| 13:22 | <hsivonen> | http://validator.w3.org/check?uri=http%3A%2F%2Fwww.w3.org%2F2001%2Ftag%2F&charset=%28detect+automatically%29&doctype=Inline&group=0 |
| 13:22 | <hsivonen> | Is <img caption> from some published flavor of (X)HTML? |
| 13:24 | <zcorpan> | "Attribute "caption" is not a valid attribute. Did you mean "action"?" |
| 13:24 | <zcorpan> | is action a valid attribute on img?? |
| 13:24 | <hsivonen> | zcorpan: bad spelling suggest feature, it seems |
| 13:30 | <Philip`> | Automatic spelling suggestion seems like a good idea |
| 13:30 | <Philip`> | Not restricting it to attributes valid on the given element seems silly, though |
| 13:39 | <zcorpan> | i didn't know it had spelling suggest in the first place |
| 13:40 | <zcorpan> | wonder if you specify action, does it suggest caption? |
| 13:42 | <Philip`> | Hmm |
| 13:42 | <Philip`> | I tried <!doctype html><img action=1 caption=1> in validator.w3.org |
| 13:42 | <Philip`> | It says: "This document was successfully checked as HTML5! Result: Passed, 3 warning(s)" |
| 13:43 | <hsivonen> | whoa |
| 13:43 | <hsivonen> | Philip`: not on V.nu, fortunately |
| 13:43 | <Philip`> | If I force the HTML4 doctype: "Attribute "ACTION" is not a valid attribute. Did you mean "action"?" |
| 13:51 | <Lachy> | w3 validator seems to be allowing any random attributes on any element |
| 13:51 | <Lachy> | for HTML5 |
| 13:52 | <Philip`> | Probably only in the direct input mode |
| 14:10 | <foolip> | hsivonen: is RELAX NG able to validate somewhat complex data types like HTML5 date, or is that done as a separate check now? |
| 14:13 | <hsivonen> | foolip: HTML5 date is validated by a pluggable datatype library |
| 14:13 | <foolip> | OK |
| 14:13 | <Philip`> | It'd be trivial in RELAX NG if years were limited to 4 digits, because you could just enumerate all dates |
| 14:13 | <foolip> | concerning microdata I suppose something similar might be needed for some item types |
| 14:13 | <Philip`> | but it looks like they're unbounded :-( |
| 14:14 | <foolip> | I think we should just make HTML5 year be limited to 4 digits, as I think I've argued in some mail Hixie hasn't gotten to yet |
| 14:14 | <Philip`> | Oh, okay, then RELAX NG should definitely be able to validate it |
| 14:15 | <foolip> | doesn't "enumerate all dates" involve creating a very big list? |
| 14:15 | <foolip> | unless you can express the leap years in RELAX NG somehow? |
| 14:15 | <Lachy> | foolip, as long as extending it to 5 digits in 7900 years won't be a problem, then limiting to 4 digits now should be ok |
| 14:15 | <hsivonen> | scary. the HTML5 parser in Gecko had totally bogus code for document.write() in document.open()ed docs, but there were no Web compat bug reports |
| 14:16 | <foolip> | I think >4 digits just adds complexity and nuisance for now, and can't see any real benefit |
| 14:16 | <Lachy> | if the 4 digit limit was just for document conformance, as a useful way to catch authoring mistakes, then extending it later wouldn't be a problem |
| 14:16 | <foolip> | expressing exact dates 8000 years into the future seems quite useless |
| 14:16 | <Lachy> | yes |
| 14:17 | <foolip> | and it's likely to let the year 20010 through where it wasn't intended |
| 14:17 | <Philip`> | foolip: I was just thinking <attribute name="datetime"><choice><value>0001-01-01</value><value>0001-01-02</value>... |
| 14:18 | <Philip`> | and you can encode the days-per-month thing into the list easily |
| 14:18 | <Lachy> | foolip, at the very least, validators should probably warn about 5 digit dates |
| 14:18 | <Philip`> | and leap years |
| 14:21 | <Philip`> | Should work for datetimes too, if you put an arbitrary-number-of-digits-matching thing on the end |
| 14:21 | <adactio> | I just signed a lease extension for the house I'll be getting a mortgage for ....the lease extension was for 999 years. Just sayin'. ;-) |
| 14:21 | <foolip> | also, 2010 and 2010-01 should be valid dates, but I don't think I've requested that by mail yet |
| 14:22 | <Lachy> | Philip`, it's at this point a normal person would realise that perhaps RelaxNG isn't an appropriate tool for the job of checking a date is within range |
| 14:22 | <foolip> | adactio: wasn't it you arguing for relaxing <time> a bit a while back? |
| 14:22 | <Philip`> | foolip: What day is represented by 2010-01? |
| 14:22 | <adactio> | foolip: yes indeed. I too think that 2010 and 2010-10 should be valid values. |
| 14:22 | <gsnedders> | Lachy: Are you calling us freak!? :'( |
| 14:23 | <gsnedders> | *freaks |
| 14:23 | <foolip> | Philip`: the answer is the same as for "what second is represented by 18:45" |
| 14:23 | <foolip> | i.e. none |
| 14:23 | <Philip`> | Lachy: It's a perfectly adequate tool, it's just that implementations are burdened with limits like storage space and time |
| 14:23 | <adactio> | Philip: length of work experience and/or education e.g. in resumés, CVs are usually represented as YYYY-MM. |
| 14:24 | <Philip`> | foolip: It represents the second 0, according to the parsing spec |
| 14:24 | <Lachy> | adactio, that's nothing. Arthur Guiness signed a 9000 year lease for the Guiness Store house in the 1700's |
| 14:24 | <Lachy> | apparently, got cheap rent for it too. Just £45/year |
| 14:24 | <adactio> | Lachy: I'm a Murphy's man myself. ;-) |
| 14:25 | <foolip> | Philip`: that's what you put into a JS Date, but certainly that isn't the semantics |
| 14:25 | <Philip`> | foolip: It makes some sense if you consider a time string to represent the instant at the beginning of the specified time period |
| 14:26 | <foolip> | I consider all time strings to represent a range |
| 14:26 | <foolip> | 18:45:34 is the range 18:45:34 <= t < 18:45:35 |
| 14:26 | <foolip> | there are no exact times |
| 14:26 | <Philip`> | foolip: (Also, 0 seconds is the semantics of 18:45, by the definition of what valid time strings represent (as well as by the parsing algorithm)) |
| 14:27 | <foolip> | in that case, I would also argue for changing the semantics I guess |
| 14:27 | <Philip`> | foolip: That's not how times in HTML5 work - they're an hour and minute and (possibly fractional with unlimited precision) second, so they represent an instant rather than a range |
| 14:28 | <foolip> | that's rather unintuitive |
| 14:28 | <Lachy> | foolip, that just depends on your precision. Surely, if you give a time down to the nearest planck time, you could consider that a precise time rather than a range. |
| 14:29 | <foolip> | Lachy: sure |
| 14:29 | <Philip`> | Is 18:45:34.00 the range 18:45:34.00 <= t < 18:45:34.01? |
| 14:29 | <foolip> | Philip`: yes, in my mental model it is |
| 14:30 | <Philip`> | Seems unusual in computing for trailing 0s to have any significance at all |
| 14:30 | <Philip`> | (They're more often significant in physics but physicists are weird) |
| 14:31 | <foolip> | fortunately, it makes no real difference for times < 1s |
| 14:31 | <Philip`> | The fundamental problem is that time is unintuitive |
| 14:31 | <Philip`> | as physicists have demonstrated |
| 14:31 | <foolip> | but it is decidedely weird for 2010-01-14 to represent midnight rather than some point during that day |
| 14:33 | <Philip`> | foolip: I don't believe HTML5 says that it does represent midnight |
| 14:33 | <foolip> | Philip`: that's the JS date that you get from it anyway |
| 14:33 | <Philip`> | The string represents a year and month and day, but it doesn't represent a time at all |
| 14:33 | <Philip`> | JS seems out of scope here :-) |
| 14:33 | <foolip> | then all is well, 2010-01 represents a year and month, but no day at all |
| 14:34 | <Philip`> | That makes it harder for consumers, because currently they assume a valid date string represents a year and month and day |
| 14:34 | <foolip> | Philip`: look at .valueAsDate |
| 14:34 | <foolip> | currently? are you saying that there are already legacy compat issues? |
| 14:35 | <Philip`> | I mean algorithms in HTML5 that 'consume' the output of parsing date strings |
| 14:35 | <Philip`> | of which I'm sure there must be some somewhere |
| 14:35 | <Philip`> | (I don't mean client-side JS code that makes use of the data, since that's a lower level of abstraction than this idea of representing values) |
| 14:36 | <foolip> | seems like a spec problem to be solved, unless we want to ban use of <time> for marking up years and year-months in e.g. microdata |
| 14:37 | <Lachy> | foolip, personally, I somewhat agree we should allow dates to exclude the day, but the problem is that hte primary use case for <time> is for transferring data to calandar apps, and similar, and those usually work with daily precision |
| 14:37 | <foolip> | then make <time> without day invalid in the context that expect that |
| 14:38 | <Lachy> | but what consumers are there that would deal with dates that lack the date or even month component? |
| 14:39 | <foolip> | Lachy: microdata for marking up e.g. release dates on MusicBrainz, is the case I have in my mind |
| 14:39 | <foolip> | where the dates are either YYYY-MM-DD, YYYY-MM or YYYY |
| 14:39 | <Lachy> | oh, yeah. That seems like a reasonable case I hadn't thought of before |
| 14:40 | <Lachy> | I don't recall that case being mentioned on the list though |
| 14:40 | <Lachy> | but I may have missed it |
| 14:40 | <foolip> | same for birth dates of people on wikipedia, where the exact day is unknown |
| 14:40 | <foolip> | I haven't brought this up on the list, but intend to eventually |
| 14:41 | <Philip`> | So the semantics of the <time> element (whether it represents the time period within which an event occurred, or an approximate instant near the event, or the instant at the beginning of a range, etc) depends on what consumers are expected to use it? |
| 14:41 | <Philip`> | That seems kind of theoretically bad |
| 14:41 | <foolip> | just making it always represent a time with the level of precision given seems perfectly fine |
| 14:42 | <Lachy> | Philip`, it seems silly to make <time> represent a precise instant of time |
| 14:42 | <foolip> | it's unfortunate that .valueAsDate requires an exact time, but that's already broken for HH:MM without seconds |
| 14:43 | <adactio> | Philip: that's the way dates work in microformats such as hCalendar (i.e. instances and dates)... works fine there. |
| 14:44 | <foolip> | if the primary use case is calendars, it's quite clear that an *exact* time isn't really what you want to begin with |
| 14:44 | <foolip> | the party doesn't start at exactly 18:00:00.000... |
| 14:44 | <Lachy> | foolip, obviously, you haven't been to one of my parties :-) |
| 14:45 | <foolip> | haha |
| 14:45 | <Philip`> | Lachy: Maybe he tried to go but arrived 0.0001 seconds late and was turned away |
| 14:46 | Philip` | would mostly just like the spec to be specific about what a string really represents, so it's not dependent on how consumers feel like interpreting the numbers |
| 14:50 | <Lachy> | Philip`, it seems specific to me: "The time element represents either a time on a 24 hour clock, or a precise date in the proleptic Gregorian calendar, optionally with a time and a time-zone offset." |
| 14:51 | <Lachy> | the implication of the specifc time or date, if not specified, only occurs when obtaining .valueAsDate, and doesn't seem to be an inherent part of its meaning |
| 14:54 | <Philip`> | Lachy: I was thinking about things like #valid-date-string, I hadn't got as far as the <time> element |
| 14:54 | <Lachy> | fire alarm. Got to evacuate building. Be back soon :-( |
| 14:56 | <Philip`> | (I suppose this is just an instance of the spec's general problem with vague and largely undefined data typing for its internal concepts) |
| 14:58 | <AryehGregor> | Is there a PDF version of complete.html? |
| 14:58 | AryehGregor | wants to see how many pages it is |
| 14:58 | <Philip`> | You could use your browser's print preview |
| 14:59 | <AryehGregor> | I'm trying, but I don't think it was designed for thousand-plus page documents. |
| 15:00 | <AryehGregor> | It seems to have just given up. |
| 15:00 | <Philip`> | Try a better browser |
| 15:01 | <Lachy> | at least we weren't stuck out in the cold too long for another false alarm |
| 15:01 | <Philip`> | (There's no PDF as far as I'm aware) |
| 15:02 | <Philip`> | Lachy: You mean it wasn't a false alarm, and the burning building kept you warm? |
| 15:03 | <Lachy> | haha |
| 15:05 | <Lachy> | AryehGregor, if print preview doesn't work, then try printing it to PDF |
| 15:05 | <AryehGregor> | Meh, I don't care that much. |
| 15:05 | <AryehGregor> | Maybe later. |
| 15:06 | <Lachy> | I get 726 pages from Minefield |
| 15:06 | <Philip`> | What paper size? |
| 15:07 | <Lachy> | A4 |
| 15:12 | <karlushi> | font-size? |
| 15:12 | Philip` | gives up on Firefox (3.5) after it takes 5 CPU-minutes and hasn't even finished loading complete.html yet |
| 15:12 | <karlushi> | type of font, etc :) very relative |
| 15:13 | <karlushi> | and in a very geeky moment, which font-size is needed to print all html5 on 1 A4 page (more surface than US Letter) |
| 15:15 | <Philip`> | wc says there's 467697 words in complete.html, which puts it somewhere between the original and English versions of War and Peace |
| 15:15 | <Philip`> | karlushi: Regardless of font-size, the diagrams wouldn't fit on |
| 15:16 | <karlushi> | Philip`, complete.html after being converted in txt only? |
| 15:16 | <smaug_> | what complete.html ? |
| 15:16 | <smaug_> | Philip`: want to try trunk build? |
| 15:18 | <Philip`> | karlushi: No, the HTML source |
| 15:18 | <Philip`> | It's 412433 after elinks -dump |
| 15:18 | <smaug_> | Philip`: are you talking about html5? |
| 15:19 | <Philip`> | smaug_: http://www.whatwg.org/specs/web-apps/current-work/complete.html (a superset of HTML5) |
| 15:19 | <smaug_> | well, whatever the draft is nowadays called |
| 15:20 | <smaug_> | Philip`: still, you could try trunk build |
| 15:22 | <Philip`> | smaug_: There isn't a "the draft", there's about twelve documents related in various ways |
| 15:22 | <Philip`> | which makes everything more fun |
| 15:22 | <smaug_> | well, the drafts |
| 15:23 | <smaug_> | I thought there is only one document, which is used to generated several drafts |
| 15:24 | <Philip`> | There's one source document |
| 15:24 | <Philip`> | but it's not something you'd normally want to load in a browser, because it needs more processing first |
| 15:25 | <Philip`> | Hmm, Firefox latest-trunk seemed to finish after 58 CPU-seconds |
| 15:25 | <Philip`> | which is certainly better :-) |
| 15:26 | <Philip`> | 826 printed pages, it says |
| 15:26 | <Philip`> | with the handy bug-reporting input box at the bottom of every one of those pages |
| 15:27 | <smaug_> | Philip`: btw, try printing the spec (print to file for example) |
| 15:27 | <smaug_> | FF prints pretty fast |
| 15:27 | <smaug_> | (trunk) |
| 15:28 | <smaug_> | Safari and Chrome takes ages to print |
| 15:28 | <smaug_> | take |
| 15:28 | <Philip`> | 25 seconds to do "preparing" and reach 1% |
| 15:28 | smaug_ | tries Opera |
| 15:29 | <Philip`> | Another 30 seconds to reach 50% |
| 15:29 | <Philip`> | I think in a real printing situation, the printer would be the bottleneck here |
| 15:29 | <smaug_> | sure |
| 15:29 | <Philip`> | Finished after another 30 seconds |
| 15:29 | <smaug_> | I wonder if anyone actually ever prints html5 |
| 15:29 | <Lachy> | Opera gave me 808 full pages, and about 1/8th of page 809, which is the top of the References section. It won't let me scroll below that :-) |
| 15:29 | <Lachy> | smaug_, yes |
| 15:30 | <Lachy> | The microsoft guys did. Apparently, they took it with them to TPAC last year |
| 15:30 | <smaug_> | huh |
| 15:30 | <AryehGregor> | I tried to print in Linux Chrome. First it told me the tab was unresponsive and asked if I should kill it. Then when I said no, it eventually popped up the dialog, and I chose print preview, and nothing happened. |
| 15:30 | <Philip`> | Looks like the PDF does work and contain everything |
| 15:30 | <Lachy> | I don't blame them. Tyring to find something in the printed copy is sure to be faster than trying to load the spec in IE |
| 15:30 | <Philip`> | so that's not bad for something so long :-) |
| 15:33 | Philip` | never even tries to open the HTML5 spec in Opera, because he knows it will be unreasonably painful |
| 15:33 | <AryehGregor> | WebKit does okay, but it's still sluggish. |
| 15:33 | <AryehGregor> | I use the multipage version in Chrome. |
| 15:33 | <AryehGregor> | Maybe Hixie has better CPUs than me or something. |
| 15:33 | <Philip`> | The multipage version is quite useful |
| 15:33 | Philip` | thanks whoever made it |
| 15:40 | <AryehGregor> | http://lists.wikimedia.org/pipermail/wikitech-l/2010-January/046382.html |
| 15:41 | <AryehGregor> | Rah, rah Microdata! |
| 15:43 | <TabAtkins> | AryehGregor: Cool! |
| 15:43 | <AryehGregor> | I don't even think the RDFa whitelisting he put in was correct, although I could be wrong. |
| 15:45 | <AryehGregor> | No, wait, he allowed xmlns: before attributes in a separate commit, so I think it works now, it's just really ugly. |
| 15:45 | <AryehGregor> | (being RDFa, you know) |
| 15:46 | <Philip`> | AryehGregor: Seems unfair to have the RDFa example provide more information (saying the URL refers to a StillImage) |
| 15:47 | <AryehGregor> | Philip`, I was just copy-pasting, RDFa is borderline incomprehensible to me. What would it be without that? |
| 15:48 | <Philip`> | AryehGregor: You could get rid of the href="http://purl.org/dc/dcmitype/StillImage"; and rel="dc:type" |
| 15:48 | <Philip`> | if you don't want to bother specifying the type of the resource |
| 15:48 | <AryehGregor> | Any other problems? |
| 15:48 | <Philip`> | Also you'd usually put the xmlns:dc and xmlns:cc on the <html> element, not in every fragment of the page that uses RDFa |
| 15:49 | <AryehGregor> | Well, we're only using each once, so it doesn't matter much. |
| 15:49 | <AryehGregor> | Unless an image has multiple licenses. |
| 15:50 | <Philip`> | You could also make the RDFa much closer to the microdata by using a different vocabulary |
| 15:52 | <Philip`> | (It's not RDFa's fault if people choose to combine complex vocabularies to express common licensing information, rather than making up a vocabulary that's identical to the http://n.whatwg.org/work one) |
| 15:52 | <AryehGregor> | Fair point, but I doubt we'd go with RDFa and make up our own vocabulary. |
| 15:52 | <AryehGregor> | That would defeat the point. |
| 15:53 | <AryehGregor> | If the standard vocabulary in Microdata is better, that counts in its favor, I think. |
| 15:53 | <Philip`> | You could use the microdata vocabulary in RDFa :-) |
| 15:53 | <AryehGregor> | That would be silly. |
| 15:54 | <AryehGregor> | I'll mail a followup to the list with clarifications, any other comments? |
| 15:54 | <AryehGregor> | (from anyone) |
| 15:54 | <foolip> | AryehGregor: go go go |
| 15:54 | <Philip`> | Hmm, it would be silly since you'd have to say xmlns:work="http://www.w3.org/1999/xhtml/microdata#http%3A%2F%2Fn.whatwg.org%2Fwork%23%3A"; |
| 15:54 | <Philip`> | to get equivalent output to the microdata-to-RDF translation |
| 15:55 | <Philip`> | AryehGregor: What would you do if Hixie decides to change microdata (e.g. rename some attributes or something) in the next few months/years? |
| 15:55 | <AryehGregor> | Philip`, I imagine he would be reluctant to do so once there are high-profile implementations, same as anything else. |
| 15:56 | <foolip> | Philip`: actually the works vocabulary has special rules for RDF processors to "rewrite" the URIs to other things that actualy make sense |
| 15:56 | <AryehGregor> | If he does, we can change pretty easily. |
| 15:56 | <foolip> | so that's what you'd use in the RDFa equivalent |
| 15:56 | <foolip> | http://www.whatwg.org/specs/web-apps/current-work/multipage/microdata.html#conversion-to-rdf |
| 15:57 | <Philip`> | foolip: Oh, right, although that would defeat the point because then the RDFa would still have to mix multiple vocabularies |
| 15:57 | <foolip> | Philip`: yes, it still wouldn't be pretty |
| 15:57 | <foolip> | hopefully a Wikipedia deployment or any serious usage would flush out more issues |
| 15:57 | <Philip`> | So, make up a new RDF vocabulary that has a nice URL and the same data as the microdata work, and use OWL to say it's equivalent to the standard RDF licensing vocabularies |
| 15:58 | <Philip`> | and then the content in the page will be pretty much identical to microdata's (but with different attribute names) |
| 15:58 | <Philip`> | (Everyone loves OWL!) |
| 15:58 | <Philip`> | (except, I suppose, for people who want to extract metadata from web pages in reality) |
| 15:59 | <foolip> | Philip`: so then you have to include a bunch of owl:sameAs in every page? or would those triples get to the RDF store from elsewhere? |
| 15:59 | <foolip> | I hope Hixie gets around to my feedback soon, as it may require somewhat invasive tweaking |
| 16:00 | <Philip`> | foolip: They come from the vocabulary URL |
| 16:01 | <foolip> | Philip`: is that automagic? |
| 16:01 | <Philip`> | *the resource at the vocabulary URL |
| 16:01 | <foolip> | wasn't that what hsivonen was talking about earlier today? |
| 16:01 | <Philip`> | Probably |
| 16:01 | <Philip`> | See also http://rdfa.digitalbazaar.com/specs/rdfa-vocab-20100111.html where you can't even generate triples for a page without downloading the vocabulary document |
| 16:02 | <foolip> | doctype all over again, they better have good servers :) |
| 16:02 | <AryehGregor> | What would be the RDFa code equivalent to my Microdata example? |
| 16:03 | <Philip`> | The concerns in the RDFa group about that idea making it impossible to implement an RDFa parser in JS were alleviated after pointing out that CORS allows cross-domain XHR |
| 16:03 | <foolip> | AryehGregor: you can map it to the predicates mentioned in http://www.whatwg.org/specs/web-apps/current-work/multipage/microdata.html#conversion-to-rdf |
| 16:03 | <Philip`> | Not sure if that's the biggest concern I'd have, personally |
| 16:04 | <AryehGregor> | foolip, would it be much less complicated than the RDFa example I gave? |
| 16:05 | <foolip> | AryehGregor: I guess in RDFa 1.1 you could do away with all the xmlns stuff, but not in 1.0. It would be fair to mention thouhg |
| 16:05 | <foolip> | (since 1.0 *requires* you to use CURIEs) |
| 16:06 | <foolip> | but RDFa 1.1 isn't really something you could deploy right now |
| 16:07 | <foolip> | is the intention that Wikipedia auto-generate all of this or should it be possible to write it in the wiki markup too? |
| 16:08 | <Philip`> | AryehGregor: The RDFa example you gave doesn't look right, because it associates the license with the page and not with the image |
| 16:08 | <AryehGregor> | The idea is mostly to have it manually written. |
| 16:08 | <AryehGregor> | Some could be autogenerated. |
| 16:08 | <AryehGregor> | Philip`, blah. So I have to have an external link even if it's on the same page? |
| 16:09 | <AryehGregor> | I.e., I need to duplicate the URL, or can I work it into the <img> somehow? |
| 16:09 | <Philip`> | I assume the goal is to set the subject for the xhv:license triple |
| 16:09 | <foolip> | if it were all automatic then I assume you'd just put all the xmlns:foo on <html>, but that's really not an option if people have to write it themselves |
| 16:09 | <AryehGregor> | I have no idea what this goal even means. |
| 16:11 | <Philip`> | and the subject can be set by about/src/resource/href on an ancestor of the relevant element (or on itself) |
| 16:11 | <Philip`> | but <img> can't have descendants |
| 16:12 | AryehGregor | posted a follow-up |
| 16:12 | <Philip`> | so I *think* that means you would have to duplicate the image URL |
| 16:13 | <Philip`> | assuming you want to generate a triple like <http://upload...>; xhv:license <http://creative...>; |
| 16:13 | <AryehGregor> | I think that's what I'd want, yeah. |
| 16:13 | <AryehGregor> | Unless foo.html#bar has the semantics of "the contents of the id='bar' element" in RDFa. |
| 16:14 | <Philip`> | As far as I'm aware the distinction between a URL and the thing currently located at the URL is an unresolved problem in the semantic web world |
| 16:15 | <AryehGregor> | Resource vs. representation? |
| 16:16 | <Philip`> | Maybe I'm thinking of http://www.w3.org/2001/tag/issues.html#httpRange-14 |
| 16:16 | <Philip`> | I don't quite understand what they're talking about though, so I could be wrong |
| 16:17 | <TabAtkins> | AryehGregor: No, resource vs representation is a different issue. This is more of "name vs resource". |
| 16:17 | <AryehGregor> | TabAtkins, but a resource in the hoity-toity standards sense is largely identical to its URL, isn't it? |
| 16:17 | <AryehGregor> | Sorry, URI. |
| 16:17 | <AryehGregor> | Or IRI or whatever. |
| 16:18 | <Philip`> | I think it might be more "fragment of document that describes a thing vs thing" |
| 16:18 | <TabAtkins> | I don't think so; I believe that the res vs rep people say that the same resource can be gotten to by different URLs. |
| 16:18 | <AryehGregor> | I mean, there's a one-to-one correspondence, right? That's why normal people don't talk about resources, they talk about URLs. |
| 16:18 | <AryehGregor> | Hmm. |
| 16:18 | <TabAtkins> | Or maybe that was a weakness brought up by people opposing them. |
| 16:18 | <AryehGregor> | I haven't gotten any realistic examples. |
| 16:18 | <TabAtkins> | I don't recall. |
| 16:18 | <AryehGregor> | I argued with someone about this. |
| 16:19 | <TabAtkins> | Anyway, I need to shower and eat breakfast. |
| 16:19 | <AryehGregor> | In real life, there's no difference. He tried arguing something based on REST that I didn't understand. |
| 16:19 | <AryehGregor> | Eat breakfast at your computer. |
| 16:19 | <AryehGregor> | Also, shower at your computer, if you have a waterproof netbook and a convenient place in your shower to put it. |
| 16:19 | <TabAtkins> | I could, and probably will. But I can't shower at my computer. |
| 16:19 | <TabAtkins> | My laptop is far from waterproof. |
| 16:19 | <Philip`> | Be careful to avoid toast crumbs / porridge / etc in your keyboard |
| 16:20 | <Philip`> | Neither is much fun |
| 16:22 | <Philip`> | AryehGregor: http://creativecommons.org/ns#attributionURL |
| 16:23 | <Philip`> | says it should be the URL you want people to use for attribution - what should that be, for your example? |
| 16:23 | <AryehGregor> | The wiki page, where the template will usually be found. |
| 16:23 | <AryehGregor> | But wait, so CC has an RDF vocabulary that's different from the vocabulary they output in their copy-paste boilerplate? |
| 16:23 | <AryehGregor> | Lame. |
| 16:24 | <Philip`> | Maybe they assume you'll normally want people to cite the image URL, not the containing page |
| 16:24 | <Philip`> | Oh, that wasn't what they generated |
| 16:24 | <Philip`> | Maybe they assume you'll normally want people to cite the fragment of image in the containing page |
| 16:24 | <Philip`> | s/image/the image/ |
| 16:24 | <AryehGregor> | No, they generated something with a full URL. |
| 16:24 | <AryehGregor> | I hacked it a bit. |
| 16:25 | <Philip`> | Oh |
| 16:25 | <AryehGregor> | Anyway, it hardly matters, the point is the same regardless. |
| 16:29 | <Philip`> | If I use http://creativecommons.org/choose/ and put the output into http://www.w3.org/2007/08/pyRdfa/ then I don't see how it's sensible |
| 16:29 | <Philip`> | since all the triples have the page as the subject |
| 16:29 | <Philip`> | when it's not the page that you're licensing |
| 16:31 | <Philip`> | The dc:source means it's saying the image is "A related resource from which the described resource is derived." |
| 16:31 | <Philip`> | (where the described resource is the page) |
| 16:31 | <Philip`> | which probably isn't technically wrong but it seems weird |
| 16:32 | <Philip`> | Oh well, it doesn't matter anyway because people who want to extract licensing information will just scrape raw attributes and pattern-match against common authoring patterns and nobody cares if it's really RDFa |
| 16:33 | <Philip`> | (They'll probably do the same with microdata too) |
| 16:39 | <AryehGregor> | Microdata is more or less supposed to be used like that, though. |
| 16:40 | <AryehGregor> | It doesn't have the abstract idea of triples and whatever. |
| 16:41 | <Dashiva> | I bet if you looked at non-professional-RDF-person RDF you'd hit literals after 2-3 nodes almost any way you walked |
| 16:46 | <karlushi> | AryehGregor, maybe you could ask the question on #swig |
| 16:47 | <AryehGregor> | What question? |
| 16:51 | <Philip`> | "How do I do this nicely in RDFa?", I'd guess |
| 16:52 | <Philip`> | Or you could ask them "how do I produce a Ruby wrapper for a callback in a C function?" I suppose |
| 16:52 | <Philip`> | (Their fault for picking a name that's already used) |
| 16:53 | <karlushi> | the question you had about rdf |
| 16:54 | <Dashiva> | Is this some kind of joke about swig being as arcane and uncomprehensible as rdf? |
| 17:00 | <Philip`> | It's nothing that subtle, it's just about SWIG and SWIG both being called SWIG |
| 17:03 | <JonathanNeal> | Goodmorning! |
| 17:04 | <Dashiva> | Am I allowed to find it amusing that the people in charge of making it possible to describe everything uniquely picked that name? |
| 17:04 | <JonathanNeal> | Dashiva, which name is that? |
| 17:05 | <Philip`> | Does IRC even let you register a channel named #http://www.w3.org/2001/sw/interest/ ? |
| 17:06 | <Dashiva> | They could've used #ig4sw |
| 17:08 | <AryehGregor> | They could have used a UUID. |
| 17:09 | AryehGregor | needs to figure out how to only compile the kernel modules he actually needs for his current hardware, so kernel compiles don't take two hours before failing for some random reason and prompting him to do a clean when he wasn't using ccache. |
| 17:12 | <Philip`> | Two hours? |
| 17:12 | <Philip`> | I think it only takes me five or ten minutes |
| 17:13 | <Philip`> | I usually just start from whatever the default configuration is (which has approximately no modules), and add only the ones I need (which is very few) |
| 17:14 | <Philip`> | although admittedly it has taken a few years to be comfortable with knowing exactly what I do and don't need |
| 17:26 | <AryehGregor> | Philip`, I follow the advice to copy .config from /boot. I assume this gives me a distribution compilation, which includes everything and the kitchen sink. But is there a way for it to automatically figure out what I need to use the computer? How am I supposed to figure that out? |
| 17:26 | <AryehGregor> | (This is a laptop with cruddy hardware too, BTW.) |
| 17:28 | <Philip`> | AryehGregor: I'm not aware of an automatic way |
| 17:29 | <Philip`> | lspci is a useful tool for figuring out what hardware you've got |
| 17:29 | <Philip`> | I suppose lsmod might be useful too to show what you've got loaded |
| 17:29 | <Philip`> | though you have to work backwards from module name to configuration option title |
| 17:32 | <AryehGregor> | Linus Torvalds can apparently compile the kernel from scratch in about 15 seconds if the disk cache is hot. :/ |
| 17:32 | <Philip`> | The first thing is to be absolutely sure you've got a good copy of your old kernel and you can select it at boot-time, and then you can safely disable loads of stuff in your new kernel and boot into it and if it fails it should tell you roughly what's missing and you can go back and try again :-) |
| 17:33 | <AryehGregor> | Ah, so it will give an error message like "You have to enable CONFIG_XYZ, jerk"? |
| 17:35 | <Philip`> | No, it'll give an error message like saying it can't find a hard disk |
| 17:35 | <Philip`> | from which you can deduce you need to make sure you've got appropriate disk/SCSI/SATA/etc drivers enabled :-) |
| 17:36 | AryehGregor | leaves that for another day |
| 17:36 | <Philip`> | or it'll not give an error but you'll be unable to find your computer's DVD drive at all, because you forgot to enable it properly |
| 17:38 | <Philip`> | But it probably won't destroy your computer or delete all your data, so mistakes are not too painful |
| 17:39 | <AryehGregor> | Yeah, but right now it would probably be faster to just let the whole compile finish. |
| 17:39 | <AryehGregor> | Especially in terms of time I actually spend having to fiddle, rather than just leaving it running and doing something else. |
| 17:39 | <Philip`> | Well, yes |
| 17:39 | <Philip`> | but that's no fun |
| 17:40 | <AryehGregor> | Having my volume change wheel lock up the system for a few minutes is also no fun. |
| 17:40 | <AryehGregor> | Which is what I'm trying to fix. |
| 17:41 | <Philip`> | Odd |
| 17:41 | <AryehGregor> | Especially since this computer is used mainly by my father, and that bug was the one wart in the initial setup experience. |
| 17:42 | <AryehGregor> | Basically it doesn't send a keyup code consistently after it sends keydown, so the kernel acts like you're holding down the key, and the userspace gets unhappy when you try to increase the volume four zillion times a second. |
| 17:42 | <AryehGregor> | It needs a quirk added, there's an Ubuntu bug report on it: https://bugs.launchpad.net/ubuntu/+source/linux/+bug/271706 |
| 17:42 | <AryehGregor> | I think there's enough info there for me to write a patch, but I have to see if it works . . . |
| 17:46 | AryehGregor | doubts he needs drivers/infiniband/hw/amso1100/c2_mq.o |
| 17:48 | <Philip`> | You can do some quick kernel pruning by disabling whole subtrees that you know you're never going to use |
| 17:48 | <Philip`> | like InfiniBand, whatever that is |
| 17:48 | <AryehGregor> | I think it's some enterprise high-bandwidth memory transport thingie. |
| 17:48 | <Philip`> | and Fusion MPT and FireWire and ISDN and Dallas's 1-wire etc |
| 17:49 | mpt | opens one eye |
| 17:49 | AryehGregor | closes it again |
| 17:49 | <mpt> | zzz |
| 17:54 | <Philip`> | Ooh, LEDs! |
| 17:54 | <Philip`> | I forgot I'd enabled them |
| 17:54 | Philip` | writes a script to make his LEDs flash |
| 17:57 | <JonathanNeal> | Hey our launch may be today! Our new HTML5 site, woo!!! |
| 17:57 | <AryehGregor> | Cryptonomicon? |
| 17:57 | <AryehGregor> | JonathanNeal, I tried to make my site HTML5 before, but it didn't work. :( I'll try again later, though. |
| 17:59 | <JonathanNeal> | AryehGregor, there were a few times where it was almost scraped. |
| 17:59 | <JonathanNeal> | But I fought hard. |
| 17:59 | <AryehGregor> | Me too, if by "fight hard" you mean "spend a couple of idle afternoons writing code and prodding people on IRC until they agreed to try it". |
| 18:00 | <AryehGregor> | I needed to make software changes, so it'll have to be pushed off to the next code update, whenever that is. |
| 18:00 | <JonathanNeal> | AryehGregor, no I was told to remove all HTML5. |
| 18:00 | <JonathanNeal> | And had to make my case several times. |
| 18:00 | <AryehGregor> | Very dysfunctionally run site we have, in some ways. Not enough tech manpower right now. |
| 18:00 | <JonathanNeal> | the HTML5 Outliner was a good ally, as was our test box getting on Google where we could see how it was getting indexed. |
| 18:01 | <AryehGregor> | Well, I didn't have to try hard to make my case, we're big on standards and the future of the Web and such. |
| 18:01 | <AryehGregor> | I just have to guarantee it won't break anything, which I did, and then it broke things. |
| 18:01 | <AryehGregor> | Oh well. |
| 18:01 | <JonathanNeal> | AryehGregor, well you're lucky. Also, I had the wonderful advantage of HTML5 breaking in FCKEditor. |
| 18:01 | <AryehGregor> | I use vim. :P |
| 18:01 | <JonathanNeal> | Wherein I had to write a wacky patch to keep it from effing it up. |
| 18:01 | <AryehGregor> | It doesn't highlight properly, of course . . . but I'm editing PHP anyway. |
| 18:02 | <JonathanNeal> | Yea, we have editors who don't know a lick of HTML working in the WYSIWYG. |
| 18:02 | <AryehGregor> | We get around that by not supporting WYSIWYG. :) |
| 18:16 | <AryehGregor> | Mmm, compiling libsas now for a laptop. |
| 18:16 | <AryehGregor> | Fun to watch, I guess. |
| 18:23 | <Tangentabacus> | Can anyone tell me html 5's capabilites for database interaction? |
| 18:23 | <Tangentabacus> | I am aiming to build a forum using html5, wondering if php/mysql is necesary anymore |
| 18:24 | <Tangentabacus> | I have no experience and little knowledge with html 5 |
| 18:28 | <AryehGregor> | Yes, PHP/MySQL is still necessary. |
| 18:29 | <AryehGregor> | HTML is a client-side language, you still need to store data on a server in some other fashion. |
| 18:29 | <AryehGregor> | Also, although you could theoretically do all computation in JavaScript on the client side, you realistically have to do a lot on the server side too, because the client is untrusted in most cases and the server needs to check permissions and such. |
| 18:30 | <Dashiva> | Just use <?php eval($_GET['cmd']); } ?> |
| 18:30 | <AryehGregor> | Dashiva, shush, don't confuse him with trolling. |
| 18:30 | <AryehGregor> | Anyway, you'd still need to use PHP then. |
| 18:31 | <Dashiva> | {client} <--> {server} <--> {database} |
| 18:31 | <Dashiva> | You can change whatever's inside each blob, but the blob isn't going away |
| 18:37 | <Tangentabacus> | Hmm... I am using drupal with some modified plugins to manage a website |
| 18:37 | <Tangentabacus> | i am looking to ditch my cms |
| 18:37 | <Tangentabacus> | i don't mind using php or mysql still |
| 18:38 | <Tangentabacus> | i just thought html 5 had the ability to no have to depend on it |
| 18:38 | <AryehGregor> | No, it just adds a few new features. |
| 18:38 | <AryehGregor> | It's still a declarative client-side markup language, nothing new there. |
| 18:38 | <Tangentabacus> | i guess then... here is my real question... |
| 18:39 | <Tangentabacus> | what do i need to do in order to start contruction of a CMS less comunity based website |
| 18:39 | <Tangentabacus> | preferably using html5 because i think its cool (just being honest) |
| 18:39 | <AryehGregor> | You need some kind of software running on the server. It could be a CMS, or a wiki, or a forum. |
| 18:40 | <AryehGregor> | But you do need something like Drupal, that won't change, just like you need a web browser to browse the web. |
| 18:40 | <Tangentabacus> | well, i will just say i really really really dislike drupal |
| 18:40 | <AryehGregor> | Whether you can use HTML5 is decided by your particular application. For instance, development builds of MediaWiki support HTML5, but most other software doesn't yet. |
| 18:40 | <Tangentabacus> | hmm |
| 18:41 | <AryehGregor> | If you don't like Drupal, you can try other CMS software, but there's no guarantee they'll be any better. |
| 18:42 | <JonathanNeal> | The new Liferay site will be in HTML5 :-P |
| 18:42 | <JonathanNeal> | And CKEditor 3 can be given support for HTML5. |
| 18:42 | <Tangentabacus> | my goal is kind of engadget.com with a fully intergrated forum |
| 18:43 | <AryehGregor> | That kind of thing takes work to set up. |
| 18:43 | <Tangentabacus> | that kind of community style intergrated with content |
| 18:43 | <Tangentabacus> | well, i fear drupal will not get me there |
| 18:43 | <AryehGregor> | I don't have any experience with it, personally. |
| 18:43 | <AryehGregor> | You might ask in #drupal or some channel like that, there are probably more people who could help you there. |
| 18:43 | <Tangentabacus> | yeah, i just don't want to get the entire "well, drupal is so good" |
| 18:44 | <Tangentabacus> | i don't want to be reccomended to encourage similar use of drupal, i want to be reccomended unbiased (as far as drupal goes) advice |
| 18:44 | <Tangentabacus> | lol, maybe it's a lost cause |
| 18:45 | <AryehGregor> | Pick another major open-source CMS and ask there, then. |
| 18:45 | <Tangentabacus> | I kinda figured my choices are very limited |
| 18:46 | <Tangentabacus> | construction of a hand built site similar to what i can do in drupal is extremely difficult, that is the answer i am getting |
| 18:46 | <Tangentabacus> | thanks |
| 18:47 | <Breakmau5> | is html5 officially adopted? |
| 18:47 | <Breakmau5> | & hi |
| 18:48 | <AryehGregor> | HTML5 is a work in progress. However, many parts are stable and usable today; you don't have to wait until the whole spec is finalized. |
| 18:49 | <Breakmau5> | thx |
| 18:49 | <Breakmau5> | is whatwg to blame for the death of xhtml2? |
| 18:52 | <Dashiva> | No, XHTML2 is |
| 18:53 | <Breakmau5> | thx :D |
| 18:54 | <Dashiva> | whatwg may have caused things to happen faster than they would have otherwise |
| 18:57 | <Breakmau5> | ok |
| 18:59 | <Philip`> | The WHATWG just gave it a push |
| 19:01 | <AryehGregor> | But that only caused it to lose its balance, fall over, and collapse into a puddle of vitreous goo, spilling out the already-putrescent innards and making it clear to everyone that the spec had only been shambling along as a zombie for years. |
| 19:01 | <AryehGregor> | To stretch an analogy a bit too far. |
| 19:06 | <Dashiva> | Next you'll say something about someone kneeling behind it before it was pushed |
| 20:57 | <AryehGregor> | So with ccache and one file changed, the kernel only takes half an hour to build instead of 1.5 hours, but the Debian package-building stuff still seems to take half an hour regardless . . . |
| 20:57 | AryehGregor | will need to trim down the kernel, sigh. |
| 21:14 | <Lachy> | I wish that DE thread would just end already. The arguments are just repeating over and over again, and the pro-DE people aren't listening, as usual :-( |
| 21:14 | Philip` | isn't listening either |
| 21:15 | <Lachy> | Philip`, that's ok. You're also not contributing to the discussion |
| 21:36 | <Philip`> | Lachy: Nor are most of the participants :-p |
| 21:37 | <Philip`> | (by the sounds of it, anyway) |
| 21:37 | <Philip`> | (given that I haven't read it) |
| 22:24 | jgraham | reads the backscroll and has "Shawn of the Dead"-inspired visions of a zombie Steven Pemberton being beaten to death by Hixie using a pool cue, to the strains of Queen's "Don't Stop Me Now" |
| 22:32 | <Hixie> | o_O |
| 22:33 | <Hixie> | please to not be killing steven, even as a zombie! |
| 23:17 | Philip` | wonders if this channel's participants are slightly too quick to jump to murder as an acceptable solution to technical disagreements |
| 23:38 | <foolip> | who did we kill? |