| 01:18 | <Hixie> | AryehGregor: you tell me: http://www.whatwg.org/issues/data.html |
| 02:27 | <johnnyk> | I have a question about HTML 5 <video> - For some reason I can play videos by direct URL, but they do not play when embedded in the <video> tag. I'm using Fx 3.5... |
| 02:28 | <johnnyk> | It's sort of strange, I have never come across anything exactly like this before... |
| 02:28 | <inimino> | what does "embedded in the <video> tag" mean specifically? |
| 02:28 | <johnnyk> | <video src="URL"> |
| 02:29 | <johnnyk> | I know that src is referencing the right file, because if I (in Firefox) right click the video element and click "View Video", the video is downloaded just fine... |
| 02:29 | <mpilgrim> | reading the backscroll from the weekend |
| 02:29 | <mpilgrim> | sorry i missed the lively chat about content sniffing |
| 02:30 | <doublec> | johnnyk, is the video an Ogg Theora video? Is it being served with the correct mime type? |
| 02:31 | <johnnyk> | doublec - It is ogg Theora, with ogv extension. How can I know if it is being served with the correct mime type? |
| 02:31 | <doublec> | look at the HTTP headers |
| 02:31 | <doublec> | maybe try: wget -S url.ogv |
| 02:31 | <johnnyk> | cool, I'll try... |
| 02:31 | <doublec> | and look for the Content-Type header |
| 02:32 | <mpilgrim> | is there a reference implementation of the proposed content sniffing algorithm? |
| 02:32 | <johnnyk> | Content-Type: text/plain :-( |
| 02:32 | <johnnyk> | But I guess that's my answer... |
| 02:32 | <doublec> | yep |
| 02:32 | <doublec> | do you control the server? |
| 02:33 | <johnnyk> | Why is it being served up like that? I've had this happen on multiple hosts, right now on XAMPP |
| 02:33 | <johnnyk> | I'm using XAMPP right now, so yes |
| 02:33 | <doublec> | what is xampp? |
| 02:33 | <johnnyk> | Just a little program to run a server on localhost |
| 02:33 | <johnnyk> | For testing purposes, etc |
| 02:34 | <johnnyk> | I guess it's not limited to that, but yeah |
| 02:36 | <miketaylr> | so Modernizr is telling me that the latest webkit supports type=color, which is cool. |
| 02:37 | <miketaylr> | anyone know if its going to get any native chrome/ui? |
| 02:37 | <aspireonescs> | thats the mac version of WAMP right johnnyk? |
| 02:38 | <tkent> | miketaylr: it's not have any fancy UI. Just a text field with type validation. |
| 02:38 | <johnnyk> | Nope, Linux. Thanks for the help by the way |
| 02:38 | <miketaylr> | tkent: ok, thanks. |
| 02:38 | <johnnyk> | Actually |
| 02:38 | <aspireonescs> | ah |
| 02:38 | <miketaylr> | i s'pose i can make my own fancy ui. ;) |
| 02:38 | <johnnyk> | I see what you mean now. Yes. XAMPP is the generic name. When installed on Linux it is called LAMPP, Windows is WAMPP, etc |
| 02:39 | <aspireonescs> | ah ya i use WAMP |
| 02:39 | <jt6195> | me too |
| 02:39 | <johnnyk> | Yup yup, pretty much the same thing. |
| 02:39 | <doublec> | johnnyk, you should be able to configure apache to serve the correct mime type |
| 02:39 | <doublec> | http://en.flossmanuals.net/TheoraCookbook/MyHosting |
| 02:39 | <doublec> | explains how |
| 02:39 | <aspireonescs> | ya it makes installing/configing a web server esasy |
| 02:40 | <johnnyk> | cool stuff. Thanks alot for all the help doublec and aspireonescs |
| 02:40 | <doublec> | np |
| 02:43 | <johnnyk> | doublec - Worked like a charm. Sorry if this isn't typically a venue for questions, which is becoming apparent by the other conversaitons going on. :-) But anyway, thanks again for the tip! |
| 02:43 | <doublec> | no worries, glad it helped |
| 02:44 | <mpilgrim> | i searched for 'mime sniff barth python' and came up empty |
| 02:44 | <mpilgrim> | or rather, i came up with a link to my own writing in the top 3 results |
| 02:44 | <mpilgrim> | which is never a good sign when you're looking for something you know you haven't written yet |
| 02:45 | <roc> | every so often I search for answers to some technical question and find my own posts about it in the top hits |
| 02:45 | <roc> | that is always depressing |
| 02:48 | <mpilgrim> | i would also love to know if there are any test cases built around http://tools.ietf.org/html/draft-abarth-mime-sniff-01 |
| 02:48 | <mpilgrim> | or if i need to write those too |
| 06:37 | <heycam> | hsivonen, ping? |
| 06:38 | <hsivonen> | heycam: pong |
| 06:38 | <heycam> | hi. i was just wondering if you have an online version of the validator that does the foreign content stuff. |
| 06:39 | <hsivonen> | heycam: in text/html? The parser does but the HTML5 schema doesn't yet |
| 06:39 | <heycam> | hsivonen, yeah |
| 06:39 | <heycam> | oh ok |
| 06:39 | <heycam> | so i was just trying an example with the live dom viewer, and it gave some unexpected behaviour |
| 06:40 | <heycam> | wanted to check if it was valid |
| 06:40 | <hsivonen> | what was the unexpected behavior? |
| 06:40 | <heycam> | i'll get a link |
| 06:41 | <heycam> | http://livedom.validator.nu/?%3C!DOCTYPE%20html%3E%0A%3Ctable%3E%0A%20%20%3Ctr%3E%0A%20%20%20%20%3Ctd%3E%0A%20%20%20%20%20%20%3Csvg%3E%0A%20%20%20%20%20%20%20%20%3CforeignObject%3E%0A%20%20%20%20%20%20%20%20%20%20%3Ctable%3E%0A%20%20%20%20%20%20%20%20%20%20%20%20%3Ctd%3E%0A%20%20%20%20%20%20%20%20%20%20%3C%2Ftable%3E%0A%20%20%20%20%20%20%20%20%3C%2FforeignObject%3E%0A%20%20%20%20%20%20%3C%2Fsvg%3E%0A%20%20%20%20%3Ctd%3E%0A%20%20%20%20%20%20Hi%0A%20%20%3C%2Ft |
| 06:41 | <heycam> | r%3E%0A%3C%2Ftable%3E%0Athere |
| 06:41 | <heycam> | apologies for the bonus whitespace |
| 06:41 | <heycam> | actually that might've got cut off |
| 06:41 | <heycam> | http://bit.ly/HtImi |
| 06:42 | <heycam> | i was surprised that the third <td> start tag didn't create a third <td> element |
| 06:43 | <heycam> | i wondered how the parser was handling nested foreign/html/foreign/etc. content with respect to the secondary insertion mode |
| 06:43 | <heycam> | since it seems to be just a single variable, and not a stack |
| 06:43 | <heycam> | do you know if that fragment should be valid? |
| 06:43 | <hsivonen> | I'm surprised "Hi" didn't get foster-parented above the table |
| 06:43 | <heycam> | why are you not surprised the third <td> didn't create a second <td> in the outer table? |
| 06:44 | <hsivonen> | heycam: actually, I whould be surprised about that first. |
| 06:44 | <hsivonen> | looks like a bug |
| 06:44 | <heycam> | :) |
| 06:44 | <hsivonen> | heycam: thanks |
| 06:44 | <heycam> | so i don't know if the secondary insertion mode needs to become a stack of insertion modes |
| 06:45 | <heycam> | or if it needs to "reset the insertion mode appropriately" or whatever it is where it guesses the insertion mode based on the current state of the open tag stack |
| 06:45 | <mpilgrim> | hixie: yt? |
| 06:45 | <mpilgrim> | i'm trying to implement http://tools.ietf.org/html/draft-abarth-mime-sniff-01 |
| 06:45 | <hsivonen> | Hixie: ^ See what heycam says |
| 06:46 | <mpilgrim> | section 4 "text or binary" algorithm, step 5, says: |
| 06:46 | <mpilgrim> | "If the first bytes of the resource match one of the byte sequences in the "pattern" column of the table in the unknown type section below, ignoring any rows whose cell in the "security" column says "scriptable" (or "n/a"), then the /sniffed type/ of the resource is the type given in the corresponding cell in the "sniffed type" column on that row; abort these steps." |
| 06:46 | <mpilgrim> | this line is incredibly unclear (to me) |
| 06:47 | <heycam> | hsivonen, i'm wondering if this problem also manifests itself when the insertion mode just before entering foreign content is something else special |
| 06:47 | <heycam> | as in, things that aren't just "in body" and do funny parsing things |
| 06:47 | <mpilgrim> | first of all, do i ignore or not-ignore rows whose "security" column is "n/a"? |
| 06:48 | <mpilgrim> | second of all, do i apply the mask given in the first column? |
| 06:48 | <mpilgrim> | hmm |
| 06:48 | <mpilgrim> | maybe it doesn't matter, because i think all the "safe" patterns have masks consisting entirely of FF bytes |
| 06:48 | <hsivonen> | heycam: very likely |
| 06:49 | <mpilgrim> | and thus irrelevant whether i apply the mask or not |
| 06:49 | <mpilgrim> | but still, quite a confusing sentence, and subject to potentially strange bugs if a "safe" pattern is ever added with a non-FF mask |
| 06:50 | <heycam> | hsivonen, we'll be discussing svg in html stuff at our f2f tomorrow, so i'm just preparing a summary |
| 06:50 | <heycam> | i'll file a bug about this unexpected parser behaviour after tomorrow if it's not fixed by then :) |
| 06:50 | <heycam> | (that kinda sounds like a threat, though it's not meant to be :)) |
| 06:50 | <mpilgrim> | it would be clearer to simply duplicate the relevant information (a list of "safe" patterns and their corresponding sniffed types) in section 4 |
| 07:08 | <hsivonen> | heycam: it's possible that it's my bug and not Hixie's, because I implement "in foreign" as a flag. When it is set, the usual mode variable is treated as the "secondary" mode. |
| 07:08 | <hsivonen> | I suspect it's Hixie's bug, though |
| 07:08 | <heycam> | hsivonen, i see. i think that's equivalent to what's in the spec, yeah. |
| 07:26 | <drunknbass> | anyone use github? |
| 07:52 | <Hixie> | regarding the variety of earlier comments, no idea off hand |
| 07:53 | <Hixie> | file a bug |
| 07:53 | <Hixie> | send e-mail |
| 07:53 | <Hixie> | i'll get to it when i'm awake sometime |
| 07:54 | <boblet> | I wish the note on <section> use (“not a generic container element”) had an id so I could link to it directly |
| 07:55 | <boblet> | @adactio was right re: <section>/<article> confusion in the trenches |
| 07:56 | <Hixie> | file a bug and i'll add an id when i next go through the bugs |
| 08:17 | <boblet> | Hixie: thanks. would that be enhancement or trivial? |
| 08:17 | <Hixie> | doesn't matter i ignore everything but the comments |
| 08:20 | <boblet> | k |
| 08:36 | <jgraham> | FWIW html5lib gives the same parsetree as the validator.nu parser |
| 08:37 | jgraham | always wants to call it parser.nu |
| 09:26 | Philip` | thinks we need to replace URLs with BOBLs, as the address of a bag of bits |
| 09:28 | jgraham | could subscribe to that if we can bacronym the full word BOBBLE somehow |
| 09:29 | <hsivonen> | I hate it how Paypal poisons Firefox's password autofill |
| 09:30 | <hsivonen> | surely Firefox should be better at telling if I've hit a phishing site than I am |
| 09:30 | <hsivonen> | in the world of IDNA spoofs |
| 09:30 | <hsivonen> | seems like a very bad idea to let me figure out if a merchant sent me to the real Paypal or to a fake Paypal |
| 09:34 | <Philip`> | jgraham: The idea is to pronounce it that way, without needing to make all the letters explicit |
| 09:34 | <Philip`> | like GRDDL |
| 09:35 | <Philip`> | or, um, any other example |
| 09:35 | <jgraham> | Philip`: Yeah, but I really want it to have all the letters |
| 09:35 | Philip` | can't think of any :-( |
| 09:36 | <jgraham> | I guess terribe, tortured acronyms (with accurate spelling) can't be the sole preserve of the astrophyics community |
| 09:37 | hsivonen | was ignorant enough to learn the etymology of "quasar" only this past weekend |
| 09:38 | <Philip`> | I suppose it could be a Bag Of Bits Binary Locator Endpoint |
| 09:38 | hsivonen | notes that TED talks given by British astrophysicists show a different attitude than the talks by the usual TED gurus |
| 09:41 | <Philip`> | The world would be a more interesting place if physics projects were given funding proportional to the Scrabble score of their acronym (with the requirement that it's a real word, and expands to an acceptable phrase) |
| 09:49 | <jgraham> | hsivonen: Who was talking? (and how was the attitude different?) |
| 09:52 | <hsivonen> | jgraham: David Deutsch and Martin Rees |
| 09:53 | jgraham | guesses Martin Rees would be one of them |
| 09:54 | <hsivonen> | jgraham: they seemed to be pretty convinced about the accuracy of very macro (universe scale) explanations while others tend to find doubt in conventional micro (earth-level) explanations |
| 09:54 | <hsivonen> | jgraham: also, the reaction to global warming wasn't let's go prevent it but it was probably too late to prevent it before anyone noticed it, let's go adapt to it |
| 09:55 | <hsivonen> | and they seem to think on different timescales than the others |
| 09:57 | <jgraham> | The timescales thing is easy to believe when you work in a field where one end of the important timescales range is 10^10 years |
| 09:57 | <jgraham> | s/when you/since they/ |
| 10:01 | <jgraham> | Also, it is quite possible to believe in the macro-level universe stuff being right without the earth-scale stuff being right because the universe is, in some sense, simpler than the sum of its parts |
| 10:02 | <Philip`> | Global warming has had a huge and immediate effect - just look at a graph of temperature in the region of space where the Earth is, which started at approximately 3K and then suddenly shot up to over 290K |
| 10:04 | <hsivonen> | jgraham: I'm not suggesting it's illogical. Just that their attitude feels different compared to the others. |
| 10:04 | <jgraham> | hsivonen: I didn't mean to imply that you were |
| 10:04 | <othermaciej> | physics has a track record of making some impressive quantitative predictions |
| 10:05 | <othermaciej> | but a lot of complicated interactions can't really be worked out from physics first principles, either due to computational intractability, or chaotic behavior |
| 10:05 | <Philip`> | It hasn't predicted anything more than a few hundred years in the future, though |
| 10:05 | <Philip`> | s/predicted/successfully and confirmedly predicted/ |
| 10:06 | <othermaciej> | sure it has, people do experiments every day that confirm predictions from theories over 100 years old |
| 10:06 | <jgraham> | othermaciej: Sure, which is part of why the universe is simpler that the sum of its parts. Because it turns out that at sufficently large scales a lot more problems are somewhat tractable |
| 10:06 | <Philip`> | By "a few hundred", I was only being accurate to an order of magnitude |
| 10:06 | <othermaciej> | jgraham: yeah, it's easier to compute orbits than to compute the full chemistry of a living system |
| 10:07 | <jgraham> | Philip`: It has also successfully predicted things more than a few hundred years in the past |
| 10:07 | <Philip`> | Predicting the past is easy :-p |
| 10:07 | <jgraham> | Philip`: Not if you weren't there when it happened :) |
| 10:07 | <Philip`> | I suppose it's more like postdiction |
| 10:08 | <jgraham> | (it is just as hard to predict the position of the plantes in the solar system a million years in the past as a million years in the future) |
| 10:08 | <othermaciej> | Newtonian mechanics made a lot of predictions that are accurate today to a remarkable degree, fare more so than any quantitative predictions made by economics |
| 10:09 | <hsivonen> | maybe physicists have less of an agenda than, say, the Chicago School |
| 10:09 | <othermaciej> | (not to pick on economics - I like it as a field - but it's not very good at predicting anything at more than a very general qualitative level) |
| 10:09 | <Philip`> | I still think we should wait a few billion years before judging whether physics successfully models our world |
| 10:10 | <Philip`> | just to be extra certain |
| 10:10 | <othermaciej> | economics does tend to get tied into politics, and sometimes that leads to theories with predictions that aren't even qualitatively right, e.g. Marxism |
| 10:11 | <othermaciej> | even so, economics is a lot more successful at telling us useful information about society than other social sciences |
| 10:11 | <jgraham> | Economics has the (recognised) problem that the common fundamental assumptions in quantitative economics are basically wrong |
| 10:12 | <othermaciej> | I'm pretty sure all physicists think that current physics does *not* completely and correctly model our world |
| 10:12 | <jgraham> | All the stuff about ideal markets is a dangerously poor approximation to the real world |
| 10:13 | <othermaciej> | not all economic models assume rational expecations |
| 10:14 | <jgraham> | Sure, like I said it's a known problem that people are working on |
| 10:14 | <othermaciej> | that said, microeconomics is much more successful than macro |
| 10:15 | <othermaciej> | despite arguably buying more thoroughly into rational expectations and other such assumptions |
| 10:34 | gsnedders | guesses he ought to start thinking about what he's going to take to Sweden |
| 10:35 | <jgraham> | gsnedders: Bibbi may be about to phone you |
| 10:35 | <gsnedders> | jgraham: "may" |
| 10:35 | <jgraham> | So, er, you might want to check I gave her the right number |
| 10:35 | <gsnedders> | jgraham: My mob. number? |
| 10:35 | gsnedders | looks up his own number |
| 10:36 | <jgraham> | Well she just asked for your phone number. I guess she is either going to phone you or leave your calling card in phone boxes arounf town |
| 10:37 | gsnedders | also notes she has his phone number somewhere anyway |
| 10:49 | gsnedders | blames BenMillard and Hixie for him sometimes signing emails "g" nowadays |
| 13:44 | <hsivonen> | sad. wikipedia put the space back |
| 13:47 | <jgraham> | Heh. I like how "consistent with the specification" lost to "inconsistent with other specifications" |
| 13:52 | <Philip`> | That seems stupid |
| 13:52 | <Philip`> | They ought to have used a *non-breaking* space |
| 13:53 | <Philip`> | otherwise it will go all ugly when it wraps |
| 14:09 | <Lachy> | even though they put it back in the title, the first sentence in the article begins with "HTML5 is the next major revision...", with no space in it |
| 14:10 | hsivonen | assumes that admins who perform moves don't also rewrite articles to match |
| 14:10 | <Lachy> | anyway, I'm not too concerned. It really shouldn't matter whether it's spelled with or without the space |
| 14:12 | <Philip`> | The http://html5.digitalbazaar.com/specs/html5-epb.html definition of profile seems to conflict with http://www.w3.org/TR/2009/PER-xhtml11-20090507/conformance.html |
| 14:12 | <Philip`> | and with content that uses version="-//W3C//DTD XHTML 1.1//EN" as suggested by that page |
| 14:13 | <Philip`> | (such content making up 29 of ~425K page from dotnetdotcom) |
| 14:14 | <Philip`> | (vs 30 with "XHTML+RDFa 1.0") |
| 14:14 | <Philip`> | (and a handful of others) |
| 14:14 | <Philip`> | s/profile/version/ |
| 14:22 | <mpilgrim> | woohoo, i found a genuine bug in http://tools.ietf.org/html/draft-abarth-mime-sniff-01 |
| 14:23 | <annevk2> | mpilgrim, btw, I think ref implementations might be Gecko/WebKit |
| 14:23 | <annevk2> | abarth patches both |
| 14:23 | <annevk2> | (maybe they have the same bug) |
| 14:24 | <mpilgrim> | the mask for the <H1 test is incorrect |
| 14:24 | <mpilgrim> | it's listed as "FF FF DF DF" |
| 14:24 | <mpilgrim> | but it should be "FF FF DF FF" |
| 14:25 | <hsivonen> | review FTW! |
| 14:25 | <mpilgrim> | test cases FTW |
| 14:25 | <mpilgrim> | annevk2: yeah, i figured that. i was hoping to find a standalone reference implementation |
| 14:25 | <mpilgrim> | i couldn't find one, so i wrote one |
| 14:25 | <mpilgrim> | with tests! |
| 14:26 | <annevk2> | trying to be an angel? :p |
| 14:26 | <mpilgrim> | :D |
| 14:27 | <mpilgrim> | this came about in a discussion with jcgregorio, the author of httplib2 |
| 14:28 | <mpilgrim> | httplib2 fetches resources and returns a dictionary of headers (as strings), and the body (as bytes) |
| 14:29 | <mpilgrim> | to convert the body to a string, you need to know the encoding |
| 14:29 | <mpilgrim> | to know the encoding, you need to know the content type |
| 14:29 | <mpilgrim> | (to know if it's even a text-based resource in the first place) |
| 14:30 | <mpilgrim> | (and if so, you need to know how to find an inline encoding declaration) |
| 14:31 | <annevk2> | you know that the full version of that for HTML resources depends on having a full HTML parser? |
| 14:31 | <mpilgrim> | i am aware of that too, yes |
| 14:31 | <annevk2> | k |
| 14:32 | <mpilgrim> | but someone has already written a full HTML parser |
| 14:32 | <mpilgrim> | httplib2 does the fetching |
| 14:32 | <mpilgrim> | html5lib does the HTML parsing |
| 14:33 | <mpilgrim> | i needed the piece in the middle |
| 14:33 | <Philip`> | Isn't the "<!DOCTYPE HTML" "Mask in Hex" one byte too short? |
| 14:33 | <mpilgrim> | Philip`: yes |
| 14:34 | <mpilgrim> | two bugs! |
| 14:34 | <Philip`> | Someone less lazy than me should report that |
| 14:36 | Lachy | is pondering what to do about the hasFeature() string in selectors api 2. |
| 14:37 | <Lachy> | Should I leave it as only returning true if the version is "1.0", or should I perpetuate the myth that it's useful by requiring implementations that support v2 to return true for "2.0" as well. |
| 14:37 | <jgraham> | This is getting increasingly insane |
| 14:38 | <annevk2> | hober, when you remove spam on the wiki, do you also block the creator? |
| 14:38 | <annevk2> | hobertoAtWork, ^^ |
| 14:39 | jgraham | would like to think of a good way of making hasFeature break without breaking sites |
| 14:39 | <annevk2> | Lachy, why did we introduce it in the first place? |
| 14:39 | <Lachy> | it's mostly useful for non-JS implementations |
| 14:40 | <annevk2> | Lachy, which? |
| 14:40 | <Lachy> | Java, apparently |
| 14:40 | <annevk2> | did anyone implement it? |
| 14:40 | <Lachy> | or so I was told by people claiming to want to use it in Java |
| 14:40 | <annevk2> | ugh |
| 14:40 | <Lachy> | yeah, it got implemneted in browsers |
| 14:41 | <annevk2> | you should have forbidden it for JS-based implementations |
| 14:41 | <Lachy> | I tried that |
| 14:41 | <Lachy> | that idea got rejected |
| 14:41 | <annevk2> | then we should've just dropped it completely imo |
| 14:43 | <Lachy> | it's a bit late for that, which is why I'm considering just leaving it as is so it can only be used to know if the implementation has some support for the api, rather than any specific level |
| 14:44 | <Lachy> | that way, it's a mostly harmless piece of legacy junk |
| 14:51 | <jgraham> | Maybe we should just change the spec to read "hasFeature(): mostly harmless" |
| 14:51 | <jgraham> | and by "the spec" I mean WebDOM |
| 14:52 | <Lachy> | redefine it in WebDOM to just return true for everything :-) |
| 14:55 | <jgraham> | I guess that might break sites |
| 14:55 | <jgraham> | Sadly |
| 14:56 | <Lachy> | might be interesting to find out what effect it would have in practice, if we could find sites that acutally use it |
| 15:07 | <jgraham> | It would be good to get it to a fixed list at least |
| 15:09 | <Philip`> | Lots seem to use it to check for XPath 3.0 |
| 15:11 | <hsivonen> | Philip`: must be "future proofing" considering that XPath is just getting to 2.2 and the version supported in browsers is 1.0 :-) |
| 15:11 | <hsivonen> | (yes, I realize 3.0 is the DOM level) |
| 15:14 | <Lachy> | it would be more useful if selectors api had a way of providing useful error reporting to scripts that indicated the exact cause of the problem, rather than providing an overly broad indicator like hasFeature() which says nothing about which selectors are support, or which methods are supported and what arguments they accept |
| 15:15 | <Lachy> | that functionality was requested in the past, but unfortunately the discussions revealed no practical way of doing implementing it |
| 15:15 | <Lachy> | anyway, spec checked in. http://dev.w3.org/2006/webapi/selectors-api2/ |
| 15:16 | <Lachy> | and the last major issue I need to start looking at is namespace support :-( |
| 15:19 | <Philip`> | I bet you're having fun today |
| 15:19 | <Philip`> | hasFeature and namespaces - what more could you ask for? |
| 15:20 | <annevk2> | extensibility |
| 15:20 | <annevk2> | distributed extensibility and a few comments from Roy on how you go against fifteen years of experience |
| 15:21 | <hsivonen> | and now it seems that RDFa not only comes with xmlns baggage but is also a delivery vector for @version |
| 15:23 | <Lachy> | at least hasFeaure only took 5 minutes to deal with. Namespaces are going to take much longer discussions about whether or not it's worth it, and then if it is, even longer discussions about how best to solve it |
| 15:23 | <annevk2> | hsivonen, RDFa designers took the blue pill :) |
| 15:24 | <Philip`> | Lachy: If you want to e.g. use the Selectors API on SVG (particularly SVG embedded in HTML), is there any way to make it work sensibly without adding namespace support? |
| 15:25 | <hsivonen> | annevk2: dunno about pills in this case, but it seems like RDFa pokes multiple areas that are considered harmful over here on #whatwg |
| 15:25 | <hsivonen> | annevk2: one such area could be written off as a matter of chance |
| 15:26 | <Lachy> | Philip`, sure, since in most cases, the tag names differ. |
| 15:26 | <Lachy> | the cases where they clash, like a and textarea, are a little more problematic |
| 15:27 | <Philip`> | I suppose "svg a" would work okay in most cases, except when you've got HTML embedded in SVG embedded in HTML |
| 15:27 | <jgraham> | Lachy: FWIW I find it hard to follow from the spec what the effect of refNodes is supposed to be |
| 15:27 | <Lachy> | jgraham, I explained it better with examples and use cases in a mail on public-webapps. One sec... |
| 15:28 | <Lachy> | http://lists.w3.org/Archives/Public/public-webapps/2009JulSep/1387.html |
| 15:28 | <jgraham> | I assume it is supposed to be related to filtering a node list for elements that match a selector or something, but I can't work out where it is actually specified |
| 15:30 | <Lachy> | it's specified that :reference pseudo-class matches the elements passed in the refNodes selector, and that scoped selector strings imply the presence of :reference at the beginning of each selector |
| 15:31 | <Lachy> | basically, the concept is that it allows you to select elements based on their relationship to one or more specific elements |
| 15:51 | <jgraham> | Lachy: I think youneed to be explicit when you describe the effect of the methods that refNodes is used to determine the set of contextual reference elements |
| 15:52 | <Lachy> | how is this not explicit enough? "The optional refNodes argument specifies one or more contextual reference element nodes." |
| 15:52 | <Lachy> | with a link to contextual refernece elements that clearly explains what it is and that they are matched by :reference |
| 15:52 | <jgraham> | Lachy: Because the specification reads like it operates by side effect |
| 15:53 | <Lachy> | huh? |
| 15:54 | <jgraham> | Lachy: In code terms, the specification reads like you currently set contextual reference nodes as a global variable and then call the match function which implicitly reads that global |
| 15:55 | <jgraham> | Lachy: I would prefer a style where you explicity say that refNodes is passed to the maych algorithm |
| 15:55 | <jgraham> | *match |
| 15:55 | <jgraham> | Does that make sense? |
| 15:58 | <Lachy> | ok, would it work if I specified in the Processing Selectors section something like "If the selector contains the :reference pseudo-class, the implementation must use the refNodes parameter to determine the contextual reference elements." |
| 16:00 | <zcorpan_> | hmm, http://www.w3.org/2006/02/son-of-3023/draft-murata-kohn-lilley-xml-03.html deprecates text/xml instead of fixing the issue |
| 16:00 | <annevk2> | is :reference the new :scope? |
| 16:00 | <annevk2> | zcorpan_, old news |
| 16:00 | <Lachy> | annevk2, yes |
| 16:00 | <zcorpan_> | annevk2: ok |
| 16:00 | <annevk2> | zcorpan_, Julian asked for some clarification |
| 16:00 | <jgraham> | Hmm, I think that would help, but it wouldn't really make Section 6.1 or 6.2 any clearer |
| 16:01 | <annevk2> | Lachy, I think :scope was a better term |
| 16:01 | <annevk2> | especially if browsers implement <style scoped> |
| 16:02 | <jgraham> | Lachy: Also "process this argument" doesn't really tell you that you are doing the spec equivalent of setting a variable |
| 16:03 | <Lachy> | :scope seemed innacurate when used for the purpose of filtering a collection of elements, as in document.querySelectorAll(".foo:reference", listOfElements); |
| 16:04 | jgraham | might be too used to the HTML5/ES5 style where everything is a pseudo-code algorithm |
| 16:04 | <annevk2> | oh, didn't know we needed such a feature |
| 16:04 | <Lachy> | it's bascially like $(elementCollection).filter("..."); in JQuery |
| 16:05 | <gsnedders> | jgraham, Lachy: if I end up doing stuff like Web DOM Core, I do want a fixed list, and I see no reason why that isn't possible. |
| 16:05 | <jgraham> | Lachy: Also typo in section 7 "must match the any element" |
| 16:06 | <Lachy> | fixed the typos |
| 16:09 | <Lachy> | jgraham, I'm not sure yet sure how to rewrite it to make it clearer, but I will think about it and get back to you. In the mean time, can you file a bug or send mail to public-webapps so I don't forget |
| 16:10 | <jgraham> | Where do I file bugs? W3C bugzilla? |
| 16:10 | <Lachy> | http://www.w3.org/Bugs/Public/enter_bug.cgi?product=WebAppsWG&component=Selectors%20API |
| 16:14 | <Lachy> | what if I said this in section 6.1? "When evaluating the selector, the implementation must use the refNodes argument, if any, to determine the contextual reference elements." |
| 17:12 | <mookid> | hi fans - new question about the cross domain requests mechanisms |
| 17:13 | <mookid> | anyone around to help a dummy get a clue? |
| 17:16 | <Philip`> | mookid: Someone might be, so it's probably good to just ask whatever question you have |
| 17:18 | <mookid> | ok I just don't know if this is the right place to be asking questions about that |
| 17:20 | <mookid> | trying to figure out the reason that OPTIONS was used in the handshake to figure out this 'Access-Control-Allow-Origin' header - and not HEAD |
| 17:24 | Philip` | doesn't remember, but supposes someone like annevk2 should know |
| 17:33 | <Philip`> | mookid: (HEAD doesn't seem to make sense because of its semantics - it's meant to be like an equivalent GET but without the response body, and if you're doing a pre-flight request for a POST to a resource that doesn't respond to GET then using HEAD would be weird) |
| 17:47 | <mookid> | Philip`: but the handshake is to establish apprpriate referrer domains for the resource |
| 17:48 | <mookid> | http methods are separate to this? |
| 17:49 | <mookid> | i.e. the mechanism can't define a resource that accepts GET but not POST from one domain |
| 17:50 | <mookid> | *one = another |
| 17:51 | <Philip`> | mookid: Access-Control-Allow-Methods sounds like it can do that |
| 17:52 | <mookid> | fair enough, thanks |
| 17:52 | <Philip`> | or you could just have a resource that always returns 405 Method Not Allowed for GET, and only accepts POST, and in that case it would be weird to use HEAD and expect a useful response |
| 17:52 | <Philip`> | (what with HEAD being a subset of GET) |
| 17:53 | <mookid> | it's not |
| 17:53 | <mookid> | :P |
| 17:54 | <mookid> | well |
| 17:54 | <mookid> | hmm thats a weird one |
| 17:54 | <mookid> | is it actually defined like that in the spec? |
| 17:56 | <Philip`> | "The metainformation contained in the HTTP headers in response to a HEAD request SHOULD be identical to the information sent in response to a GET request." |
| 18:14 | <Hixie> | usability study 6 |
| 18:14 | <Hixie> | it's interesting that all the paticipants so far have been getting that <img>'s data comes from src="" and <p>'s data comes from the element's contents |
| 18:14 | <Hixie> | this participant even explicitly mentioned this and said it was "obvious" |
| 18:14 | <Hixie> | excellent |
| 18:15 | <Hixie> | i was really worried that might be counter-intuitive |
| 18:17 | <tantek> | Hixie - microformats have well-established priori art and practice in that (getting data from <img src> and <p> content etc.) |
| 18:17 | <tantek> | so it's not really a surprise |
| 18:17 | <Hixie> | this participant has no microformats experience |
| 18:18 | <Hixie> | also some of the previous participants weren't able to determine if <a itemprop="" href=""></a> should get the value from the href="" or the element's contents |
| 18:18 | <tantek> | Hixie, that's because it depends on the property ("itemprop" in your example) |
| 18:18 | <tantek> | properties of type URL take their value from the href |
| 18:18 | <tantek> | whereas normal "string" properties that their value from the element's contents |
| 18:19 | <Hixie> | i don't really have any information to base the "why" on, i'm just observing what these participants have found intuitive or not |
| 18:19 | <tantek> | this is deliberate / by design and helps with efficient markup in very common cases (e.g. linking to a person and their web page by name) |
| 18:19 | <Hixie> | excellent, this participant has no problems with URL property names either |
| 18:20 | <Hixie> | so far only the participants who had trouble with the fundamentals have had trouble with full-url property names |
| 18:20 | <TabAtkins> | Awesome. |
| 18:20 | <Hixie> | which is interesting |
| 18:20 | <Hixie> | none so far have said anything about urls being too long, either |
| 18:20 | <Hixie> | which i found interesting |
| 18:21 | <tantek> | Hixie, do you mean "reverse DNS property names"? |
| 18:21 | <tantek> | or did something change recently to put full URLs as property names? |
| 18:21 | <hober> | tantek: it's been that way since the advent of microdata |
| 18:22 | <hober> | full urls, java-style property names, or short names from a predefined vocabulary |
| 18:22 | <tantek> | it's been reverse DNS property names since the advent of microdata |
| 18:22 | <hober> | yes, it's *also* been reverse DNS names |
| 18:22 | <tantek> | oh, *both* have been allowed? |
| 18:22 | <hober> | yup |
| 18:23 | <tantek> | were there any examples with URLs as property names? |
| 18:23 | <Hixie> | tantek: the study has only tested URLs and short names |
| 18:24 | <tantek> | are you going to test the reverse DNS property names? (those seemed like a much more interesting simplifying innovation of microdata) |
| 18:24 | <Hixie> | wasn't planning on it, since people have been so negative about them |
| 18:24 | <Hixie> | we don't really have a good solid proposal for dealing with names yet |
| 18:24 | <tantek> | Hixie - maybe the supporters of reverse DNS property names have simply quietly supported them, rather than trolling incessantly ;) |
| 18:25 | <Hixie> | maybe |
| 18:25 | <hober> | yes, the "moral majority" form of being Warnocked :) |
| 18:25 | <Hixie> | hah, this participant just complained about the vcard terminology ("locality is so vague, yet postal-code is so specific!") |
| 18:26 | <Hixie> | wow, she guessed <itemref ref="">'s meaning correctly without seeing any documentation |
| 18:27 | <tantek> | Hixie - not surprising, most don't "get" why vcard uses locality or region until they have to markup international addresses |
| 18:28 | <tantek> | same reason why people don't get what's wrong with 10/9/9 as a date until they travel to a country that uses a different DD MM order than they are used to. |
| 18:29 | gsnedders | assumes that means 2010-09-09 and gets on with life, thinking it isn't coming up soon |
| 18:29 | <tantek> | gsnedders :) |
| 18:29 | <Rik|work> | gsnedders: you should care even less, it was 3 weeks ago |
| 18:30 | <gsnedders> | Rik|work: Either way, no need to care about it now :) |
| 18:30 | <gsnedders> | I guess if you're one of these mad Americans you'd have to, though, but their loss. |
| 18:31 | <Hixie> | haha, interesting, i made a typo in the markup (item vs itemscope), and when i got the study moderator to fix it, the participant was like "ohh, that makes much more sense" |
| 18:31 | <Hixie> | and that was precisely what we were testing |
| 18:34 | <Hixie> | so, interesting. |
| 18:35 | <Hixie> | she understands containership this time -- she explicitly commented on how itemscope with itemprop=adr made sense |
| 18:35 | <Hixie> | (basically saying that a nested vcard makes sense) |
| 18:35 | <Hixie> | but she cannot express it! |
| 18:36 | <Hixie> | the exercise has her trying to write down the microdata from this example: http://damowmow.com/playground/microdata/004/review-annotated.html |
| 18:36 | <Hixie> | and she expressed it verbally very accurately, especially the last part with the itemprop=adr |
| 18:36 | <Hixie> | but when she tried to write it down, it lost all structure |
| 18:36 | <Hixie> | and now she's trying to mark up http://damowmow.com/playground/microdata/004/blog.html |
| 18:36 | <Hixie> | and she's getting it wrong again |
| 18:37 | <Hixie> | ooooh! wait! she's getting it! |
| 18:37 | <Hixie> | man this study more than anything tells me the power of getting the right examples |
| 18:38 | <Hixie> | the intro has <div>s and yet none of the examples she had to study carefully has <div>s |
| 18:38 | <Hixie> | so she's using a stray <body> :-( |
| 18:39 | <Hixie> | people really really don't want to include hidden metadata, that's hilarious |
| 18:39 | <Hixie> | they don't even know they don't |
| 18:39 | <Hixie> | they just don't do it |
| 18:40 | <TabAtkins> | I like these studies. Thanks for running them, Hixie. |
| 18:41 | <Hixie> | i wish we could do more! just not sure how to really test most things |
| 18:42 | TabAtkins | wishes the Dev team at his company would settle on a *single* rule for demarcing parts of a string as being replacable. The translators are getting horribly confused and translating things that should be treated as gensyms. |
| 18:43 | <Hixie> | sweet, she got that she had to use the predefined vocabularies |
| 18:44 | <Hixie> | and she got that itemid="" is how you refer to the image in a work |
| 18:44 | <TabAtkins> | That's a first so far, right? |
| 18:44 | <Hixie> | yeah pretty much! |
| 18:44 | <Hixie> | i added text to the intro to try to guide one to that conclusion |
| 18:44 | <Hixie> | maybe it worked, maybe she's just more in that mindset |
| 18:44 | <Hixie> | not sure what her background is |
| 18:44 | <Hixie> | (i could look it up, i just don't have it at hand) |
| 18:44 | <TabAtkins> | What intro did she read? |
| 18:45 | <Hixie> | http://damowmow.com/playground/microdata/004/introduction |
| 18:45 | <Hixie> | though she stopped about half way through |
| 18:46 | <TabAtkins> | Ah, that's some good text there. |
| 18:46 | <Hixie> | she replaced her stray <body> wrapping <h2>s and <p>s with a <span> |
| 18:46 | <Hixie> | it's an improvement i guess |
| 18:47 | <TabAtkins> | I think your examples use <span>s when hooking metadata to stuff that doesn't naturally have an element already. |
| 18:48 | <Hixie> | i tried to use <div>s for almost all the itemscope=""s in the intro |
| 18:48 | <Hixie> | but none of the the examples she looks at do |
| 18:48 | <Hixie> | they use <body>, <li>, and <tr> respectively |
| 18:48 | <Hixie> | oh an <span> for an itemprop=adr once |
| 18:48 | <TabAtkins> | Making *everything* start with "item" was a win I think. |
| 18:48 | <Hixie> | yeah i think that helps |
| 18:49 | <TabAtkins> | The confusion of having lots of seemingly unrelated names working together is one of the big loses in RDFa to me. |
| 18:50 | <Hixie> | someone suggested renaming them to iscope, itype, and iprop or iname; someone else suggested mdobject and mdname |
| 18:50 | <Hixie> | personally i prefer item |
| 18:50 | <TabAtkins> | Yeah, item is better. It's transparent and obvious. |
| 18:50 | <Hixie> | i mean, itemscope, itemtype, and itemprop (and <itemref>) |
| 18:50 | <TabAtkins> | And easy to type, so the extra characters over "i" aren't that significant. |
| 18:51 | <Hixie> | i love how every. single. person. who has tried to mark up blog.html ends up using the <h2>s as the itemprop=titles |
| 18:51 | <Hixie> | when in fact none of them are |
| 18:53 | <Hixie> | ok, blog-annotated.html makes a lot of sense to her |
| 18:53 | <Hixie> | that's encouraging |
| 18:53 | <TabAtkins> | I can see that. Titles are titles. |
| 18:53 | <Hixie> | it matched what she had been verbally saying she wanted, but not what she did |
| 18:53 | <Hixie> | which is interesting |
| 18:54 | <TabAtkins> | Perhaps people aren't trusting themselves to mark up 'naturally'? |
| 18:55 | <TabAtkins> | Instead assuming there has to be some more complex structure they should follow? |
| 18:56 | <TabAtkins> | I've had that "it can't be this easy" feeling before when confronted with something that just falls into place. |
| 18:57 | <TabAtkins> | Hell, I still feel that way anytime I use Markdown. |
| 18:57 | <Hixie> | yeah she even said that she was trying to use a more complex study |
| 18:57 | <Hixie> | er |
| 18:57 | <Hixie> | structure |
| 18:57 | <Hixie> | not study! |
| 18:57 | <TabAtkins> | Hixie, we need a <markdown> element. Plain-text formats ftw. |
| 18:58 | <Hixie> | hah, she said "i kinda hope that this list of predefined vocabularies isn't the real list" |
| 18:59 | <TabAtkins> | Heh, she wants more? |
| 18:59 | <Hixie> | well the list is basically just the terms i used in the examples |
| 18:59 | <Hixie> | so it's a pretty dumb list |
| 19:02 | <zcorpan_> | is abarth paying attention to htmlwg bugs? |
| 19:04 | <lmorchard> | HTML 5 spec question: Did the <datalist> element support a data attribute that later got removed? Seemed like it did in Web Forms, and HTML 5 examples show it in use. |
| 19:04 | <lmorchard> | Tried grepping though an svn log, but didn't find the answer |
| 19:04 | <zcorpan_> | yeah, it was removed in the wf2->html5 integration |
| 19:05 | <lmorchard> | Huh, oh well. It seemed useful |
| 19:05 | <zcorpan_> | yeah |
| 19:06 | <zcorpan_> | but not too hard to emulate |
| 19:06 | <Hixie> | lmorchard: yeah, it might be added back later. it had its issues, not too many advocates, and frankly not too many use cases, so i figured it would be better to trim the fat. |
| 19:06 | <Hixie> | ok, participant 5 is done |
| 19:06 | <Hixie> | 6th and final one this afternoon |
| 19:06 | <lmorchard> | Seemed nice to have a native autocomplete data source |
| 19:06 | <Hixie> | we'll just run the same 004 study again to find out if this morning was a fluke or what |
| 19:07 | <Hixie> | lmorchard: yup |
| 19:07 | <Hixie> | lmorchard: probably are better ways to do it if that's the exact use case though |
| 19:07 | <Hixie> | lmorchard: e.g. using some sort of protocol over WebSocket, so you don't have to do a TCP round trip for every keystroke |
| 19:07 | <Hixie> | er, two TCP roundtrips |
| 19:07 | <Hixie> | one to create the connection and one to get the data |
| 19:08 | <lmorchard> | Ah, true. Then you don't need to produce, deliver, and parse a whole body of autocomplete choices |
| 19:08 | <Philip`> | (Isn't the first roundtrip solved by keepalive?) |
| 19:09 | <zcorpan_> | also, oninput isn't every keystroke |
| 19:09 | <Philip`> | When I stroke my keys, nothing happens |
| 19:10 | <TabAtkins> | Philip`, you're not doing it right. Treat them like you love them, not like they're just a piece of meat. |
| 19:10 | Philip` | presumably needs a touch-sensitive keyboard |
| 19:11 | <Philip`> | If your keyboard is a piece of meat, I think you should take it back to the shop |
| 19:11 | <TabAtkins> | Meat-based computing is the future. |
| 19:11 | <lmorchard> | They're made out of meat! |
| 19:27 | <TabAtkins> | Woohoo! Got a medical study just sent to me to see if a vaccine works by actually exposing me to the virus and seeing if I get sick. Totally worth it for $1100, though. ^_^ |
| 19:28 | <mpilgrim> | TabAtkins: we should test RDFa that way |
| 19:29 | <TabAtkins> | mpilgrim: There are some things I wouldn't risk. |
| 19:39 | <zcorpan_> | TabAtkins: swine flu? |
| 19:41 | <Philip`> | TabAtkins: Zombism? |
| 19:42 | <TabAtkins> | zcorpan_: Nah, it's something called "Cruise Ship Illness and Intestinal Flu". |
| 19:43 | <Philip`> | Sounds jolly |
| 19:44 | jgraham | wonders if "#whatwg on freenode.net" would be a suitable (if non unique) answer to "where does Ian Hickson live on the web?" |
| 19:44 | <Philip`> | TabAtkins: How long would you have to be sick for, before you considered $1100 no longer worth it? |
| 19:45 | gsnedders | grumbles at having to start up word |
| 19:45 | <Philip`> | jgraham: #whatwg is on the web now? |
| 19:45 | <TabAtkins> | Philip`: Depends on the severity of the illness. |
| 19:45 | jgraham | wonders if he is along in finding the idea that people "live at" or are "identified" by some URI they happen to control to be odd |
| 19:45 | <jgraham> | Philip`: Hmm. Good point. |
| 19:45 | <jgraham> | *alone |
| 19:46 | <gsnedders> | zcorpan_: What does "trappstädning" mean? |
| 19:46 | Philip` | partly-intentionally avoids having a single URL that he considers to be anything like a homepage |
| 19:46 | <TabAtkins> | Philip`, I've done the same thing. www.xanthir.com is just a place for me to have random fun. |
| 19:46 | <TabAtkins> | And see what the current time is. |
| 19:47 | <jgraham> | gsnedders: Google translate says "stair cleaning" |
| 19:47 | jgraham | guesses you looked at that already |
| 19:47 | gsnedders | prefers to give out secret.gsnedders.com |
| 19:47 | <gsnedders> | jgraham: Yeah, I did. I guess it's right, as it does just about make sense :P |
| 19:48 | <TabAtkins> | Now, as to when $1100 becomes no longer worth it, I'd probably rate it as a bit over a week. For a general intestinal flu, I'd consider compensation similar to what I make per day at my job to be fair. |
| 19:48 | <Philip`> | gsnedders: It looks like "trap standing", so presumably it involves finding mousetraps and bear traps etc on which to stand for laughs |
| 19:48 | jgraham | is far too lazy to have anything that resembles a coherent homepage |
| 19:48 | <gsnedders> | Philip`: That doesn't really make sense. |
| 19:48 | <Philip`> | gsnedders: What, standing on traps? |
| 19:48 | <jgraham> | Or rather it never seems that interesting to make one |
| 19:48 | gsnedders | vandalizes his, apparently |
| 19:48 | <Philip`> | It's a way to pass the time |
| 19:49 | <zcorpan_> | gsnedders: what jgraham said |
| 19:49 | <jgraham> | gsnedders: Yeah, well, you have been warned about that |
| 19:49 | <gsnedders> | Philip`: So it's my landlord's responsibility, and not mine? Oh well. I guess I'm sad to miss out. :( |
| 19:49 | <gsnedders> | Also, I'm apparently "a very nice, a bit humble and easy going person", which is a bit of a worrying description. |
| 19:51 | <jgraham> | At least they didn't just call you "easy" |
| 19:52 | <gsnedders> | Well, yeah. |
| 20:03 | TabAtkins | is desperately thankful he remembered a browser game he played 10+ years ago as the answer to the secret question when recovering his password. |
| 20:04 | <zcorpan_> | http://net.tutsplus.com/articles/web-roundups/5-lesser-known-javascript-libraries-that-make-web-design-easier/ - talk about over engineering things (the Modernizr audio example) |
| 20:04 | <zcorpan_> | let's test for audio support! if it's not supported, let's hide the element! |
| 20:05 | <zcorpan_> | because <audio src></audio> does ... something bad in old browsers? |
| 20:05 | <zcorpan_> | but wait, then let's test for audio support, and if it is supported, let's hide the download link that is next to the audio! |
| 20:06 | <zcorpan_> | because this way is better than just <audio src><a href>link</a></audio> |
| 20:06 | <lmorchard> | It's got electrolytes |
| 20:07 | <Hixie> | why would they hide the download link at all |
| 20:07 | <TabAtkins> | Ooh, I didn't know about Firebug Lite. Awesome. |
| 20:07 | <zcorpan_> | Hixie: dunno, maybe they want to rely on the browser providing an easy way to download the audio if it supports <audio> |
| 20:08 | <zcorpan_> | i didn't get firebug lite to work, though i only tried to apply it as a bookmarklet |
| 20:12 | <paul_irish> | zcorpan_: yah that's kind of useless. (i maintain modernizr, btw). i'll shoot the author a message to have him do it properly. :) |
| 20:13 | <remysharp> | hi - I can't find the -data attribute in the html5 spec, has it been removed or am I just being lame at searching? |
| 20:13 | <Hixie> | search for data-* |
| 20:13 | <Philip`> | paul_irish: If you're the right person to report this to: I think there's a typo in the project's name, it should have an extra 'e' |
| 20:13 | <remysharp> | or rather "data-*" |
| 20:13 | <paul_irish> | Philip`: ah, can you file a bug? :P |
| 20:14 | miketaylr | ducks from Philip` |
| 20:14 | <Hixie> | remysharp: http://www.whatwg.org/specs/web-apps/current-work/#attr-data-* |
| 20:14 | <Philip`> | remysharp: http://www.whatwg.org/specs/web-apps/current-work/multipage/elements.html#embedding-custom-non-visible-data |
| 20:15 | <zcorpan_> | paul_irish: i've already written a comment :) |
| 20:15 | <paul_irish> | Philip`: thx for the web font optimizer btw. i'm hoping to add support for any font file to it. |
| 20:15 | <remysharp> | cheers |
| 20:16 | <Philip`> | paul_irish: The web UI is restricted to some free ones, but the code at http://bitbucket.org/philip/font-optimizer/ ought to work with any font |
| 20:16 | <Philip`> | Well, any non-CFF font |
| 20:16 | <paul_irish> | yah that's what it looked like. |
| 20:16 | <Philip`> | except ones it doesn't work with |
| 20:16 | <Philip`> | but I've tested on quite a few |
| 20:16 | <zcorpan_> | you can choose any color you want so long as it's black |
| 20:17 | <paul_irish> | very cool. i'm also hoping to support something like font segmenting. so half the glyphs go in one file and half in the rest.. its a trick typekit is doing these days. |
| 20:17 | <remysharp> | Re: custom data, if I have: <div data-lat-long="12,12"> - could I target it using this.dataset['lat-long'] (note particular use of dashes) |
| 20:18 | <zcorpan_> | remysharp: yes |
| 20:18 | <Philip`> | paul_irish: That seems much harder when you want to support ligatures (particularly for non-Western languages which depend heavily on glyph substitutions) and kerning |
| 20:18 | <Philip`> | paul_irish: As far as I'm aware, Typekit's approach is to not support those things |
| 20:18 | <paul_irish> | Philip`: i agree. we'll see how far i can reasonably take it. |
| 20:19 | <paul_irish> | yah that's what i've seen as well. |
| 20:19 | <Philip`> | paul_irish: (My code tries very hard to preserve all of that kind of information, while still optimising away as much as possible) |
| 20:19 | <Hixie> | zcorpan_: what does "("[^...]" is subtracted from Char.)" mean in http://www.w3.org/Bugs/Public/show_bug.cgi?id=7488#c3 ? |
| 20:19 | <Philip`> | paul_irish: ((The GSUB/GPOS tables can be hundreds of KB in some font, so they do need to be optimised when subsetting)) |
| 20:21 | <zcorpan_> | Hixie: [^a-z], [^#xN-#xN] |
| 20:21 | <zcorpan_> | matches any Char with a value outside the range indicated. |
| 20:22 | <Hixie> | right |
| 20:22 | <Hixie> | so? |
| 20:22 | <zcorpan_> | Hixie: so you need to check against Char, too |
| 20:22 | <zcorpan_> | Hixie: not just check for " and ' |
| 20:23 | <Hixie> | oh, i see |
| 20:23 | <Hixie> | ok, thanks |
| 20:23 | <Hixie> | sorry didn't realise we were checking anything at all yet |
| 20:25 | <Hixie> | i wonder why they used [^"]* for systemliteral, but used Char-'-' for comments |
| 20:25 | <Hixie> | if it means the same |
| 20:27 | Hixie | wonders why XML allows the second block of <control>s but no the first |
| 20:27 | <zcorpan_> | it's Char-'--' |
| 20:28 | <Hixie> | right |
| 20:28 | <Hixie> | but why not use Char-'"'-"'" or some such? |
| 20:28 | <Hixie> | whatever the right syntax is |
| 20:28 | <zcorpan_> | dunno |
| 20:33 | <Hixie> | ok tried to fix it |
| 20:33 | <Hixie> | thanks |
| 20:38 | zcorpan_ | bets there's still a case it doesn't cover |
| 20:47 | <jgraham> | Philip`: It seems like the intersting number wrt RDFa versioning is how many existing implementations actually ignore the version parameter i.e. parse version="2" identically to version="1" |
| 20:47 | <jgraham> | s/parse/interpret/ |
| 20:48 | <Philip`> | jgraham: All the ones that ignore @version entirely will do that |
| 20:49 | <hsivonen> | do any existing pieces of software claimed to be RDFa consumers modify their behavior depending on version attribute values? |
| 20:49 | <Philip`> | I suppose it's theoretically possible they accept both version-1 and absent-version |
| 20:49 | Philip` | could look at the implementation code, he guesses |
| 20:50 | zcorpan_ | notes that xml 1.0 interprets version="1.0" and version="1.9" the same but version="2.0" differently |
| 20:50 | <zcorpan_> | 5th ed, that is |
| 20:51 | <hsivonen> | zcorpan_: that doesn't count because they changed that after the 1.1 disaster |
| 20:52 | <hsivonen> | zcorpan_: a case study should consider pre-1.1 parsers that were built to self-sabotage when fed the 1.1 identifier |
| 20:52 | <jgraham> | Philip`: Yeah it's that theoretical possibility I was thinking of |
| 20:52 | <zcorpan_> | i wonder what the xml core wg will do if/when they find that no-one is adopting 5ed |
| 20:52 | <jgraham> | But the whole idea of versioning doesn't seem like it can work so I woudln't waste too much time on it |
| 21:04 | <zcorpan_> | (which resource you end up with |
| 21:04 | <zcorpan_> | depends on your Google login credentials) |
| 21:05 | <zcorpan_> | can this be exposed to other pages? |
| 21:05 | <zcorpan_> | <iframe src=google-i'm-feeling-lucky> |
| 21:05 | <zcorpan_> | window[0].URL |
| 21:08 | <TabAtkins> | Hmm, what's Neuberg's nationality? |
| 21:09 | <TabAtkins> | I'm watching his HTML5 video, and I noticed that he uses the German way of counting on fingers for his "5 steps" graphic. |
| 21:10 | <hsivonen> | TabAtkins: what's the German way? |
| 21:10 | <TabAtkins> | Starting with the thumb. |
| 21:10 | <TabAtkins> | In America we end with the thumb, only using it for "5". |
| 21:11 | <zcorpan_> | sweden does it the American way |
| 21:11 | ttepasse | can't extend little and ring finger for counting to two ??but then he's german. |
| 21:11 | <zcorpan_> | although the german way is easier |
| 21:12 | <TabAtkins> | ttepasse: Sorry, bad description. ^_^ 1 is index finger, 2 is index and middle, etc. |
| 21:12 | <zcorpan_> | ttepasse: you start with index finger |
| 21:12 | <ttepasse> | Uh. Ok. |
| 21:13 | <ttepasse> | At least it's not binary. |
| 21:13 | <TabAtkins> | Binary is really easy on the fingers. |
| 21:13 | <TabAtkins> | I use it all the time. |
| 21:14 | <zcorpan_> | 4 is harder the german way |
| 21:14 | <TabAtkins> | Frex, I can count out how long to do my mouthwash on one hand. |
| 21:14 | <zcorpan_> | (than 3 the american way) |
| 21:14 | <TabAtkins> | Yeah, the American 3 can use the thumb to hold down the pinky. |
| 21:15 | <Hixie> | final participant |
| 21:15 | <TabAtkins> | woo |
| 21:15 | <Hixie> | he just said something that directly contradicted something this morning's participant said |
| 21:15 | <Hixie> | how confoosing |
| 21:15 | <Hixie> | i want my participants to all be consistent with each other so my life is easy! |
| 21:15 | <Hixie> | ("my" participants :-P) |
| 21:15 | <ttepasse> | The new feudalism. |
| 21:16 | <zcorpan_> | you should have chosen other participants, then |
| 21:17 | <hsivonen> | Hixie: naysayers won't approve of your selection of participants even if they weren't chosen for your convenience |
| 21:18 | <Hixie> | zcorpan_: i didn't chose any of them, we have an independent team for that |
| 21:19 | <Hixie> | hsivonen: oh i don't really care about the naysayers, i'm just trying to make it better :-) |
| 21:46 | <TabAtkins> | Man, now I wanna start using Workers. |
| 21:46 | <TabAtkins> | Damn IE upgrade cycle. |
| 21:47 | <Philip`> | Why "now"? |
| 21:49 | <TabAtkins> | Neuberg's video. |
| 21:50 | <TabAtkins> | While I've vaguely tracked Workers on the list, I didn't realize how easy they were to use. |
| 22:01 | <hsivonen> | with Chrome Frame, IE8 has 7 layout modes, right? |
| 22:02 | <gsnedders> | hsivonen: Yeah, I think |
| 22:08 | <sicking> | Hixie: they should have called it "Canvas Accessibility Taskforce" instead |
| 22:14 | <Hixie> | what _did_ they call it? |
| 22:15 | <Hixie> | sicking: do you know anything about https://wiki.mozilla.org/Security/Sec-From ? |
| 22:15 | <Hixie> | specifically, do you know if "*If ancestor tree is more than 1 deep, send "null"" still applies now that we have the list version of Origin |
| 22:16 | <sicking> | Hixie: don't know what they called it, but i'm all about backronyms |
| 22:16 | <Hixie> | hehe |
| 22:16 | <sicking> | Hixie: don't know that no :( |
| 22:16 | Hixie | summons abarth |
| 22:16 | <Hixie> | that's not abarth. |
| 22:17 | Hixie | tries again with a chicken: |
| 22:17 | <Hixie> | . o |
| 22:17 | <Hixie> | ('>) |
| 22:17 | <Hixie> | LL |
| 22:17 | <Hixie> | ok that's just disturbing. |
| 22:17 | <gsnedders> | Blatantly you control annevk42. |
| 22:19 | <TabAtkins_> | Lachy, forgive me for being stupid, but *why* would we need special namespaced versions of the querySelector* functions? You can use namespaces directly in selectors. |
| 22:20 | <annevk42> | hmm, I knew I shouldn't go online |
| 22:20 | <TabAtkins_> | Haha. |
| 22:21 | TabAtkins_ | sighs. |
| 22:21 | <TabAtkins_> | Oh, Roy T. Fielding. Why you so crazy? |
| 22:22 | <Lachy> | TabAtkins_, to provide namespace resolution mechanism. |
| 22:22 | <annevk42> | no, the answer is "just say no" |
| 22:22 | <TabAtkins_> | Lachy, I assume this is something beyond CSS's normal namespace handling? If so I don't really understand it, so shrug. |
| 22:23 | <Lachy> | annevk42, just saying no doesn't help. But the question was just about why the would be needed if we do introduce namespaces. I'm not saying we will or will not do so at this stage. |
| 22:23 | <Lachy> | @namespace can't be used in the API. That applies to CSS only. |
| 22:24 | <annevk42> | Hixie, Sec-From is dead |
| 22:24 | <TabAtkins_> | Point me at an example of how the current *NS functions do that? |
| 22:25 | <Lachy> | TabAtkins_, document.getElementsByTagNameNS("http://www.w3.org/2000/svg";, "svg"); |
| 22:26 | <TabAtkins_> | Thought so. How would that work in querySelectorNS? You can potentially have multiple different namespaces in a single selector. |
| 22:26 | <Lachy> | For selectors api, we would need a way to resolve any number of prefixes and the default ns in, e.g., "svg|textArea, xh|textarea" |
| 22:26 | <Hixie> | annevk42: no, it's just renamed Origin |
| 22:26 | <TabAtkins_> | kk. |
| 22:27 | <annevk42> | Hixie, which already existed |
| 22:27 | <Hixie> | sure but i still have to define the Privacy-Sensitive Contexts stuff |
| 22:27 | <Lachy> | the original proposal that was tried with selectors api v1 was to pass an NSResolver object that implements the lookupNamespaceURI() method (or simply a function in JS) that returned the namespcae URI |
| 22:28 | <annevk42> | Hixie, anyway, afaik abarth is following what I specified which means it is only null when there's no serialization |
| 22:28 | <annevk42> | Hixie, and it's space-separated in case of redirects |
| 22:29 | <Hixie> | annevk42: it's also null for regular hyperlinks, navigations of any kind other than form submission, <img>s, and style sheets, apparently |
| 22:32 | <TabAtkins_> | Lachy: The most natural js way to do it would be to just pass an object with the keys being the ns and the values being urls. {"svg":"http://www.w3.org/2000/svg";} as the first argument, frex. |
| 22:33 | <Lachy> | that was suggested but doesn't really solve the problems that existed with the function because of the ability to define getter functions. |
| 22:33 | <TabAtkins_> | I don't know enough context to understand that sentence. |
| 22:33 | <sicking> | annevk42: no, it's null in a lot more situations |
| 22:33 | <Hixie> | lachy: get a structured clone of the value passed, then use it :-) |
| 22:33 | <Lachy> | anyway, I'm not looking at the namespace issue right now and will not look into any possible solution until the use cases have been made clear. |
| 22:34 | <sicking> | annevk42: basically it's null when we don't think we can trust where a request-uri came from |
| 22:34 | <sicking> | annevk42: for example with <a> links, it's very common that the link is generated by someone other than the site owner |
| 22:35 | <TabAtkins_> | Lachy: Sure, I was just curious and wanted to understand some issues better before I responded. |
| 22:35 | <Lachy> | Hixie, that might work. I'll keep that in mind. |
| 22:35 | <sicking> | annevk42: for example most forums allow you to insert links, as do blog comments and gmail |
| 22:38 | <Hixie> | <iframe> <embed> <applet> send origin; <img> does not |
| 22:38 | <Hixie> | should <object> ? |
| 22:39 | Philip` | notes that Gmail seems to do some magic link rewriting so you don't actually get directly linked from emails |
| 22:41 | <annevk42> | Hixie, ah right |
| 22:42 | <annevk42> | Hixie, would be nice if <img> worked with CORS so you can prevent tainting the <canvas> |
| 22:42 | <Hixie> | yes, CORS needs to be applied all over the place |
| 22:42 | <Hixie> | eventsource, img, etc |
| 22:42 | <Hixie> | but i'm waiting for cors to be deployed and proven good before i start adding it everywhere |
| 22:42 | <gsnedders> | won't that cause web compat issues? |
| 22:45 | <Hixie> | why? |
| 22:55 | <annevk42> | Hixie, fair enough |
| 22:55 | <annevk42> | gsnedders, CORS can be used in an additive way, e.g. the <img> can still load and be displayed even if the resource sharing check failed |
| 23:00 | <Hixie> | ok since nobody replied, i'm assuming <object> is not privacy-sensitive |
| 23:06 | <Hixie> | wait, we can't do CORS for <img> if we don't send Origin |
| 23:06 | <Hixie> | and similarly, that means that <video> and <audio> will need to send Origin if we are ever to do CORS |
| 23:06 | <Hixie> | so i might as well do it already |
| 23:06 | <da3d> | What is CORS? |
| 23:08 | <Hixie> | http://dev.w3.org/2006/waf/access-control/ |
| 23:09 | Hixie | presumes <input type=image> is the same as <img> |
| 23:15 | <TabAtkins_> | Lachy: Also, the hypothetical selector() function should be named $. |
| 23:28 | <Hixie> | TabAtkins: document.$? |
| 23:28 | <TabAtkins> | Hixie: jk, really. ^_^ |
| 23:28 | <Philip`> | Hmm, looks like PyRDFa emits a warning (no idea where, since I've never seen it) if a document doesn't have either an appropriate doctype or appropriate <html version> or appropriate <head profile> |
| 23:28 | <Hixie> | TabAtkins: :-P |
| 23:29 | <Philip`> | If you pass the rdfa_strict_version option then Swignition requires <html version> =~ /rdfa/i else it doesn't extract RDFa |
| 23:30 | <Philip`> | rdfQuery seems to ignore version/profile entirely |
| 23:31 | <Philip`> | rdfa.js (whatever it's meant to be called) too |
| 23:33 | <Philip`> | librdfa too, I think |
| 23:33 | <Philip`> | (For PyRDFa, "appropriate" means equal to the RDFa 1.0 specific strings) |
| 23:35 | <Hixie> | so basically the rdfa wg made the same mistake as the xml wg but weren't as successful as shooting themselves in the foot? |
| 23:35 | <Hixie> | er, at shooting, not as shooring. |
| 23:35 | <Hixie> | i give up. |
| 23:38 | <Philip`> | You could argue that they were *less* successful than shooting themselves in the foot, because they're trying to shoot everybody else's feet too |
| 23:39 | <Philip`> | (I don't know that it would be a valid or justifiable thing to argue, but that's no reason not to attempt to interpret your pre-typo text) |
| 23:40 | <Philip`> | (Um, pre-typo-fix) |
| 23:40 | <Hixie> | i was arguing that they were less successful at shooting themselves in their foot, yes |
| 23:40 | <Philip`> | (pre-attempted-typo-fix) |
| 23:41 | <Hixie> | in that they tried to make it very difficult for themselves to extend their language while supporting legacy content in new UAs, but their UAs by and large ignored that requirement, so they'll be ok |
| 23:41 | Philip` | is unsure whether success in foot-shooting should be measured in terms of hits or misses |
| 23:41 | <Hixie> | heh |
| 23:41 | <Hixie> | it's interesting to see WGs try to use versioning |
| 23:41 | <Hixie> | i wonder if there are any examples of it being successful |
| 23:42 | <Hixie> | in any multivendor technology |
| 23:42 | <Hixie> | i certainly haven't heard of any success stories |
| 23:42 | <TabAtkins> | ffs, seriously? In this day and age I have to type in a custom-format url to locate something on my computer? |
| 23:42 | TabAtkins | is angry at Microsoft right now. |
| 23:42 | <Philip`> | "they'll be ok" - not if they want to make backward-incompatible changes, like defaulting to plain text literals instead of XMLLiterals |
| 23:42 | <Hixie> | well if they want to make backwards-incompatible changes they've lost already |
| 23:43 | <Hixie> | so i was assuming they wouldn't do anything like that |
| 23:43 | <Philip`> | If they don't make the backwards-incompatible change then they've also lost, because they'll be stuck with a design mistake |
| 23:43 | <Philip`> | So either way, they're not entirely okay |
| 23:44 | <Philip`> | Presumably the only solution is to be really really sure you've got the design right at the start, before anybody's started using it |
| 23:46 | <TabAtkins> | Or just don't care all that much about the people you're messing up when you make a breaking change. |
| 23:48 | <Hixie> | or to put up with design mistakes -- html's done pretty well for itself with a whole raft of horrendous design mistakes |
| 23:50 | <Philip`> | Indeed, you need to be really really sure you've got the design right at the start, before anybody's started using it, and once somebody's started using it you have to put up with the raft of horrendous design mistakes you suddenly discover |
| 23:50 | <gsnedders> | Well, that's better than PHP, seeming that has no design |
| 23:53 | <Philip`> | "A file cannot have infinite length" - what about /dev/urandom? |
| 23:54 | <Hixie> | where does it say that |
| 23:54 | <Philip`> | On public-html |
| 23:54 | <Hixie> | oh |
| 23:55 | <Hixie> | well, that e-mail is far ore productive than most on that thread have been |