| 00:55 | <karlcow> | Hixie: i think people wants a "highsrc" |
| 00:56 | <karlcow> | as in save bandwidth/more responsive when on mobile. so if <img src="foo" lowsrc="tinyfoo"/> downloads everything foo and tinyfoo. We have not made any progress. |
| 00:57 | <karlcow> | personally I do not think 2 sizes solve a lot of use case either. |
| 00:57 | <karlcow> | the other issue is that the screen size is only one context. |
| 00:58 | <karlcow> | small screen size on a wifi is not the same thing than small screen size on 3g |
| 01:05 | <necolas> | karlcow: that's right. screen size is not an accurate proxy of available bandwidth |
| 01:53 | <AryehGregor> | In Ubuntu 11.10, you don't need to enter your password to apply updates. That is a small but amazingly awesome usability improvement. |
| 01:53 | <AryehGregor> | Because the update dialog thing shows up like every day. |
| 06:38 | hsivonen | wonders if http://lists.w3.org/Archives/Member/w3c-css-wg/2011OctDec/0239.html meets the criteria for messages that are eligible to be posted to the Member list instead of the public list under the CSS WG mailing list policy |
| 07:53 | <Taggnostr> | is there an easy-to-parse version of http://www.w3.org/TR/html5/named-character-references.html available somewhere? |
| 08:10 | <MikeSmith> | Taggnostr: maybe http://www.w3.org/2003/entities/2007xml/unicode.xml |
| 08:11 | <Taggnostr> | not sure xml is easier to parse than html -- I was hoping in a plain txt like the ones used by the Unicode consortium |
| 08:11 | <MikeSmith> | (the source of that, not the rendered version) |
| 08:11 | <Taggnostr> | but thanks anyway |
| 08:11 | <MikeSmith> | um, OK |
| 08:12 | <Taggnostr> | e.g. http://www.unicode.org/Public/6.0.0/ucd/NamedSequences.txt |
| 08:16 | <zcorpan> | Taggnostr: http://gsnedders.html5.org/html5.txt |
| 08:20 | <Taggnostr> | thanks |
| 08:21 | <Taggnostr> | the fact that some entities map to 2 chars is a bit of a problem for application that expect them to map to a single char (and e.g. use a char type for the result) |
| 08:30 | <Taggnostr> | is that list definitive or is it possible that more entities will be added in future? |
| 08:32 | <zcorpan> | it's not impossible that more entities will be added, but i would say it's unlikely |
| 08:36 | <Taggnostr> | ok |
| 08:41 | <hsivonen> | https://twitter.com/#!/glazou/status/141436497143414785 |
| 08:43 | <MikeSmith> | hsivonen: he has such a wonderful, understated way of expressing his views |
| 08:43 | <MikeSmith> | very diplomatic |
| 08:43 | <MikeSmith> | exactly what one would want in a working-group chair |
| 08:46 | <hsivonen> | Taggnostr: I use the character reference table that in the spec and split it using regular experessions |
| 08:46 | <hsivonen> | (yes, consuming HTML using regexps in order to build an HTML parser) |
| 08:46 | <hsivonen> | s/that in/that is in/ |
| 08:47 | <Taggnostr> | I used the parser that needs the entities to parse the page that contains it :) |
| 08:47 | <Taggnostr> | s/it/them/ |
| 08:49 | <Taggnostr> | btw, is there any reason why the U+XXXXX notation is used there? usually U+XXXX is used for BMP chars, and U+XXXXX and U+XXXXXX are used for non-BMP ones |
| 08:49 | <Taggnostr> | it's just a minor nit, but it looks a bit weird to me |
| 08:51 | <zcorpan> | file a bug |
| 08:52 | <zcorpan> | the spec uses U+XXXX elsewhere |
| 08:55 | <Taggnostr> | is the "feedback comments" form ok too? |
| 08:56 | <Taggnostr> | can I also provide a patch? |
| 08:56 | <zcorpan> | the feedback form is ok (it files a bug) |
| 08:56 | <zcorpan> | the table is generated so i don't think a patch on the html source will help |
| 08:57 | <Taggnostr> | then I guess the script that generates it will need to be patched |
| 08:57 | <zcorpan> | yeah |
| 08:58 | <zcorpan> | afaik that script isn't public |
| 08:59 | <Taggnostr> | ok |
| 09:13 | <tantek> | hsivonen - I specifically stated my preference to have licensing discussions in public in that thread. |
| 09:13 | <tantek> | not sure what else I can do except just continue to do the work itself in public |
| 09:52 | <david_carlisle> | Taggnostr: the 5 digit convention is probably my fault, I use that internally in unicode.xml (to allow easy naive sorting of filenames etc) but (after comments on earlier drafts) I switch to 4 digits for BMP range in the generated spec, so I guess Hixie should do same in the htnl spec |
| 11:07 | <annevk> | tantek: I emailed once explaining where I come from |
| 11:08 | <annevk> | tantek: if someone wants to discuss next steps I suggest we request that happens in public |
| 11:19 | <hsivonen> | tantek, annevk: thank you for preferring discussion in public |
| 11:20 | <annevk> | kind of painful the whole thing |
| 11:21 | <annevk> | I wish people were a bit nicer about it |
| 11:21 | <annevk> | the other thing is that the only interactions I have with these people is when they want to badmouth me for something they think I did wrong |
| 11:21 | <annevk> | it's a very negative relationship |
| 11:22 | <jgraham> | I toughtthere was only one person who wanted everything to happen behind closed doors |
| 11:22 | <jgraham> | *thought there |
| 11:25 | <jgraham> | Is there any precedent for dispatching events to window and workers |
| 11:25 | <jgraham> | ? |
| 11:25 | <jgraham> | It seems like the device orientation people are considering it |
| 11:25 | <annevk> | as long as it's async that might be fine |
| 11:26 | <annevk> | I don't think there's a precedent |
| 11:26 | <annevk> | and it might be a bit inconsistent I guess |
| 11:30 | <jgraham> | It feels a bit weird to me. I mean I can see the appeal, but it's not hard to forward the important points as messages to the worker if you want to |
| 11:30 | <jgraham> | This would be much easier with a polling-based design :| |
| 11:30 | <annevk> | are we going to do the same for scroll/hashchange etc.? |
| 11:31 | <annevk> | this is why it's bad to have people working on features in a vacuum |
| 11:31 | <annevk> | they do what makes some sense to me without considering the platform as a whole |
| 11:33 | <michel_v> | to them, you mean? |
| 11:33 | <annevk> | oops yes |
| 11:34 | <michel_v> | it always comes down to "Very clever people, working alone, make very clever mistakes" |
| 11:35 | <smaug____> | saying something happens behind the the closed doors makes one curious what is discussed there... |
| 11:36 | <annevk> | smaug____: see http://krijnhoetmer.nl/irc-logs/whatwg/20111128#l-443 onwards |
| 11:40 | <smaug____> | huh |
| 12:15 | <annevk> | http://lists.w3.org/Archives/Public/public-html/2011Oct/0108.html o_O |
| 12:19 | <zcorpan> | why do we support Link: again? |
| 12:20 | <hsivonen> | it seems this would be the easiest to solve by unsupporting Link |
| 12:21 | <hsivonen> | zcorpan: I gather it's mostly Hixie's and annevk's fault for writing tests |
| 12:21 | <hsivonen> | annevk: the CP has the usual Conformance Classes Changes: None. |
| 12:22 | <annevk> | I was willing to cave on Link btw |
| 12:22 | <annevk> | Hixie still likes it |
| 12:23 | <annevk> | not sure if it's in Acid3 or not |
| 12:23 | <zcorpan> | it's not |
| 12:23 | <annevk> | nuking it works for me |
| 12:23 | <annevk> | but I think it might be in 3 browsers by now? |
| 12:24 | <zcorpan> | did ie implement it? |
| 12:25 | <zcorpan> | http://simon.html5.org/articles/mobile-results is unstyled in chrome |
| 13:05 | <hsivonen> | who is operating DuckDuckGo and what's their business model? |
| 13:07 | <wilhelm_> | It was just one guy until recently. |
| 13:08 | <hsivonen> | is it basically an anonymizer on top of Yahoo! (which itself gets data from Bing)? |
| 13:08 | jgraham | hopes their business model is independent of whether the name is too embarassing for word-of-mouth marketing to work |
| 13:09 | <wilhelm_> | No, I think it's more fancy than that. Here's the guy: http://www.gabrielweinberg.com/ |
| 13:10 | <hsivonen> | jgraham: their name seems more catchy than e.g. Teoma |
| 13:10 | <erlehmann> | i started using duckduckgo solely because it is not as overbloated as other frontends and does not second-guess like google. |
| 13:11 | <jgraham> | I can't imagine ever telling my mother to use a search engine called DuckDuckGo. Assuming she didn't mishear me and think I was talking about a porn search engine, she would laugh at me |
| 13:11 | <erlehmann> | in before someone mentioning the gimp |
| 13:12 | <erlehmann> | also in before someone mentioning photoshop, wondering aloud if that is a program to buy color prints online. |
| 13:12 | <hsivonen> | I don't know English porn slang well enough to understand what makes duckduckgo sound like a porn search engine |
| 13:12 | <erlehmann> | just checked the yahoo front page. it is still full of crap :/ |
| 13:12 | <jgraham> | I was siomply thinking s/D/F/g |
| 13:12 | <hsivonen> | jgraham: I see |
| 13:13 | <jgraham> | (if it does become popular that site will *so* exist) |
| 13:13 | <hsivonen> | I thought the association you are supposed to make is duck, duck, goose |
| 13:14 | <erlehmann> | http://www.boardgamegeek.com/boardgame/35964/duck-duck-go |
| 13:14 | <erlehmann> | >duck! duck! GO! comes with 6 REAL rubber duckies, and a rubber bird dog. There are over 100 duckies in the set - which ones will YOU get? |
| 13:28 | <sakamal> | hello |
| 13:29 | <jgraham> | hi |
| 14:19 | <mpt> | erlehmann, Photoshop 1.0's icon: http://blog.cocoia.com/2008/the-first-photoshop-icon/ |
| 14:28 | <hsivonen> | where's the useful set of Mac OS X settings that was on github and was called something like ~/.osx? |
| 14:28 | <hsivonen> | the name of the project turns out to be ungooglable |
| 14:29 | <hasather> | hsivonen: https://github.com/mathiasbynens/dotfiles/blob/master/.osx? |
| 14:29 | <hsivonen> | hasather: yes. thank you |
| 14:34 | <hsivonen> | hmm. I was almost sure that list of tricks had a way to speed up network shares by making Finder not look inside files to generate icons |
| 14:34 | <hsivonen> | but now the list seems to be lacking that |
| 14:38 | <hsivonen> | for some reason, accessing my NAS from Lion is terribly, terrible slow compared to Snow Leopard or Ubuntu |
| 14:41 | <michel_v> | hsivonen: over AFP? there's been changes in Lion, so your NAS needs an update if the vendor released one |
| 14:42 | <hsivonen> | michel_v: I've updated the NAS |
| 14:42 | <hsivonen> | michel_v: I turned off AFP just now. Left the CIFS service running on the NAS |
| 14:42 | <hsivonen> | curiously, my pre-existing Mac aliases pointing to the NAS still work |
| 14:43 | <hsivonen> | even though I thought I was using AFP previously |
| 14:43 | <hsivonen> | I think it's still very uncool of Apple to make perf suck with unupdated AFP servers |
| 14:50 | <annevk> | volkmar: battery strength could be used across pages |
| 14:50 | <annevk> | volkmar: not sure if that is exposed though |
| 14:50 | <annevk> | volkmar: I only briefly looked at the draft |
| 14:53 | <volkmar> | annevk: battery strength? |
| 14:54 | <volkmar> | you mean level? |
| 14:54 | <annevk> | yeah seems like I mean that |
| 14:56 | <hsivonen> | hmm. maybe I've accidentally used CIFS from the Mac all along and I should switch the Mac to using AFP... |
| 14:59 | <annevk> | volkmar: in any event, it does not seem like a serious issue |
| 15:00 | <hsivonen> | how can accessing local network shares suck this badly in Lion |
| 15:08 | <volkmar> | annevk: knowing the battery level doesn't seem like a privacy treat to me |
| 15:08 | <volkmar> | I mean, the battery level is a changing information |
| 15:09 | <annevk> | it's fingerprinting of some kind but it's not exactly stable |
| 15:09 | <annevk> | / reliable |
| 15:09 | <volkmar> | annevk: yes, it could be used to fingerprint but hardly |
| 15:10 | <annevk> | hopefully |
| 15:10 | <hsivonen> | maybe for cross-site navigation. not for return visits |
| 15:10 | <annevk> | apparently deviceorientation can be used to figure out what keys you press |
| 15:13 | <jgraham> | Wow, is it really that sensitive? Or is the attack different to what I am imagining? |
| 15:14 | <annevk> | sensitive I believe |
| 15:18 | <hsivonen> | annevk: do you mean the device moves a little when you press keys and device orientation can pick that up? |
| 15:19 | <jarek> | what's the current state of http://dev.w3.org/csswg/css3-page/? |
| 15:19 | <jarek> | does any browser implement it? |
| 15:19 | <jarek> | the grammar seems to be out of sync with the examples |
| 15:20 | <miketaylr> | here's the original paper, hsivonen (i believe): www.cs.ucdavis.edu/~hchen/paper/hotsec11.pdf |
| 15:20 | <annevk> | hsivonen: see e.g. http://www.gatech.edu/newsroom/release.html?nid=71506 |
| 15:21 | <miketaylr> | oh cool, haven't seen this one |
| 15:23 | <hsivonen> | miketaylr, annevk: thanks |
| 15:29 | <divya> | no implementation yet jarek |
| 15:29 | <jarek> | divya: even in Opera? AFAIR the spec was pushed by Opera |
| 15:30 | <divya> | we have implementation for this http://dev.w3.org/csswg/css3-gcpm/ |
| 15:45 | <annevk> | argh |
| 15:45 | <annevk> | qualified name versus prefix/local name/namespace make attributes complex |
| 15:49 | <jarek> | why @page atrule can have only one pseudo-class? |
| 15:50 | <jarek> | e.g. I can have either "@page :first" or @page :left", but not @page :first:left" |
| 15:53 | <scor> | hsivonen: will validator.nu validate such markup one day? <link href="page.html" rel="foaf:page"> |
| 15:53 | <scor> | is it pending some spec to become more stable, or it is the way it is now by design? |
| 15:55 | <scor> | validator.nu choose HTML5+ARIA, SVG 1.1 plus MathML 2.0 (experimental) for the page I was testing |
| 16:07 | <annevk> | the other extremely silly thing is that setAttributeNS takes a qualified name |
| 16:10 | <annevk> | ohg |
| 16:10 | <annevk> | given |
| 16:10 | <annevk> | document.head.setAttributeNS("test", "x:a", "a") |
| 16:10 | <annevk> | document.head.setAttributeNS("test", "e:a", "b") |
| 16:10 | <annevk> | Gecko actually changes the prefix! |
| 16:11 | <annevk> | which is what the spec says :) |
| 16:11 | <annevk> | but not what Opera/WebKit do |
| 16:11 | MikeSmith | krijn |
| 16:12 | <MikeSmith> | oofs |
| 16:12 | <krijn> | Yes? |
| 16:12 | krijn | MikeSmith too! |
| 16:45 | <smaug____> | hmm, 3.5% of XHRs are still sync :/ |
| 16:45 | <smaug____> | unfortunately I don't know which sites are using sync XHR |
| 16:52 | <annevk> | smaug____: how many of those are cross-origin? |
| 16:52 | <smaug____> | dunno |
| 16:52 | <smaug____> | I could add some more probes |
| 16:52 | <annevk> | I'm happy to make that change sicking suggested |
| 16:53 | <annevk> | not sure if it's preferable for the spec to lead there though |
| 17:26 | <annevk> | maybe I should define |
| 17:26 | <annevk> | append "attribute" |
| 17:26 | <annevk> | change attribute value |
| 17:26 | <annevk> | remove attribute |
| 17:27 | <annevk> | append/remove/change |
| 17:44 | <AryehGregor> | It's absolutely astonishing how much time I wind up spending on wedding-related stuff 13 days before. |
| 17:44 | <AryehGregor> | (how much being basically all of it) |
| 17:44 | AryehGregor | has not done a single minute of work since last Wednesday |
| 17:45 | jgraham | is absolutely not astonished |
| 17:45 | annevk | wishes he had a break in the same period |
| 17:46 | <jgraham> | annevk: I thought you were just playing Zelda :p |
| 17:46 | <dglazkov> | AryehGregor: congrats! |
| 17:46 | <annevk> | AryehGregor: hopefully you can enjoy it when it's there :) |
| 17:47 | <annevk> | jgraham: that's what I should do :) |
| 17:47 | <annevk> | jgraham: finally got those damn flames, now these otherworldly powers are asking me see some dragons |
| 17:47 | <annevk> | jgraham: never stops, that game |
| 17:47 | <divya> | AryehGregor: congratulations! |
| 17:48 | <AryehGregor> | dglazkov, divya: Thanks. |
| 17:48 | jgraham | should finish the last Zelda game :p |
| 17:49 | <Philip`> | AryehGregor: 13 days of pre-wedding effort seems insignificant to the lifetime of post-wedding effort that will be needed |
| 17:50 | <AryehGregor> | Philip`, well, right, but I'm expecting that the post-wedding effort won't be so much as to actually preclude me from working. I think my wife would take exception to that. |
| 17:50 | <jgraham> | Depends, maybe his new bridge will push a toaster in the bath on their honeymoon |
| 17:50 | <jgraham> | *bride |
| 17:50 | <jgraham> | I assume you are not marrying a bridge… |
| 17:50 | AryehGregor | doesn't take baths, so is safe |
| 17:51 | <AryehGregor> | No, but if you're interested, I have a lovely bridge to sell you. |
| 17:51 | <AryehGregor> | Scenic view, busiest vehicular bridge in the world, roughly two blocks from my house. |
| 17:51 | <AryehGregor> | Only $50,000. |
| 17:54 | <AryehGregor> | I don't have the deed handy, but I'll be able to get it to you in six months if you pay now. |
| 17:55 | <AryehGregor> | Just need to find it buried somewhere under the sofa cushions. |
| 19:32 | <rniwa> | good morning #webkit |
| 19:33 | <rniwa> | oops I mean #whatwg |
| 19:33 | <AryehGregor> | rniwa, FYI, if you didn't know, I'm being really slow to respond to anything because I'm getting married in 13 days. |
| 19:33 | <AryehGregor> | So if something is urgent, be sure to ping me and I can look. |
| 19:34 | <rniwa> | AryehGregor: oh, congratulations! |
| 19:34 | <AryehGregor> | But otherwise it will probably stay on the backburner for a couple of weeks. |
| 19:34 | <AryehGregor> | Thanks! |
| 19:34 | <rniwa> | AryehGregor: well, it seems like everyone's slow in responding to anything in december so no problem :) |
| 19:34 | <rniwa> | AryehGregor: have you moved to Israel already? |
| 19:35 | <AryehGregor> | rniwa, no, that will be a few months after we get married. |
| 19:35 | <rniwa> | AryehGregor: ah, ok/ |
| 19:49 | <gsnedders> | Hah, so IE10pp4 finally has XHR+CORS. Wonder if they still have XDR? |
| 19:52 | <erlehmann> | gsnedders, what is the „pp“ for? |
| 19:56 | <gsnedders> | erlehmann: Platform preview |
| 19:57 | <erlehmann> | gsnedders, meaning what exactly? |
| 19:58 | <erlehmann> | is that microsoftspeak for “beta” ? |
| 20:00 | <gsnedders> | erlehmann: pre-alpha, really |
| 20:00 | <gsnedders> | Or alpha, depending on how you define terms |
| 20:01 | <erlehmann> | gsnedders, what do they say for “beta” ? in before odd release numbers. |
| 20:04 | <gsnedders> | "Beta" |
| 20:04 | <gsnedders> | And then "Release Candidate" |
| 20:04 | <gsnedders> | The platform preview isn't really a browser. |
| 20:05 | <gsnedders> | It's just a thin wrapper around mshtml.dll |
| 20:05 | <gsnedders> | The platform previews really have little relation to where they are in the dev phase |
| 20:05 | <gsnedders> | IE9 had PPs after the Beta |
| 20:35 | <jgraham> | Right, platform preview is actually quite literal |
| 20:35 | <jgraham> | It is a preview of the *platform* i.e. the rendering engine/scripting engine/etc. |
| 20:35 | <jgraham> | Not of the browser |
| 20:36 | <jgraham> | i.e. it doesn't have any UI |
| 20:36 | <jgraham> | (or any UI that will actually be released) |
| 20:56 | <L04f3r> | Is IE10 really that great as they claim it to be: http://samples.msdn.microsoft.com/ietestcenter/#svg11e2 |
| 20:57 | <L04f3r> | SVG compliance beating both Chrome and Firefox |
| 20:59 | <miketaylr> | based on the 74 submitted tests, probably |
| 20:59 | <L04f3r> | Well actually on the same page they claim IE9 Svg is not that bad either, |
| 21:00 | <L04f3r> | I think they just pick and choose what to test against. |
| 21:07 | <hober> | gaaaah |
| 21:08 | <hober> | I always expect bugs.w3.org or bugzilla.w3.org to go to the right place |
| 21:09 | <hober> | MikeSmith: do you think we could get a redirect or two in place? |
| 21:11 | <bga_> | gsnedders http://mobile.twitter.com/kangax/status/141623286927851521 |
| 21:13 | <jgraham> | hober: Oh, I thought it was just me |
| 21:13 | <jgraham> | I *always* type bugs.w3.org |
| 21:13 | <jgraham> | Showing a disturing inability to learn |
| 21:20 | <gsnedders> | bga_: known |
| 21:22 | <timeless> | AryehGregor: mazal tov |
| 21:24 | <timeless> | L04f3r: what microsoft does is pick certain areas to do work on |
| 21:24 | <timeless> | and then they do work on them |
| 21:24 | <timeless> | in those areas, they also advertise that they've done work |
| 21:24 | <timeless> | this is all perfectly rational |
| 21:27 | <roc> | Microsoft tends to not submit tests that they don't pass |
| 21:28 | <jgraham> | It is also true that they often pass tests that they have written. Or do well in benchmarks they have written. This is not really surprising. In general be very worried about attaching too much weight to single-vendor tests of any kind. Or any tests really. But at least if it is something where multiple people have independly contributed to there is a hope it won't suck |
| 21:28 | <dbaron> | roc, neither does any browser vendor other than Mozilla, I think |
| 21:28 | <roc> | the fairest way to interpret their test results is to compare the non-IE browsers against each other and ignore the IE results :-) |
| 21:29 | <gsnedders> | dbaron: We release tests we fail… along with a greater number we pass. :) |
| 21:29 | <jgraham> | We have even been known to release whole testsuites we do badly at (XHR2) |
| 21:30 | <gsnedders> | Well, we did equally badly when we released that, no? |
| 21:30 | <dbaron> | hmmm, I think somebody from Opera once told me they're not supposed to do that, but maybe they do, and that's good :-) |
| 21:30 | <gsnedders> | It's just others improved more quickly than us. :P |
| 21:31 | <jgraham> | gsnedders: I don't think so. But I might be wrong. |
| 21:31 | <astearns> | I've been submitting tests that both WebKit and IE fail, but should pass soon :) |
| 21:31 | jgraham | is generally more upset by crappy performance tests than crappy regression tests because they seem to have more PR value |
| 21:33 | <gsnedders> | But how quickly you can shift off an empty array is *totally* important! |
| 21:33 | <jgraham> | e.g. apparently all of Microsoft's canvas tests essentially tested blitting speed. But there were lots of them and they were pretty |
| 21:33 | <jgraham> | And sunspider is kind of a joke |
| 21:33 | <gsnedders> | (That's PeaceKeeper 2) |
| 21:34 | <jgraham> | Well yeah, Peacekeper is just very buggy |
| 21:34 | <gsnedders> | Also another favourite thing to benchmkar is timer resoution. |
| 21:34 | <gsnedders> | *benchmark |
| 21:34 | <jgraham> | 15:36 < jgraham> Well yeah, Peacekeper is just very buggy |
| 21:35 | <gsnedders> | I meant that more generally than PEaceKEpper. |
| 21:35 | <jgraham> | (it is doubly sad because good performance tests are really useful in targeting optimisation work) |
| 21:35 | <gsnedders> | *PeaceKeeper |
| 21:35 | gsnedders | really can't type decently right now |
| 21:35 | <gsnedders> | jgraham: Well, Kraken is fairly decent. But it's massively dominated by whether you impl a small number of things or not, CSE especially. |
| 21:37 | <gsnedders> | (Which, to be fair, is interesting. But if you care about perf it's something you can trivially do yourself.) |
| 21:38 | <gsnedders> | Testing things that can't be done at a source level are more interesting. |
| 21:38 | <jgraham> | Right, Kraken has actually made some attempt to target realistic workloads, which is nice |
| 21:46 | <timeless> | writing good tests is hard |
| 21:46 | <timeless> | plus you'll eventually lose |
| 21:46 | <timeless> | it's better to write and publish tests which let you look good in marketing :) |
| 21:47 | <timeless> | oh wait, i don't think i've ever had an employer that tried to do that |
| 21:47 | timeless | sighs |
| 21:50 | <roc> | gsnedders: actually in JS you can't CSE everything yourself, because there are a bunch of CSE-able runtime checks |
| 21:53 | <gsnedders> | roc: I'm assuming the author of the JS knows the types of arguments of the function and items higher up in the lexical scope are. |
| 21:54 | <gsnedders> | roc: Oh, you mean the runtime checks are CSE-able? |
| 21:54 | <gsnedders> | Oh, duh. |
| 21:54 | <roc> | yeah |
| 21:54 | <gsnedders> | I need to read what you write, and not read four things at once :) |
| 21:55 | <gsnedders> | roc: Depending on design, that may be done separately to CSE of code, though |
| 21:55 | <annevk> | IE has CORS now |
| 21:55 | <annevk> | sweet |
| 21:55 | <gsnedders> | annevk: Old news. |
| 21:55 | <annevk> | care |
| 21:59 | <annevk> | so I think Sam Ruby announced earlier today that the W3C HTML spec |
| 22:00 | <annevk> | of which the editor's draft was last updated November 4 |
| 22:00 | <annevk> | will not see updates until at least somewhere in January |
| 22:00 | <annevk> | way to make yourself irrelevant |
| 22:02 | <timeless> | wasn't it already irrelevant? |
| 22:08 | <annevk> | I wish there was a way to fix a typo in a tweet other than deleting it and posting a new one |
| 22:08 | <gsnedders> | +1 |
| 22:09 | hober | must have missed that announcement |
| 22:11 | <annevk> | it follows from when they plan to work on <time> again |
| 22:11 | <annevk> | which seems mid-January, maybe later |
| 22:11 | <timeless> | well, we're almost @december anyway |
| 22:11 | <timeless> | and that's a holiday somewhere |
| 22:11 | <annevk> | see e.g. http://lists.w3.org/Archives/Public/public-html/2011Nov/0229.html |
| 22:12 | <annevk> | more likely end of January / mid-February given the time they give for counter proposals and all |
| 22:17 | <gavinc> | Yeah, it's so horrible to take a little bit of time to let people read, think, and respond. I mean a whole four weeks will have gone by! (two of which will likely to be spent on holidays) |
| 22:19 | <_bga> | https://gitorious.org/qmlweb/pages/Home |
| 22:19 | <annevk> | gavinc: the problem is giving people time to comment, that's great |
| 22:20 | <annevk> | gavinc: the problem is having an unmaintained forked draft |
| 22:20 | <annevk> | I'm missing a "not" in that first sentence :) |
| 22:21 | <karlcow> | annevk: I'm pretty the Web will survive :) |
| 22:21 | <timeless> | annevk: unmaintained drafts, oh my! |
| 22:21 | <timeless> | wait, doesn't that describe 10-15 years of w3 docs? |
| 22:21 | <karlcow> | I have been watching for 20 years its soap-opera. It will survive for a few weeks, months even |
| 22:22 | <karlcow> | hurricane in teapots |
| 22:22 | <gavinc> | So the W3C is making itself irrelevant by having forked the WHATWG HTML document in order to revert an element that that WHATWG added back after taking a few days to consider having removed it. |
| 22:23 | <annevk> | karlcow: there's always things that will outlast other things |
| 22:23 | <jgraham> | gavinc: That seems to be an inaccurate summary of the situation, yes |
| 22:24 | <annevk> | karlcow: you are probably right that this does not matter much |
| 22:25 | <annevk> | karlcow: at least we are past the point where if I said "specification" on my weblog I would get a lengthy public personal letter from someone |
| 22:25 | <gavinc> | annevk: but a good chance to bash the W3C can't go wasted |
| 22:25 | <gsnedders> | TabAtkins: Is it a goal of selectors to be able to do something like //*[substring(., 0, 4) == "this"] or //comment()? |
| 22:25 | <annevk> | gavinc: I did not bash the W3C |
| 22:26 | <annevk> | gavinc: I am somewhat upset by the not caring of the HTML WG co-chairs |
| 22:26 | <annevk> | gavinc: and them not giving any indication they are doing something here |
| 22:26 | <othermaciej> | we do care |
| 22:26 | <annevk> | gavinc: or communicating about the problem |
| 22:26 | <othermaciej> | and will do something |
| 22:26 | <annevk> | hey othermaciej :) sweet |
| 22:26 | <othermaciej> | though I'm honestly unable to predict what |
| 22:26 | <annevk> | heh |
| 22:28 | <TabAtkins> | gsnedders: The first selects all elements with names beginning with "this", right? |
| 22:29 | <jgraham> | gsnedders: I don't think anyone is seriously arguing that selectors will ever be as powerful as XPath |
| 22:29 | <TabAtkins> | gsnedders: Almost certainly not. JS can do the former just as easily, and comment nodes are mostly irrelevant. |
| 22:29 | <jgraham> | The only discussion is whether it is worth adding API for that extra power |
| 22:29 | <TabAtkins> | jgraham: I seriously argue that they're basically identical in power, especially when combined with JS and NodeArray or whatever. |
| 22:30 | <jgraham> | TabAtkins: In that case I claim that DOM is just as powerful and selectos are unnecessary |
| 22:30 | <annevk> | fwiw, unlike Selectors, XPath can select text nodes, attribute nodes, and things like that |
| 22:31 | <gavinc> | fragments of text nodes too |
| 22:31 | <jgraham> | Right, XPath is clearly theoretically better than selectors. |
| 22:31 | <annevk> | if "better" means it can do more |
| 22:32 | <gsnedders> | TabAtkins: No, names whose textContent begins with this |
| 22:32 | <zewt> | more widely-known, i'd expect, which is a plus in and of itself |
| 22:32 | <annevk> | but it would be interesting to see how often XPath is used today and how much convenience APIs have been created for it |
| 22:32 | <gsnedders> | TabAtkins: That argument applies to selectors too. |
| 22:32 | <gsnedders> | TabAtkins: (that it can just be done in JS) |
| 22:32 | <zewt> | i've used it, though mostly just for local greasemonkey scripts; never really had a need for pages i've authored myself |
| 22:33 | <zewt> | and for xml stuff, of course, though not for a long time |
| 22:33 | <gavinc> | TabAtkins: In fact plenty of sites still use @style and JS to do styling |
| 22:33 | <gsnedders> | JS is Turing complete, so you can do, any processing you want of data you have access to. |
| 22:33 | <zewt> | (actually, not that long, now that I think about it) |
| 22:33 | <jgraham> | "Better" means something like "has been designed with a small number of ideas that work with the whole DOM data model" |
| 22:33 | <zewt> | (i know nothing at all about selectors) |
| 22:34 | <gsnedders> | Anolis certainly used to use XPath for things that don't work with CSS Selectors. |
| 22:34 | <jgraham> | Whereas selectors has celarly just been made up as people went along to solve CSS use cases |
| 22:34 | <jgraham> | *clearly |
| 22:34 | <zewt> | (is that just the name for css matching?) |
| 22:34 | <gsnedders> | (It didn't use CSS Selectors at all, because in lxml they're just compiled to XPath, so there's little gain, just cost) |
| 22:34 | <zewt> | (i abuse parentheses) |
| 22:34 | <jgraham> | zewt: Yeah |
| 22:34 | <zewt> | gotcha |
| 22:34 | <jgraham> | (you do) |
| 22:34 | <zewt> | heh |
| 22:35 | <gavinc> | systems for crawling the public web use xpath a fair amount. Not a browser UA, but still a UA. |
| 22:35 | <timeless> | that reminds me, i should pick up one of the other specs and review it |
| 22:35 | timeless | gets distracted by chasing credit cards and bills |
| 22:35 | <zewt> | yeah, i use it for finding random nodes for gm scripts pretty often (but that's not a use case for a web api) |
| 22:36 | <annevk> | gavinc: but not really relevant for developing browser-based APIs |
| 22:36 | <timeless> | anyone here ever ask United Airlines for a duplicate of a receipt for an extra baggage fee? |
| 22:36 | <zewt> | timeless: i might ask such a thing if i want to see a blank, bewildered stare |
| 22:36 | <jgraham> | I use XPath all the time with html5lib + lxml but again tht's not quite the web use case |
| 22:36 | <timeless> | zewt: it's blocking a >1000$ reimbursement |
| 22:37 | <timeless> | the fee was 25% |
| 22:37 | <timeless> | s/%/$/ |
| 22:38 | gsnedders | thinks unless browsers are going to drop XSLT and the existing DOM Level n XPath impls, we may as well have a clean API for it |
| 22:38 | <jgraham> | I tend to agree |
| 22:38 | <jgraham> | Which never happens :) |
| 22:38 | <jgraham> | Although I don't include XSLT in my argument |
| 22:39 | <jgraham> | Because, ugh |
| 22:39 | <gsnedders> | jgraham: Well, we can't drop XPath and not XSLT. |
| 22:39 | <zewt> | i tend to find css selectors for the common cases more readable; but that's probably because when i use xpath, it's generally for more complex cases anyway |
| 22:39 | <jgraham> | zewt: I think it is generally agreed that for simple cases selectors are more readable |
| 22:40 | <jgraham> | foo.bar is better than //foo[@class='bar'] |
| 22:40 | <jgraham> | (and that isn't even right) |
| 22:40 | <zewt> | also, i've never tried using css selectors for GM scripts; i probably should |
| 22:41 | <TabAtkins> | gsnedders: Ah, in that case, maybe. Selectors based on text content have been rejected as too slow for CSS, but if we do a "batch processors profile" of Selectors, it's perfectly reasonable. |
| 22:41 | <zewt> | yeah, i have to check references a lot with xpath, which IMO implies a poor design (also because I don't use it often enough, but even so) |
| 22:42 | <gavinc> | can selectors select positionally? //foo[42] I know I never have but not sure I haven't |
| 22:42 | <TabAtkins> | gsnedders: I don't think "You dont' need Selectors at all if you have JS" is a reasonable argument. |
| 22:42 | <TabAtkins> | gavinc: Yes, but in a different way than XPath does. |
| 22:42 | <gavinc> | TabAtkins: nth child stuff? yeah okay |
| 22:42 | <TabAtkins> | gavinc: Selectors can select based on index among siblings. XPath selects based on document-order index in the current context nodeset. |
| 22:43 | <gavinc> | TabAtkins: Which is often the parent and thus often the same |
| 22:43 | <gavinc> | TabAtkins: Yeah |
| 22:43 | <TabAtkins> | gavinc: However, with JS involved (which is the context we care about here), it's just "document.find('foo')[42]" |
| 22:43 | <annevk> | and Selectors ignores non-elements |
| 22:43 | <TabAtkins> | Or rather, findAll. |
| 22:44 | <jgraham> | (kind of sucks if you don't want to create and destroy a big array of elements though) |
| 22:45 | <jgraham> | document.selectSingleNode("//foo[42]") avoids that |
| 22:45 | <TabAtkins> | A little bit, sure. I don't know if it's a big deal. The fact that map() and filter() are separate functions also sometimes means you're creating intermediaries. |
| 22:45 | <TabAtkins> | But that's good enough for most functional languages. |
| 22:45 | <zewt> | python has spoiled me against the grossness of map and filter functions |
| 22:46 | <TabAtkins> | I'll admit, list comprehensions are awesome. |
| 22:46 | <gavinc> | Personally I know I would have used XPath in the browser more with a better API. But then the back end was often XML |
| 22:46 | <TabAtkins> | And I'm excited to get them in JS. |
| 22:46 | <zewt> | (list comprehensions and generator expressions) |
| 22:46 | <jgraham> | Anyway, I am quite skeptical that the line in the sand between "what I should have to write js for" and "what I can use a DSL for" is "what selectors can do" |
| 22:46 | <gavinc> | Didn't use CSS selectors much in JS till JQuery either |
| 22:46 | <timeless> | and i'm expecting them to ask me for 25$ for the receipt |
| 22:47 | <jgraham> | timeless: Sucks to be you... |
| 22:47 | <TabAtkins> | jgraham: Sure. But I don't think the answer for "the line is misdrawn" is "support a new language with a different syntax but 95% the same functionality" |
| 22:47 | <karlcow> | is there an implementation of selectors in python. |
| 22:47 | <jgraham> | TabAtkins: It's not "support a new language" though |
| 22:47 | karlcow | is using a lot XPath in my scripts |
| 22:47 | <jgraham> | karlcow: Yeah lxml has one |
| 22:47 | <gavinc> | karlcow: It compiles XPaths ;) |
| 22:47 | <jgraham> | But just using xpath is easier |
| 22:47 | <TabAtkins> | gavinc: Yeah, jQuery made Selectors-in-JS popular, paving the way for querySelector and now find. |
| 22:47 | <paul_irish> | hey #whatwg.. friends and i are working on a site to launch tomorrow that's all about getting more developer/author involvement in standards & browser development... http://h5bp.github.com/igotmybeanie/ would love any feedback you have. areas where the content is a little WTF or could use more detail.. |
| 22:47 | <karlcow> | ahah |
| 22:48 | <gavinc> | TabAtkins: I do remember however wishing that jQuery had also implemented XPath selectors :\ |
| 22:48 | <jgraham> | Apparenty it did at one point |
| 22:48 | <TabAtkins> | jgraham: XPath is a new language for most devs. If qSA had never existed, it would still probably be a bad idea to have XPath selectors in JS and Selectors selectors in CSS, again because of the 95% overlap. |
| 22:48 | <TabAtkins> | gavinc: They did. Then they removed it because nobody used it. |
| 22:49 | <gavinc> | MMm, no, it didn't do math, nor work with context |
| 22:49 | <gavinc> | Looked like XPath, wasn't. At least that's my memory |
| 22:49 | <TabAtkins> | gavinc: Oh, ok, yeah. It was a subset. |
| 22:49 | <jgraham> | TabAtkins: Everything is new to people the first time. |
| 22:49 | <jgraham> | And the overlap is far from 95% |
| 22:50 | <TabAtkins> | jgraham: Your first statement is true. Your second statement is arguable. |
| 22:50 | <TabAtkins> | The vast majority of Selectors usage is trivial tagname/class/id selection, perhaps with a combinator involved, and maybe some :hover. |
| 22:50 | jgraham | finds that after you get away from the simple things XPath is relatively OK because the model is clean whereas CSS is a mess |
| 22:50 | <gavinc> | jgraham: +1 |
| 22:50 | <jgraham> | Right, if people only do trivial things selectors are better |
| 22:50 | <TabAtkins> | Weighted by usage on the web, the percentage is probably greater than 95%. |
| 22:51 | <karlcow> | "<TabAtkins> jgraham: XPath is a new language for most devs." HIHI. houhou ahaha. :D I'm going to the toyshop to find webtools for the 5 years old :) |
| 22:51 | <jgraham> | But it is not clear if people only do trivial things because that's all that is easy with selectors |
| 22:51 | <TabAtkins> | jgraham: The models are isomorphic. |
| 22:51 | <annevk> | paul_irish: WHATWG blog also covers W3C WebApps to some extent |
| 22:51 | <gavinc> | TabAtkins: Weighted by usage on the web there was no need for img ;) metrics will only get you so far |
| 22:52 | <annevk> | paul_irish: and other random stuff that interests me |
| 22:52 | <jgraham> | TabAtkins: But CSS Selectos looks like line noise. That's a difference |
| 22:52 | <TabAtkins> | jgraham: XPath lets you save on functions by reusing the same things for all string handling, for example, and just letting you specify the source of the string. |
| 22:52 | <paul_irish> | annevk: truth. ill refine that wording |
| 22:52 | <TabAtkins> | jgraham: Um, and XPath selectors aren't line noise? |
| 22:52 | <jgraham> | Well I don't have to rememebr the difference between : and :: or + and ~ |
| 22:52 | <zewt> | not really |
| 22:52 | <paul_irish> | karlcow: i linked to your weekly summaries as well. :D |
| 22:52 | <annevk> | paul_irish: also may want to add www-dom⊙wo to the mailing lists |
| 22:53 | <paul_irish> | k |
| 22:53 | <TabAtkins> | Shrug, okay. I don't see much difference in line-noise-ness, particularly in the abbreviated syntax. |
| 22:53 | <miketaylr> | paul_irish: you can add a link to ODIN for people to keep up to speed with Opera updates, ;) http://my.opera.com/ODIN/blog/ |
| 22:53 | <TabAtkins> | Both require a sophisticated mental parser to understand non-trivial examples. |
| 22:53 | <karlcow> | paul_irish: damned! :) more pressure. I should finish write one today or tomorrow morning. It is almost done. |
| 22:53 | <miketaylr> | in (How do I keep up with what’s landing in browsers?) |
| 22:54 | <annevk> | paul_irish: "See the latest commits to the HTML and CSS specs by following their respective working groups on Twitter:" has the twitter accounts in the wrong order, either fix that or drop respective |
| 22:54 | <paul_irish> | kk |
| 22:55 | <annevk> | paul_irish: for advanced, http://wiki.whatwg.org/wiki/Bugzilla_conventions might also be of help for those diving into existing bugs |
| 22:56 | <annevk> | paul_irish: nice initiative |
| 22:56 | <paul_irish> | thanks. excited about it :) |
| 22:56 | <paul_irish> | conventions page is excellent |
| 22:57 | <miketaylr> | :hover on the blue buttons could use some more contrast for my old eyes |
| 22:57 | <divya> | agreed |
| 22:57 | <divya> | fixing. |
| 22:57 | <karlcow> | miketaylr: me too :) oldies |
| 22:57 | <karlcow> | :p |
| 22:57 | <jgraham> | paul_irish: Woudl be nice if the browser updates thing was more cross browser. Someone mentioned the Opera ODIN. Maybe the IE blog and some Mozilla blog (about:mozilla?) |
| 22:57 | <gavinc> | sigh, totally non spec related question: ... are there any non evil registrars these days? |
| 22:58 | <karlcow> | gavinc: using gandi.net myself |
| 22:58 | <paul_irish> | jgraham: Yeah I'm gonna end up making a better post that summarizes browser update resources, but it's gonna land after this ships (tomorrow morning). |
| 22:58 | <paul_irish> | but i'll add in the browser outlets explicitly for now |
| 22:59 | <jgraham> | paul_irish: Great |
| 22:59 | <annevk> | gavinc: https://www.transip.nl/ is pretty awesome |
| 22:59 | <annevk> | gavinc: not sure if it works outside the Netherlands though, don't know anyone who tried |
| 23:00 | <annevk> | oh, and it might only be in Dutch :) |
| 23:00 | gavinc | can read a bit of dutch ... but not sure about co-workers |
| 23:00 | jgraham | also used gandi.net |
| 23:01 | <jgraham> | I didn't notice any evilness so far |
| 23:02 | <paul_irish> | thx everyone |
| 23:02 | <jgraham> | paul_irish: WebApps (and CSS I guess) also have testsuites that would benefit from contributers |
| 23:03 | <paul_irish> | jgraham: have any links handy? |
| 23:03 | <annevk> | http://lists.w3.org/Archives/Public/public-webapps-testsuite/ |
| 23:03 | <annevk> | http://lists.w3.org/Archives/Public/public-css-testsuite/ |
| 23:03 | <annevk> | are entry points of some kind |
| 23:04 | <jgraham> | http://www.w3.org/2008/webapps/wiki/Testing |
| 23:04 | <jgraham> | http://wiki.csswg.org/test |
| 23:15 | <annevk> | holy shit |
| 23:15 | <annevk> | "IE10 Preview 4 introduces an updated quirks mode that is more consistent and interoperable with the way quirks modes works in other browsers like Firefox, Chrome, Safari, and Opera." |
| 23:16 | <annevk> | not sure what it means exactly, but that might be bigger than CORS |
| 23:18 | <gsnedders> | We needs tests for quirks mode, obviously. |
| 23:21 | <roc> | how do you activate it? |
| 23:21 | jgraham | expects it involves goat entrails under a full moon |
| 23:21 | <franksalim> | will IE10 have both quirks modes? |
| 23:22 | <timeless> | typically ie would automatically decide to use quirks mode |
| 23:23 | <timeless> | or you could use F12 to force it |
| 23:26 | timeless | downloads the new ie10pp |
| 23:28 | <timeless> | Debug> Force {ie5 quirks, ie7 doc, ie8 doc,ie9 doc, standards, quirks} |
| 23:28 | <timeless> | so currently it has 2 modes labeled as quirks |
| 23:28 | <timeless> | plus 3 doc modes and a standards mode |
| 23:28 | <franksalim> | paul_irish: this might mean you could revise your estimate of 72 IE flavors upwards |
| 23:28 | <franksalim> | timeless: wow |
| 23:28 | <timeless> | and yes, you can change the mode in f12 |
| 23:28 | <timeless> | if you haven't used f12, you need to get it and try |
| 23:28 | timeless | goes home |
| 23:29 | <franksalim> | so how do you specify which quirks mode you want? is there a quirks mode doctype /s |
| 23:29 | <timeless> | presumably it tries to match the rules gecko and co use |
| 23:30 | <annevk> | whoa, complex |
| 23:30 | <annevk> | must suck to do QA on that |
| 23:30 | <miketaylr> | DOCTYPEs are for suckers |
| 23:33 | <annevk> | heard it here first |
| 23:48 | <roc> | our rules are quite simple |
| 23:48 | <roc> | the question is, what mode do they use for a document without a DOCTYPE? |
| 23:48 | <roc> | I find it hard to believe they'll switch such documents from "IE6" mode to something else |
| 23:51 | <jamesr_> | time to update the IE mode flowchart? |