 12:34 | <jgraham> | *if |
| 12:35 | <hsivonen> | jgraham: In Gecko, it doesn't |
| 12:35 | <hsivonen> | jgraham: but Hixie WONTFIXed the spec bug: http://www.w3.org/Bugs/Public/show_bug.cgi?id=10803 |
| 12:35 | <hsivonen> | I didn't change Gecko according to WONTFIXing |
| 12:35 | <jgraham> | Ah. I had a feeling it was something like that |
| 12:35 | <jgraham> | Any idea what webkit or IE do? |
| 12:36 | <hsivonen> | jgraham: testing other browsers and reopening certain bugs is on my todo list |
| 12:42 | <hsivonen> | whoa. the "in head" handling of the charset attribute seems bogus as far as the UTF-16 part goes |
| 12:42 | <hsivonen> | sigh. |
| 12:43 | <hsivonen> | the difference between the spec and Gecko is larger than I thought |
| 12:44 | <hsivonen> | oh. UTF-16 gets handled later |
| 13:14 | <annevk> | damnit http://www.ragingplatypus.com/ is offline |
| 13:14 | <annevk> | fix it |
| 14:01 | <hsivonen> | aargh. what's wrong with validator.nu *now*? |
| 14:06 | <hsivonen> | looks like a network problem |
| 14:27 | <matjas> | hsivonen: Fixed already? Works fine here/now |
| 14:27 | <hsivonen> | matjas: does validator.nu work or only html5.validator.nu? |
| 14:28 | <hsivonen> | matjas: the latter is unaffected iff you happen to have CSS and JS cached already |
| 14:28 | <matjas> | hsivonen: both worked just a minute ago, but now only html5.validator.nu seems to work |
| 14:29 | <matjas> | hsivonen: does validator.nu have a cache manifest? |
| 14:29 | <hsivonen> | matjas: no |
| 14:30 | <matjas> | must've been some weird browser caching then |
| 14:31 | <hsivonen> | I guess I should make html5.validator.nu load CSS and JS from itself instead of loading them from the server that runs about.validator.nu and validator.nu |
| 14:36 | <hsivonen> | hmm. maybe bugzilla.validator.nu has a bad robots.txt after reinstall and a crawler is indexing it |
| 14:51 | <fortes> | What's the best way to express a boolean property using html5 microdata? |
| 14:52 | <annevk> | there's no design patterns afaik |
| 14:52 | <annevk> | HTML5 itself uses presence/absence of attributes for booleans |
| 14:52 | <fortes> | annevk: So perhaps I should mint a data- attribute to indicate my boolean property? |
| 14:53 | <fortes> | Something like <div itemscope data-foo> ? |
| 14:53 | <Lachy> | fortes, depends what you're trying to use it for |
| 14:53 | <annevk> | data- != microdata |
| 14:54 | <fortes> | I'm using metadata to construct a table of contents |
| 14:54 | <fortes> | Where each item is an article |
| 14:54 | <Lachy> | microdata is useful when you want to share data with 3rd parties via a shared vocabulary. data-* is useful if you just want extra metadata for your own scripts to use internally on the page. |
| 14:54 | <fortes> | And I need the ability to set a boolean flag or two on the articles |
| 14:55 | <fortes> | Microdata also seems to be useful when the content is the data as well, no? |
| 14:55 | <fortes> | Like an article's title |
| 14:57 | <Lachy> | yes, it can be, depending on what use cases and problems your trying to solve. |
| 14:57 | <fortes> | From what I've played around with it seemed that <h2 itemprop="title">Article Title</h2> was a better choice than <div data-title="..."><h2>.... |
| 14:57 | <Lachy> | you could just use class="title" |
| 14:58 | <fortes> | True, although I was concerned with name conflicts |
| 14:58 | <fortes> | goal is also to eventually be able to use native document.getItems |
| 14:59 | <Lachy> | that's not a goal, that's a specific solution for achieving some other goal. |
| 15:01 | <Lachy> | you could just as easily use document.querySelectorAll("[data-title]"); or .querySelectorAll(".title"); |
| 15:02 | <fortes> | slightly more complicated since there is a bit of structure. each article has a title, author, etc |
| 15:02 | <fortes> | but yes, since doc.getItems doesn't exist, I've been implementing it manually |
| 15:03 | <Lachy> | ok, so then you have a requriement for obtaining easily scriptable and structured data from your page? |
| 15:05 | <fortes> | yes. balancing performance with ease of authoring |
| 15:07 | Philip` | sees PNaCl, and wonders what the "native" in its name means |
| 15:08 | <Philip`> | since it's no longer got native machine code (except to the extent that e.g. JavaScript does, via JITs), and you can't use native APIs or anything like that |
| 15:09 | <Philip`> | Seems the main thing it adds that's fundamentally impossible with JS is shared-memory concurrency, which everyone nowadays hates |
| 15:12 | <matjas> | Weird, document.images is 70% slower than getElementsByTagName('img') in Chrome. In every other browser the images collection is faster. |
| 15:12 | webr3 | mentions something about RDFa Core 1.1 and that kind of data.. |
| 15:13 | <matjas> | Well, /s/Chrome/WebKit/ |
| 15:15 | <gsnedders> | So, hmm… fun doing analyis of Sputnik failures. The two engines that have actually tried to impl ES5 have 74 and 75 failures, but only 50 of those are shared between the two |
| 15:15 | <gsnedders> | Of those 50 failure, there are seven of those that both SpiderMonkey and V8 pass. |
| 20:50 | <AryehGregor> | Interesting: http://blog.benstrong.com/2010/11/google-and-microsoft-cheat-on-slow.html |
| 20:50 | <AryehGregor> | Probably a good idea. |
| 20:50 | <AryehGregor> | (Cheating on slow start, I mean.) |
| 23:43 | <JonathanNeal> | And hello. |
| 23:43 | <_bga> | hi |