| 06:34 | <hsivonen> | gsnedders: HTML 4 didn't deprecate <b>! beware of the propaganda! |
| 08:39 | <MikeSmith> | hsivonen: |
| 08:39 | <MikeSmith> | ./htmlparser/src/nu/validator/htmlparser/impl/Tokenizer.java:38: cannot find symbol |
| 08:39 | <MikeSmith> | symbol : class Auto |
| 08:39 | <MikeSmith> | location: package nu.validator.htmlparser.annotation |
| 08:39 | <MikeSmith> | import nu.validator.htmlparser.annotation.Auto; |
| 08:39 | <MikeSmith> | ^ |
| 08:39 | <MikeSmith> | ./htmlparser/src/nu/validator/htmlparser/impl/Tokenizer.java:39: cannot find symbol |
| 08:39 | <MikeSmith> | symbol : class CharacterName |
| 08:39 | <MikeSmith> | location: package nu.validator.htmlparser.annotation |
| 08:39 | <MikeSmith> | import nu.validator.htmlparser.annotation.CharacterName; |
| 08:39 | <MikeSmith> | ^ |
| 08:40 | <MikeSmith> | etc. |
| 08:41 | <hsivonen> | MikeSmith: hmm. have I forgotten to hg add stuff... |
| 08:41 | <MikeSmith> | yeah, figured that might be it |
| 08:41 | <hsivonen> | yes, I have. I suck. sorry |
| 08:41 | <MikeSmith> | no problem man :) |
| 08:42 | <MikeSmith> | no loss of life caused |
| 08:42 | <MikeSmith> | yet |
| 08:44 | <hsivonen> | MikeSmith: added & pushed |
| 08:45 | <MikeSmith> | thanks |
| 08:45 | MikeSmith | rebuilds |
| 08:46 | <MikeSmith> | hsivonen: ah, back to still getting these: |
| 08:46 | <MikeSmith> | ./htmlparser/src/nu/validator/htmlparser/impl/TreeBuilder.java:1229: method does not override a method from its superclass |
| 08:46 | <MikeSmith> | @Override public void zeroOriginatingReplacementCharacter() |
| 08:46 | <MikeSmith> | ^ |
| 08:46 | <MikeSmith> | ./htmlparser/src/nu/validator/htmlparser/impl/TreeBuilder.java:5220: method does not override a method from its superclass |
| 08:46 | <MikeSmith> | @Override public boolean cdataSectionAllowed() throws SAXException { |
| 08:46 | <MikeSmith> | ^ |
| 08:47 | <hsivonen> | let's see... |
| 08:51 | <hsivonen> | MikeSmith: how about now? |
| 08:52 | hsivonen | wonders why the compiler wasn't whining at me |
| 08:53 | <annevk> | http://www.benedelman.org/hardcoding/ -- not unexpected |
| 08:57 | <hsivonen> | MikeSmith: now I see what's happening. I'm using Java 1.6 compiler and you are probably using a Java 1.5 compiler |
| 08:58 | <hsivonen> | MikeSmith: and the semantics of @Override changed between 1.5 and 1.6 to allow the use that I had in the source file |
| 09:01 | <MikeSmith> | hsivonen: yeah |
| 09:06 | <MrWax> | does anyone have an overview of the new css3 attributes that come with html5? |
| 09:08 | <roc> | MrWax: CSS3 is not part of HTML5 |
| 09:09 | <MrWax> | Well, a lot people seem to define CSS3 WITH the HTML5 standard |
| 09:09 | <MrWax> | i know its not the same |
| 09:09 | <ako> | http://www.brucelawson.co.uk/2010/meet-newt-new-exciting-web-technologies/ |
| 09:09 | <ako> | NEWT :> |
| 09:19 | <annevk> | MrWax, but now you're asking people defining the HTML5 standard |
| 09:28 | <annevk> | it seems we need to set up our own charset registry |
| 09:28 | <annevk> | reasoning with these guys is not really feasible |
| 09:29 | <annevk> | tried twice now |
| 09:29 | <annevk> | i actually think it should just be one spec that has all the mappings |
| 09:29 | <annevk> | no registry needed |
| 09:29 | <annevk> | it's all legacy anyway |
| 09:30 | <hsivonen> | annevk: who are "these guys" |
| 09:30 | hsivonen | is lacking context |
| 09:31 | <hsivonen> | oh charset registry |
| 10:19 | <annevk> | is markp member of the WG? |
| 10:19 | <annevk> | I think that is another req |
| 10:19 | <jgraham> | Oh, it could be |
| 10:20 | <jgraham> | If not he should join |
| 10:20 | <jgraham> | Or get someone else to do the commit |
| 10:20 | <annevk> | http://www.w3.org/2000/09/dbwg/details?group=40318&public=1 -- he is not |
| 10:22 | jgraham | will self-reply |
| 11:07 | <annevk> | hmm, CSS blog uses en-EU |
| 11:08 | <erlehmann> | can anyone tell me what difference gecko and webkit have so that webkit does load the script and gecko (20100916) does not? http://daten.dieweltistgarnichtso.net/src/blinkenworld/index.html |
| 11:22 | <hsivonen> | annevk: it's a valid language tag |
| 11:22 | jgraham | is lost in WebIDL (again) |
| 11:23 | <hsivonen> | erlehmann: OpenLayers does UA sniffing for "Safari" and "MSIE" |
| 11:23 | <jgraham> | Where does it define how readonly maps into ECMAScript |
| 11:23 | <hsivonen> | erlehmann: Firefox 4 has neither of those substrings, so the script breaks |
| 11:23 | <jgraham> | i.e. which ES properties it corresponds to |
| 11:23 | <hsivonen> | erlehmann: http://trac.osgeo.org/openlayers/ticket/2933 |
| 11:24 | <annevk> | hsivonen, can you do en-NL and such too? |
| 11:24 | <annevk> | or en-FI |
| 11:24 | <hsivonen> | annevk: sure |
| 11:24 | <erlehmann> | hsivonen, UA sniffing bad. but this does not seem to be my problem? |
| 11:24 | <annevk> | good times |
| 11:25 | <hsivonen> | erlehmann: I don't know of other problems with OpenLayers |
| 11:26 | <erlehmann> | >please run the document.write()-based code path in all browsers |
| 11:26 | <erlehmann> | ahahahaha |
| 11:28 | <hsivonen> | well, if you want to block the parser, document.write is what HTML5 gives you |
| 11:28 | <erlehmann> | hsivonen, you seem to be an expert on javascript execution order. is there any gooy way to ensure that OpenLayers is loaded until I load my script? |
| 11:29 | <hsivonen> | erlehmann: ooh. you have defer in there |
| 11:29 | <hsivonen> | erlehmann: that won't work |
| 11:29 | <hsivonen> | *never* use defer on a script that might call document.write |
| 11:30 | <hsivonen> | and we know this script calls document.write in some cases |
| 11:30 | <erlehmann> | i didn't. |
| 11:30 | <erlehmann> | hmmpf |
| 11:31 | <hsivonen> | erlehmann: is there an OpenLayers tutorial recommending defer, or did you use defer on your own initiative? |
| 11:31 | <erlehmann> | i thought it was some canvas based solution. until i looked. will probably encounter many more “interesting” things. |
| 11:32 | <erlehmann> | hsivonen, i used it on my own initiative because i wanted to defer script execution. after all, the document should be fully ready, my thought. |
| 11:32 | <jgraham> | OK, lazyirc, is there any reason that calling delete on a readonly property should not return false? Assuming we are not in strict mode. |
| 11:32 | <hsivonen> | erlehmann: anyway, I can't answer your question with certainty without examining all the code in OpenLayers, but my guess is your script will load after OpenLayers if you remove defer from both |
| 11:32 | <jgraham> | e.g. delete document.body |
| 11:33 | <hsivonen> | erlehmann: defer is rarely any good |
| 11:33 | <hsivonen> | erlehmann: it still defers the load event, so it doesn't help with perceived performance |
| 11:33 | <hsivonen> | erlehmann: defer mainly exists for backwards compat at this point |
| 11:34 | <hsivonen> | async actually makes sense for some use cases, but unfortunately async scripts are specced to delay the load event, too |
| 11:34 | <erlehmann> | hsivonen, i removed it, but it does not seem to help. thanks anyway |
| 11:35 | <hsivonen> | erlehmann: not helping in which browser? |
| 11:36 | <erlehmann> | firefox 3.6 |
| 11:37 | <hsivonen> | oh. 3.6 |
| 11:37 | <hsivonen> | no idea what's going on then |
| 11:37 | <hsivonen> | erlehmann: if you put "Safari" in your UA string, does the script work? |
| 11:40 | <hsivonen> | jgraham: fwiw, my support case about the Ubuntu jumpiness got closed without any real solution except not trying to do so much stuff on this hardware |
| 11:41 | <hsivonen> | jgraham: however, on the linux kernel mailing list, there's now a patch that could make my case better |
| 11:41 | <hsivonen> | jgraham: so the options seem to be waiting for natty or do-it-your-self kernel patching and compiling |
| 11:42 | <virtuelv> | HTMLMediaElement.playbackRate fails to define what should happen to the audio when increasing the playback rate |
| 11:42 | <erlehmann> | hsivonen, changes nothing. seems to be something else that makes OpenLayers not werk. |
| 11:43 | <hsivonen> | erlehmann: ok |
| 11:43 | <hsivonen> | virtuelv: worth filing a spec bug, I guess |
| 11:43 | <virtuelv> | hsivonen: probably, but I'm unsure which is the right behavior |
| 11:44 | <virtuelv> | for me, it would have been decidedly useful if the playback rate altered the pitch, instead of working as a timestretch |
| 11:44 | <virtuelv> | but I can see why this is not sensible for the general case |
| 11:47 | <jgraham> | hsivonen: yeah, I saw think thing about the kernel patch |
| 11:47 | <jgraham> | *the thing |
| 11:47 | <erlehmann> | hsivonen, got it. openlayers cannot into empty divs. |
| 11:48 | <erlehmann> | if i insert a node, everything is fine. |
| 11:48 | <erlehmann> | weird. |
| 11:48 | <erlehmann> | even a text node does work. |
| 11:48 | <erlehmann> | :( |
| 12:33 | <zcorpan> | i wonder if mozilla people can agree on whining in the error console for websockets |
| 12:56 | <hsivonen> | http://www.w3.org/Bugs/Public/show_bug.cgi?id=11326 looks so totally out of scope to me |
| 12:57 | <hsivonen> | esp. considering the drive to LC |
| 12:58 | <annevk> | it's filed after the LC date |
| 13:22 | <virtuelv> | hsivonen: that proposal looks like it belongs in at least a number of other wg's |
| 13:23 | <virtuelv> | probably webapps |
| 13:24 | <virtuelv> | or even an entirely new WG |
| 13:24 | <hsivonen> | virtuelv: is the proposal asking for features one would put in a general purpose browser? |
| 13:25 | <hsivonen> | virtuelv: or is it asking for features one would put in a set top box that reuses a browser engine to implement a cable TV UI? |
| 13:26 | <virtuelv> | I see the value in having local service discovery in a local browser |
| 13:26 | <virtuelv> | s/local/general purpose/ |
| 13:26 | <hsivonen> | I see. I hope the local services aren't trying to be entangled into MPEG stuff |
| 13:26 | <virtuelv> | so, say you have a web-based media player service |
| 13:26 | <hsivonen> | (4 or 2) |
| 13:27 | <virtuelv> | the local media server could potentially offer video sources for playback through the browser |
| 13:27 | <hsivonen> | virtuelv: so if you have a local box that serves WebM video, you could already navigate to it, no? |
| 13:28 | <virtuelv> | hsivonen: the point is that your local videos could be directly integrated with youtube |
| 13:29 | <virtuelv> | so: go to youtube, see a mix of local and remote media in the sidebar |
| 13:29 | <hsivonen> | virtuelv: oh. |
| 13:29 | <Philip`> | It sounded to me like a general purpose desktop browser should be able to discover set top boxes on the local network, so e.g. you could visit a TV channel's web site and it could find you've got a box and do <video src="rtmp://192.168.12.34"> and call some API to tell the box to switch to the right channel, or something |
| 13:29 | <virtuelv> | that, however, sounds downright scary |
| 13:29 | <virtuelv> | and/or fun |
| 13:30 | <Philip`> | (with appropriate security controls etc) |
| 13:30 | <virtuelv> | then again, I am only commenting on the UPnP service discovery part here, not that |
| 13:32 | <hsivonen> | my knee-jerk reaction is that it sounds scary that a Web site out there could query for services on the other side of the firewall |
| 13:32 | <hsivonen> | even with user authorization |
| 13:39 | <heycam> | jgraham, if you didn't get an answer earlier: readonly attributes in IDL currently map to configurable accesor properties with a getter but no setter |
| 13:39 | <heycam> | jgraham, and thus calling delete on them will succeed |
| 13:40 | <heycam> | (previously, when Web IDL targetted ES3, these properties were ReadOnly ones and you couldn't delete them) |
| 13:41 | <hsivonen> | what would be a good tool for creating nice-looking flowcharts in SVG? |
| 13:42 | <hsivonen> | stuff like http://hsivonen.iki.fi/doctype/ie8-mode.png but in SVG |
| 13:42 | <heycam> | hsivonen, hmm not really sure. i know lots of people like OmniGraffle for mac. (which i believe can output SVG.) |
| 13:42 | <hsivonen> | heycam: last time I tried OmniGraffle Pro, its SVG export was rather broken |
| 13:42 | <heycam> | hsivonen, was that done with graphviz? |
| 13:42 | <hsivonen> | maybe I should try again |
| 13:42 | <hsivonen> | heycam: OmniGraffle non-Pro |
| 13:42 | <heycam> | ah |
| 13:42 | <paul_irish> | http://svg-edit.googlecode.com/svn/trunk/editor/svg-editor.html maybe? |
| 13:43 | <annevk> | that is kind of neat |
| 13:43 | <hsivonen> | paul_irish: looks like a generic drawing tool. does it support anchoring lines to flowchart boxes? |
| 13:43 | <annevk> | even supports fn+backspace |
| 13:44 | <paul_irish> | i doubt it. :) |
| 13:47 | <hsivonen> | hmm. $99.95 for SVG export... |
| 13:47 | <jgraham> | heycam: In what sense suceed? |
| 13:47 | <jgraham> | *succeed |
| 13:47 | <hsivonen> | (upgrade from OmniGraffle 5 to 5 Pro) |
| 13:47 | <heycam> | jgraham, succeed as in return true and remove the property |
| 13:47 | <heycam> | hsivonen, ouch |
| 13:47 | <jgraham> | heycam: That seems very wrong |
| 13:47 | <jgraham> | delete document.body shouldn't have any effect |
| 13:48 | <jgraham> | But it should return true |
| 13:48 | <jgraham> | AFAICT |
| 13:48 | <jgraham> | So I am very confused |
| 13:48 | <heycam> | jgraham, orly? |
| 13:48 | heycam | thinks a bit |
| 13:49 | <heycam> | does the delete operator only delete own properties? |
| 13:49 | <jgraham> | heycam: http://software.hixie.ch/utilities/js/live-dom-viewer/saved/706 |
| 13:49 | <jgraham> | Good question; I will look |
| 13:50 | <heycam> | because Web IDL now specifies that the accessor properties for IDL attributes go on the prototype |
| 13:53 | <jgraham> | heycam: Ah, it does seem to be an own property only operation |
| 13:54 | <heycam> | jgraham, ok cool then no problem :) |
| 13:54 | <jgraham> | So delete document.__proto__.body should work? |
| 13:54 | <heycam> | yes |
| 13:57 | <jgraham> | http://software.hixie.ch/utilities/js/live-dom-viewer/saved/707 |
| 13:57 | <jgraham> | heycam: Did I do something wrong? |
| 13:58 | <heycam> | jgraham, no but probably the spec doesn't match all browsers yet |
| 13:58 | <gsnedders> | jgraham: Do you foresee anyone actually making use of a expected_results field? |
| 14:14 | <jgraham> | heycam: Is it desirable to be able to delete DOM properties from the prototype? |
| 14:51 | <heycam> | jgraham, it's desirable that they can be reconfigured -- just as you can overwrite the properties that correspond to IDL operations |
| 14:54 | <jgraham> | heycam: Why? |
| 14:54 | <jgraham> | It seems like it would be poor form to ever write a script that actually did that |
| 14:54 | <heycam> | jgraham, so you can do monkey patching of the dom |
| 14:55 | <jgraham> | Hmm |
| 14:55 | <jgraham> | Anyway, I have way I need |
| 14:55 | <jgraham> | *what |
| 14:55 | <heycam> | ok |
| 14:55 | <heycam> | someone was suggesting that a select few properties would be non-configurable |
| 14:55 | <matjas> | Does anyone know why/when the text/javascript MIME type was declared “obsolete”? http://tools.ietf.org/html/rfc4329#section-7.1 |
| 14:55 | <heycam> | like window.location (i think) |
| 14:56 | <matjas> | annevk: I’ve read your old blog posts (plural) on the JavaScript MIME type but couldn’t find the answer there |
| 14:56 | <jgraham> | heycam: (thanks, btw :) |
| 14:56 | <heycam> | jgraham, np :) |
| 14:57 | <jgraham> | matjas: I think it is some silly reason like people thinking it is prettier to use application/ for scripts |
| 14:58 | <zcorpan> | isn't the reason the same as text/xml being deprecated? |
| 14:58 | <matjas> | jgraham: I couldn’t think of any good reason either. text/javascript has great browsers support, according to http://krijnhoetmer.nl/stuff/javascript/mime-types/ |
| 14:58 | <matjas> | zcorpan: What was the reason for that? |
| 14:59 | <annevk> | JavaScript media types are somewhat messed up |
| 14:59 | <zcorpan> | matjas: encoding must be us-ascii if there's no charset= parameter, according to some RFC |
| 14:59 | <annevk> | servers do not need to use them, for instance |
| 15:01 | <zcorpan> | matjas: in practice, nobody implements the us-ascii requirement, servers traditionally use application/x-javascript for .js files, browsers ignore the server-provided mime type completely for <script src>, and <script type> has best browser compat with text/javascript (or omitted type altogether) |
| 15:02 | <matjas> | annevk: Oh, the MIME type can be blank? Would be interesting to test browser support for that. |
| 15:03 | <matjas> | zcorpan: I don’t understand the reasoning there. charset (as specified through the Content-Type header) has nothing to do with the plain MIME type AFAICS |
| 15:05 | <zcorpan> | matjas: yes it does. for text/* the default encoding (according to the unimplemented RFC, don't know the number) is us-ascii unless the charset parameter is set, and the message body can't override it |
| 15:06 | <zcorpan> | http://tools.ietf.org/html/rfc2046 has the us-ascii requirement |
| 15:06 | <MikeSmith> | hsivonen: thanks for the parser changes |
| 15:07 | <MikeSmith> | everything seems to be working as expected now |
| 15:07 | <zcorpan> | http://tools.ietf.org/html/rfc3023 has details around text/xml |
| 15:08 | <hsivonen> | MikeSmith: nice |
| 15:08 | <MikeSmith> | I'm updating the validator.w3.org instances now |
| 15:09 | <MikeSmith> | they were pretty far out of date with upstream |
| 15:09 | <matjas> | zcorpan: I see. Thanks for your explanation. |
| 15:12 | <zcorpan> | matjas: http 1.1 has a different encoding default, and the html and css specs require that the message body can set the encoding for text/html and text/css (actually text/css and text/javascript can inherit the encoding from the embedding document) |
| 15:13 | <zcorpan> | matjas: so what mime and http 1.1 say are mostly academic since higher level specs don't follow it (except text/xml, but nobody implements it that way) |
| 15:18 | <hsivonen> | bah. gedit can't handle named-character-references.html from the multipage spec |
| 15:18 | <hsivonen> | crashed |
| 15:32 | <matjas> | zcorpan: So basically, text/javascript can still be used, even though it’s officially “obsolete”. |
| 15:33 | <matjas> | It doesn’t really matter much, but `text/javascript` is obviously shorter than `application/javascript` or `application/x-javascript`. |
| 15:33 | <matjas> | Good to know! |
| 15:35 | <zcorpan> | matjas: yeah. my conclusion is "don't touch the server config for .js files since it doesn't matter anyway" and "just use "<script>", type="" is not necessary" |
| 15:36 | <matjas> | I knew about the latter ;) |
| 16:30 | <Philip`> | About submitting tests: Given that it's a DVCS, maybe mpilgrim should just push to a public copy of the repository and then someone with W3C access can push them across |
| 16:30 | <Philip`> | Seems a bit silly to get hung up by lack of access permissions |
| 16:32 | <zcorpan> | "2. Microsoft engineers working on Internet Explorer 9 could have been using the SunSpider Benchmark and unintentionally over-optimized the JavaScript engine for the SunSpider Benchmark." - http://digitizor.com/2010/11/17/internet-explorer-9-caught-cheating-in-sunspider-benchmark/ |
| 16:32 | <zcorpan> | how can you over-optimize something? |
| 16:32 | <jgraham> | People seem a bit concerned about having the logs correspond to who submitted what |
| 16:32 | <jgraham> | zcorpan: If you optimize for a specific case at the expense of the general case |
| 16:33 | <jgraham> | It's a classic thing to do wrong in machine learning |
| 16:33 | <jgraham> | where you make something that works perfectly for the training data but poorly for other data |
| 16:34 | <zcorpan> | ah. ok. |
| 16:35 | <jgraham> | (I am not really sure why people are worried about the accuracy of the logs or whether having someone else push mpilgrim's changes would be a problem) |
| 18:59 | <tabatkin1> | Hixie: You around? |
| 19:00 | <annevk> | hsivonen, did you donate in a way that made money not show up at local chapters? |
| 19:01 | <annevk> | hsivonen, I am undecided yet, but I do not really care for my money ending up at the Dutch chapter of Wikipedia |
| 19:02 | <annevk> | anyone taken a look at what markp submitted? |
| 19:43 | <annevk> | omg so much grandstanding again in http://blogs.msdn.com/b/ie/archive/2010/11/17/html5-and-real-world-site-performance-seventh-ie9-platform-preview-available-for-developers.aspx |
| 19:45 | <miketaylr> | i like how html5 is the first word...for an article on chakra |
| 19:48 | <jgraham> | I like the fact that they bemoan microbenchmarks and then only present sunspider results, the microest of all microbenchmarks |
| 19:49 | <jgraham> | If there was any more irony, the article would have serious problems walking past magnets |
| 19:55 | <espadrine> | Just been on http://ie.microsoft.com/testdrive/, and there it was, written in big bitmap letters, "Cool, you're using a Chrome 9 nighly build!" But what exactly would have been "un-cool"? Internet Explorer. |
| 19:56 | <jgraham> | mpilgrim: It looks like you just changed the mode of some Microsoft tests to me. Or did I miss something |
| 19:57 | <mpilgrim> | ah, that would explain it |
| 19:57 | <mpilgrim> | i've backed it out |
| 19:57 | <mpilgrim> | and will resubmit from a fresh clone |
| 19:59 | <jgraham> | The new values looked more sensible :) |
| 19:59 | <mpilgrim> | i was testing privately on a remote server and running into permission problems |
| 20:00 | <mpilgrim> | some of my tests re-use Microsoft's common/media.js to get a supported video or audio URI |
| 20:05 | <jgraham> | Hmm, I see that it is not sufficiently obvious how I intended the harness to be used |
| 20:05 | <jgraham> | This is my fault of course |
| 20:07 | <jgraham> | But there are clearly cases where more of the test setup could be in the step() |
| 20:07 | <karlcow> | http://norman.walsh.name/2010/11/17/deprecatingXML |
| 20:14 | <mpilgrim> | jgraham: i am happy to update tests if i'm using the harness incorrectly, or even just suboptimally |
| 20:15 | <mpilgrim> | presumably these will be rolled into an actual harness someday |
| 20:15 | <mpilgrim> | all the tests i just submitted were designed to be run automatically |
| 20:16 | <mpilgrim> | but i haven't done anything wild and crazy like actually testing that or anything |
| 20:17 | <jgraham> | mpilgrim: It is mostly a style point. My "vision" (although it was really nothing so grandiose) was that one would wrap almost all code for a particular test in either test.step(function(){}) or test(function(){}) (for sync tests) |
| 20:18 | <jgraham> | That way unexpected failures in some part of the setup code would be caught by the harness and other tests on the page could still run |
| 20:18 | <mpilgrim> | i am happy to do that |
| 20:18 | <mpilgrim> | but i've opted for one test per page |
| 20:18 | <jgraham> | I like one test per page :) |
| 20:19 | <jgraham> | But it doesn't always make sense |
| 20:19 | <mpilgrim> | as it will make submitting bugs easier |
| 20:19 | <jgraham> | e.g. if you are autogenerating 10,000 tests from some javascript |
| 20:19 | <jgraham> | Making 10,000 files a quite the overhead |
| 20:20 | <jgraham> | So the harness(which should have been called framework) supports both |
| 20:20 | jgraham | remembers that the reason it was not called "testframework.js" was that we already had something with that name on an internal server |
| 20:20 | <mpilgrim> | yes, i saw AryehGregor's reflection tests |
| 20:21 | <jgraham> | :) |
| 20:21 | <mpilgrim> | one file that generates enough tests to expose a dozen bugs in 3 browsers... that's impressive |
| 20:22 | <mpilgrim> | so far i've only found bugs in IE |
| 20:22 | <mpilgrim> | and webkit, only insofar as we don't support the <video audio> attribute yet |
| 20:23 | <mpilgrim> | no, i take it back, i found 1 bug in opera |
| 20:23 | <mpilgrim> | tests/submission/Google/video/canPlayType/canPlayType_application_octet_stream.html |
| 20:23 | <mpilgrim> | fails in opera |
| 20:23 | <zcorpan> | that's known :) |
| 20:24 | <zcorpan> | (the spec used to require 'maybe' for that case) |
| 20:24 | <mpilgrim> | yes, it's a strange thing |
| 20:24 | <mpilgrim> | my test case was actually the exact opposite until i read the spec three more times |
| 20:25 | <zcorpan> | heh |
| 20:25 | <mpilgrim> | in one place it says "The MIME type "application/octet-stream" with no parameters is never a type that the user agent knows it cannot render." |
| 20:25 | <mpilgrim> | but elsewhere it says "The canPlayType(type) method must return the empty string if type is a type that the user agent knows it cannot render or is the type "application/octet-stream"" |
| 20:26 | <mpilgrim> | tests are hard |
| 20:27 | <mpilgrim> | let's create a grammar to express assertions and replace the spec with a bunch of those |
| 20:28 | <espadrine> | seems like the complexity is externalized to the spec in this case! |
| 20:31 | <zcorpan> | mpilgrim: some tests have 30 or 60 seconds timeout if i'm reading it right. do they need to be that long? |
| 20:32 | <mpilgrim> | make media.w3.org faster |
| 20:32 | <mpilgrim> | whoa |
| 20:33 | <mpilgrim> | apparently when you join the working group, you start getting email |
| 20:33 | <mpilgrim> | i was wondering where i had gone wrong in my life that roy fielding was in my inbox |
| 20:34 | <mpilgrim> | Matches: list:"<public-html.w3.org>" Do this: Skip Inbox |
| 20:37 | <annevk> | :) |
| 21:21 | <zcorpan> | http://www.favbrowser.com/browser-adoption-rate-looks-good/ - i wonder how much of the "other versions" is nightly/alpha/beta/dev of the *next* release |
| 21:55 | <jgraham> | zcorpan: I'm pretty sure it is almost none |
| 21:56 | <jamesr_> | i hope they mean chrome >= 7.0 |
| 21:58 | <Hixie> | hsivonen: yeah, those bugs were filed in the same era that i was in support of xhtml2 and where acid2 was written assuming we couldn't change the spec |
| 21:58 | <Hixie> | and where i thought quirks mode was a good idea |
| 21:58 | <Hixie> | and where dbaron and i thought the way to fix css was to define the inline box model and margin collapsing to be what CSS2 implied it had to be, rather than doing something compatible with UAs |
| 22:02 | <jgraham> | Hixie: This multi-day lag is confusing. It's like getting echos from the past |
| 22:02 | <Hixie> | heh |
| 22:02 | <Hixie> | sorry, was offlien yesterday |
| 22:02 | <Hixie> | offline, too |
| 22:03 | <jgraham> | :) |
| 22:04 | <jgraham> | jamesr_: Why would you assume that? They explicitly say "equal to" |
| 22:04 | <jamesr_> | well that would mean their article is worthless |
| 22:05 | <jgraham> | It doesn't even seem to properly account for point releases |
| 22:05 | <jgraham> | Their article probably is worthless, but not because some tiny fraction of people are running dev versions |
| 22:06 | <jamesr_> | the % of chrome users running 6.0 and lower is way less than 10% |
| 22:07 | <jgraham> | Well if you already know that, the article is clearly worthless because you already know the result :) |
| 22:07 | <jamesr_> | i only know the %s for chrome |
| 22:08 | <jgraham> | Well they claim that 90% of people run chrome exactly equal to 7 |
| 22:08 | <jgraham> | If that doesn't match your numbers then one or both of you are wrong |
| 22:09 | <TabAtkins> | I'm confused: http://www.w3.org/Bugs/Public/show_bug.cgi?id=11333 What the hell does that even mean? |
| 22:09 | <jamesr_> | there's a decent number of users on chrome 8 and chrome 9 currently |
| 22:10 | <zcorpan> | TabAtkins: i read it to mean that "title" is outdated (no idea why), and thus we should mint another name |
| 22:11 | <jamesr_> | who filed that? |
| 22:13 | <TabAtkins> | jamesr_: Someone random. |
| 22:16 | <jgraham> | TabAtkins: If you are going to try to decipher every lunatic bug report, we could be here a while |
| 22:16 | <TabAtkins> | I was wondering if it was a lunatic, or it was actually saying something reasonable. |
| 22:17 | <jgraham> | Well given that they appear to want to rename the <title> tag for no particular reason, my crankometer is showing a definite reading |
| 22:19 | <Philip`> | I assumed they meant title="", not <title> |
| 22:20 | <jgraham> | Would that explain why it is filed agianst the 2d context spec? |
| 22:21 | <jgraham> | (I guess @title would make marginally more sense than <title>. But not a lot more) |
| 22:21 | <Philip`> | Isn't the component derived automatically by Hixie's script? |
| 22:22 | <Philip`> | (presumably incorrectly in this case) |
| 22:22 | <TabAtkins> | This is a #top bug, so it's filed on a W3C-version spec. |
| 22:22 | <TabAtkins> | Presumably the canvas 2d one? |
| 22:25 | <Philip`> | If they mean the title attribute, maybe they mean the spec's own usage of the attribute |
| 22:25 | <Philip`> | since it's used for crossreferences rather than for titling |
| 22:25 | <jgraham> | I thought maybe they mean the usage of <title> in the example. Although I see nothing wrong with it |
| 22:27 | <jgraham> | But the @title usage is more reasonable |
| 22:27 | <jgraham> | Sadly W3C pubrules prevent the obvious solution being employed |
| 22:27 | <jgraham> | (data attributes) |
| 22:28 | <Philip`> | Maybe we could guess an email address based on the hostname of the IP address of the submitter, and ask for clarification |
| 22:46 | <Hixie> | you can tell which spec the bug was filed from from the Specification: link which is derived from the Referer: header sent to my script |
| 22:49 | <Philip`> | Where does the Component field come from? |
| 22:49 | <Hixie> | it's in the form |
| 22:55 | <Philip`> | Ah |
