| 05:48 | Hixie | ponders various better extension mechanisms for microdata |
| 06:02 | <Hixie> | i suppose we could have a first-come-first-served list of identifiers |
| 06:03 | <Hixie> | but that wouldn't work well with the vcard vocabulary... |
| 06:03 | <Hixie> | maybe we should make each item only have one type? |
| 06:03 | <Hixie> | then the identifiers can be scoped to the type... |
| 06:08 | <Hixie> | hmm |
| 06:08 | <Hixie> | i wonder if there are good use cases for giving an item two types |
| 08:34 | jgraham | wonders what problem Hixie is trying to solve |
| 08:48 | <Hixie> | jgraham: making the extension mechanism better than com.example.bar |
| 08:54 | <jgraham> | Hixie: Define "beter" |
| 08:54 | <jgraham> | Like, what qualities are you tring to optimize? |
| 08:55 | <Hixie> | usability. |
| 08:55 | <Hixie> | so probably shortness of identifiers |
| 08:56 | <Hixie> | we're going to experiment and see if com.example.bar is simple enough |
| 08:56 | <Hixie> | (See my e-mail to the whatwg list recently) |
| 08:56 | <Hixie> | but to do that i need a variety of syntaxes |
| 08:58 | <jgraham> | Is that the specific thing that you are testing? Or is there more? Will you test prefix-based solutions (the disadvantages in that case might not come out in a short test) |
| 09:03 | <Hixie> | i think prefix-based systems have been shown to be unusable pretty convincingly already |
| 09:08 | othermaciej | is reading Jeni's blog post |
| 09:13 | <gavin_> | that's a good post |
| 09:14 | <gavin_> | he reply to shelley's comment, in particular |
| 09:39 | <Hixie> | odd, why did a bunch of IDs just change when i regenned? |
| 09:39 | <Hixie> | jgraham: any changes to anolis i should know about? |
| 09:40 | <jgraham> | Hixie: Er, I have been poking it a bit but nothing should have changed the ids |
| 09:45 | <Hixie> | how about this: we restrict items to one type per item, and then allow items to be one word long |
| 09:45 | <Hixie> | and those names are automatically scoped to the type |
| 09:45 | <Hixie> | kinda like what we do now with predefined vocabularies |
| 09:46 | <othermaciej> | I did not realize there was so much discussion of microdata at public-rdf-in-xhtml-tf |
| 10:23 | Philip` | isn't sure what Hixie means |
| 10:24 | <Philip`> | Did you mean "allow *itemprops* to be one word long" or something? |
| 10:24 | <Hixie> | er yes |
| 10:28 | <Philip`> | <section item=org.example.animal.cat><h1 itemprop=name>... etc, and anyone processing the document should detect 'name' properties in 'org.example.animal.cat' items, rather than detecting 'org.example.animal.cat' properties in 'org.example.animal.cat' items |
| 10:31 | <Philip`> | What happens when you want the equivalent of <section item=org.example.animal.cat><div item=vcard><h1 itemprop='org.example.animal.cat.name fn'>... with nested items? |
| 10:56 | <Hixie> | Philip`: do you mean there to be an itemprop on your vcard? |
| 10:57 | <Hixie> | Philip`: if you want to extend items whose type is vcard with other properties that aren't from vcard, then you use rev-dns or uris |
| 10:57 | <Hixie> | (for any value of "vcard") |
| 10:58 | <Hixie> | nn |
| 11:12 | <foolip> | Hixie: gsnedders has been telling me that you did some testing of IEs "special" treatment of prefixed elements/attributes (xmlns-style), are those tests available somewhere? |
| 11:12 | gsnedders | points out the "nn" to foolip |
| 11:12 | <foolip> | gsnedders: "nn"? |
| 11:13 | <gsnedders> | night night |
| 11:13 | <foolip> | ah, right |
| 11:13 | <foolip> | so I finished my giant post on the microdata stuff, wonder when is the best time to publish... |
| 11:13 | <foolip> | silly americans are sleeping now |
| 11:13 | <gsnedders> | When digg and /. have fallen quiet? |
| 11:14 | <gsnedders> | :P |
| 11:14 | <Philip`> | Hixie: Um, I probably do mean that |
| 11:15 | <foolip> | meh, there's a reason I didn't go into marketing: http://blog.foolip.org/2009/08/23/microformats-vs-rdfa-vs-microdata/ |
| 11:15 | <Philip`> | but I forget how microdata works |
| 11:15 | <gsnedders> | foolip: It's called "sanity". |
| 11:15 | <foolip> | haha, right you are |
| 12:14 | <jgraham> | foolip: If you had waited untill the best time we wouldn't have got to see it until Tuesday or so. So being no good at amrketing is nice for the rest of us |
| 12:15 | <foolip> | jgraham: Is that so? I'll just have to write something else for Tuesday then :) |
| 12:16 | <foolip> | jgraham: I'm thinking that converting the microdata tree to a DOM tree or JavaScript object might be a better way of visualizing/testing markup than the full JSON/RDF |
| 12:24 | <jgraham> | Someone should remister mantics.se for use in reverse DNS identifiers |
| 12:25 | <jgraham> | *register |
| 12:26 | <jgraham> | foolip: I'm not sure what context you have in mind |
| 12:26 | <jgraham> | Or what type of tree/object you would like |
| 12:26 | <foolip> | did you read my post? |
| 12:27 | <foolip> | the microdata example could be made into this XML: |
| 12:27 | <foolip> | <vcard fn="Philip Jägenstedt" url="http://foolip.org/">; |
| 12:28 | <foolip> | I didn't bother to mention that because it would probably be confusing though |
| 12:31 | <foolip> | both of mantic.se and mantics.se are free :) |
| 13:04 | <foolip> | jgraham: I just registered mantic.se :) Just 99 SEK for so much fun! |
| 13:07 | <jgraham> | Heh I would probably have done it for that price :) |
| 13:09 | <foolip> | too bad many TLDs require at least 3 characters in a domain name, so many nice ideas failed because of that |
| 13:10 | Philip` | wonders what a SEK is in real money |
| 13:11 | Philip` | looks it up |
| 13:11 | <Philip`> | Ah, right, that's not much :-) |
| 13:11 | <foolip> | http://lmgtfy.com/?q=99+SEK+in+USD :P |
| 13:11 | Philip` | was confused by the "99" which sounded like a lot |
| 13:12 | <foolip> | meh, I'm too slow |
| 13:14 | <Philip`> | I suppose it still seems like an incredibly expensive mechanism for supplying globally unique identifiers (which is all that's needed for microdata) |
| 13:15 | <foolip> | well, there aren't that many unique things to choose from. you have DNS and UUIDs and... then what? |
| 13:16 | <Philip`> | There's loads of ways |
| 13:16 | <Philip`> | TinyURL has been supplying unique identifiers for free for lots of years |
| 13:17 | jgraham | makes the pedantic note that only DNS+a date is really globally unique |
| 13:17 | <foolip> | jgraham: true |
| 13:17 | <foolip> | Philip`: tinyurl is DNS though |
| 13:18 | <Philip`> | You just need a server and a CGI script and a database, so people can register names without collisions |
| 13:18 | <Philip`> | foolip: Um... How so? |
| 13:18 | <foolip> | Philip`: because it relies on "tinyurl.com" for uniqueness |
| 13:18 | <jgraham> | tinyurl seems to combine the disadvantages of uuids with the disadvantages of dns |
| 13:18 | <foolip> | hehe |
| 13:19 | <foolip> | I'm not a fan of URL shortening services, completely opaque URLs isn't a lot of fun |
| 13:21 | <Philip`> | foolip: The identifier space provided by TinyURL is completely independent of DNS, and a spec could just say item=xxx refers to the identifier xxx as registered with TinyURL and there's no need for DNS |
| 13:22 | <Philip`> | just like lots of RFCs reference IANA registries, which has nothing to do with DNS (even though you typically use DNS if you want to read the registries) |
| 13:22 | <foolip> | well, you're right |
| 13:22 | <foolip> | but how would that work with microdata |
| 13:23 | <jgraham> | Philip`: I thought that tinyurl provided identifiers that were deterministic based on url |
| 13:23 | <jgraham> | But I guess they might not |
| 13:23 | <Philip`> | You can make things less opaque by letting people choose identifiers (avoiding collisions) rather than doing it randomly |
| 13:23 | jgraham | suggests using twitter usernames as identifiers |
| 13:24 | gsnedders | wonders how to identify jgraham then |
| 13:24 | <foolip> | I guess you could use DNS for the top-level item and something proprietary for the rest |
| 13:24 | <gsnedders> | jgraham, foolip: What are you doing inside on such a lovely sunny day? |
| 13:24 | gsnedders | heads out |
| 13:24 | <jgraham> | gsnedders: Same as you it seems |
| 13:24 | <foolip> | gsnedders, out? seriously? |
| 13:24 | <foolip> | well, back to killing zombies for me (resident evil) |
| 13:25 | <Philip`> | foolip: Instead of saying "Types are identified ... As reversed DNS labels", the spec could say "As identifiers registered on the WHATWG Registry (currently located at http://registry.whatwg.org/)"; |
| 13:26 | <Philip`> | and that registry can be a first-come-first-served collection of arbitrary Unicode strings |
| 13:26 | <foolip> | Philip`: meh, even I would object to that, imagine what the semwebbers would say |
| 13:26 | <jgraham> | Philip`: It could but it probably wouldn't satisfy anyone |
| 13:26 | <Philip`> | Why would it be worse than reversed DNS labels? |
| 13:27 | <jgraham> | because e.g. if microdata took off there would be a landgrab on all the sensible names |
| 13:27 | <Philip`> | (other than sys-admin issues of actually running the service) |
| 13:27 | <foolip> | at least DNS has (small) barrier to hogging all kinds of crufty ids without using them |
| 13:27 | <jgraham> | Like I could just register all strings < 10 characters |
| 13:28 | <Philip`> | Hmm, true |
| 13:28 | <gsnedders> | jgraham: What? Changing travel plans in September because you found out people you were meant to be meeting have had their plans changed under them by other organizations? |
| 13:28 | <jgraham> | gsnedders: Yes. |
| 13:28 | <gsnedders> | jgraham: Then I guess you are doing the same. |
| 13:28 | Philip` | was surprised when he realised Amazon EC2 bucket names are just arbitrary strings and you can register whatever you want, but in practice people just pick long names (because they don't need to type them frequently) |
| 13:28 | <jgraham> | Well at least if the executive summry of that was "wasting my life away" |
| 13:29 | <gsnedders> | No. |
| 13:29 | <jgraham> | Hmm |
| 13:30 | <Philip`> | On the subject of going outdoors and killing zombies: Plants vs Zombies is quite a nice game |
| 13:30 | <gsnedders> | Speaking of zombies, I should buy a copy of Pride and Prejudice and Zombies |
| 13:32 | <jgraham> | Not speaking of zombies I recentlt discovered that Leonard Nimoy is an accomplished photographer |
| 14:06 | <Philip`> | (Did I say EC2 buckets? I meant S3 buckets) |
| 14:15 | Philip` | sees http://www.gog.com/en/gamecard/the_settlers_2_gold_edition and remembers gsnedders having talked about it a lot |
| 14:15 | <gsnedders> | Philip`: Not the gold edition :P |
| 14:15 | <gsnedders> | Philip`: Wow. It actually works on 64-bit Windows _and_ on NT! |
| 14:16 | <gsnedders> | How the hell does it require a 1GHz CPU on a game from '96? |
| 14:17 | <Philip`> | Probably because it's emulating a machine from '96 |
| 14:17 | <gsnedders> | That was my guess. |
| 14:18 | <Philip`> | and because you're running XP/Vista in the background |
| 14:49 | Philip` | attempts to read Dive Into HTML 5 |
| 14:49 | <Philip`> | Is there some way I can disable web fonts in Opera? It's impossible to read like this because all the punctuation is wrongly positioned |
| 14:49 | <gsnedders> | Philip`: Do you have a bug on that? |
| 14:49 | gsnedders | ducks |
| 14:50 | <gsnedders> | It looks fine in O10 beta 3 here |
| 14:51 | Philip` | scrolls down a bit |
| 14:52 | <Philip`> | Hmm, great - it randomly draws the paragraph text in a monospace font, or in a serif font with the monospace font's spacing, depending on how I scroll |
| 14:52 | <gsnedders> | What platform? |
| 14:53 | <Philip`> | This build 4566 on Linux |
| 14:53 | <Philip`> | *is |
| 14:53 | <gsnedders> | Is 4566 the beta? :P |
| 14:53 | <gsnedders> | (Honestly, I can't remember build numbers.) |
| 14:54 | <Philip`> | No, it was just the latest one that existed when I looked yesterday |
| 14:54 | <gsnedders> | Ah. |
| 14:54 | <gsnedders> | Then I dunno. |
| 14:56 | Philip` | switches to Firefox 3.0, which works much better because it doesn't support web fonts |
| 18:55 | <Lachy_> | what's the point? http://doctype.be/402.html |
| 18:56 | <Dashiva> | If you love doctype-based validators? |
| 18:57 | <Philip`> | s/doctype/DTD/ |
| 18:57 | <Philip`> | (since something that has hard-coded doctype-dependent modes could still be considered a doctype-based validator, but wouldn't be helped by this) |
| 18:58 | <Philip`> | (i.e. I'm not being entirely pedantic, just largely) |
| 18:58 | <Dashiva> | I agree |
| 18:59 | <Philip`> | I suppose this is one of those DTDs where you're meant to ignore the validator's complaints about xmlns:* attributes |
| 19:00 | <Dashiva> | Validators always complain |
| 19:00 | <Dashiva> | Can't they say something nice for once? |
| 19:00 | <Philip`> | They do, they give you cute little tick-mark buttons when you've been good |
| 19:01 | <Dashiva> | Only if you're good |
| 19:01 | <Dashiva> | Would it hurt to say something like "Okay, you have two errors, but you clearly tried very hard and I appreciate that" |
| 19:05 | <Philip`> | We should remove the pass/fail distinction, and have validators report "congratulations, your page is in the top $n percent of pages for correctness!" where $n ranges from 0 to 100 |
| 19:06 | <Philip`> | Then nobody will be disheartened by rejection, and they will be rewarded for incremental improvements |
| 19:08 | <Philip`> | (To calculate $n, just find the percentage rank of the page's number of errors vs a precomputed random sample of the web, and then divide by 2 and add 50) |
| 19:09 | <Philip`> | (...then subtract from 100, obviously) |
| 19:09 | <Dashiva> | Great idea |
| 19:09 | <Philip`> | Thank you |
| 19:10 | <Philip`> | I consider it to be in the top 50 percent of ideas for greatness |
| 19:10 | <Dashiva> | And then when people are used to it, you can slowly bias the random sample towards higher quality |
| 19:10 | <Dashiva> | And people will think everyone else is improving, so they'll try to improve too |
| 19:11 | <gsnedders> | Why base it upon a random sample? Why not just choose an algorithm to create the number? |
| 19:11 | <Dashiva> | That would be lying |
| 19:12 | <Philip`> | The point is to distort the truth, not to just make something up |
| 19:13 | <Philip`> | You've got to base the decisions on data, so you can simply hide your biases in your analysis of the data |
| 19:14 | <Dashiva> | Ssh, you're giving away the secrets |
| 19:48 | <gsnedders> | What DOM does http://software.hixie.ch/utilities/js/live-dom-viewer/?%3C!DOCTYPE%20html%3E%0D%0A%3Cscript%3E%0D%0Avar%20title%20%3D%20document.getElementsByTagName(%22title%22)%5B0%5D%0D%0Atitle.parentElement.removeChild(title)%0D%0A%3C%2Fscript%3E give in IE? |
| 19:50 | <Dashiva> | "Internet Explorer has modified this page to help prevent cross-site scripting" |
| 19:51 | <gsnedders> | If you manually set the src to that? |
| 19:51 | <Dashiva> | If I paste it in manually, html > head, body; head > script |
| 19:52 | <gsnedders> | k |
| 19:52 | <gsnedders> | And if you remove the script there is a title? |
| 19:52 | <Dashiva> | Yes. Before the script. |
| 19:52 | <gsnedders> | Does w(title.firstChild == null) give true? |
| 19:53 | <Dashiva> | Then? |
| 19:53 | <Dashiva> | Or with script |
| 19:53 | <gsnedders> | Without the second line of script |
| 19:53 | <gsnedders> | (without the first title isn't set) |
| 19:55 | <Dashiva> | w() doesn't seem to work |
| 19:55 | <gsnedders> | At all? |
| 19:55 | <gsnedders> | It should output it in the log |
| 19:56 | <gsnedders> | Just use alert then |
| 19:56 | <Dashiva> | There seems to be a comment in the source about it not working too |
| 19:57 | <Dashiva> | true both before and after removing the title |
| 19:58 | <gsnedders> | And if you add <title>foo</title> false? |
| 19:59 | <gsnedders> | (That's the final question, I promise) |
| 19:59 | <Dashiva> | Before the script tag? |
| 20:00 | <gsnedders> | Actually, it doesn't matter |
| 20:00 | <Dashiva> | firstChild is null still |
| 20:01 | <gsnedders> | WTF? |
| 20:01 | <Dashiva> | outerHTML is <TITLE>foo</TITLE> |
| 20:01 | <gsnedders> | Ergh. |
| 20:01 | <gsnedders> | and innerHTML? |
| 20:01 | <Dashiva> | childNodes has length 0, for consistency |
| 20:01 | <Dashiva> | innerHTML is foo |
| 21:35 | <Dashiva> | "I concluded the W3C has ceded the development of HTML to a cartel of browser vendors with no interest in the needs of authors and users (not even their own users)" |
| 21:35 | <Dashiva> | It's a cabal, not a cartel |
| 21:39 | <takkaria> | Rob is really good at writing rhetoric |
| 21:40 | <Dashiva> | The overuse of "ignoring users" kinda grates for me |
| 21:40 | <takkaria> | rhetoric always grates on the people it's not aimed at |
| 21:41 | <Dashiva> | Fair enough |
| 21:43 | <TabAtkins> | Question: is Rob Burns on crack? |
| 21:44 | <takkaria> | if he was I'm not sure he'd bother with the htmlwg |
| 21:45 | <TabAtkins> | All right, then assume a crack-like drug that still produces a compulsion to comment on technical mailing-lists. |
| 21:45 | <TabAtkins> | Is he on this theoretical drug? |
| 21:45 | <Dashiva> | He's just himelf |
| 21:45 | <Dashiva> | *himself |
| 21:46 | <TabAtkins> | That's not a very fun self to be, filled with constant paranoia that the browser vendors hate everyone and presumably want to kill blind people. |
| 21:47 | <Dashiva> | As far as behavior goes, posting on a mailing list is greatly preferable to other activities society may frown upon |
| 21:48 | <TabAtkins> | While true, that doesn't make it any easier to read his email without facing the hard decision of whether to laugh my coke through my nose or dissolve into furious apoplexy. |
| 21:49 | <Dashiva> | If you find yourself in that situation, there is a third way: not reading it |
| 21:49 | <TabAtkins> | I took that third way after two paragraphs. |
| 21:50 | <Dashiva> | I think it's good that he posts, really |
| 21:51 | <Dashiva> | Better to get the controversy out in the open so it can be resolved |
| 21:51 | <TabAtkins> | I dunno if that sort of post is useful, though. I mean, yeah, when no one's disagreeing you're either a genius or no one around you cares. |
| 21:52 | <TabAtkins> | But how do you *respond* to something like "I like your draft because the other guy is part of a shadowy conspiracy with browser vendors to make the internet suck." |
| 21:53 | <TabAtkins> | Might as well just blame interop issues on aliens for all the good it does. |
| 21:54 | <TabAtkins> | Like, okay, I dislike Shelley Power. I *really* dislike her. But she performs a very valuable service by strongly disagreeing with stuff, and offering reasons why she disagrees that you can actually address. I rarely agree with her reasons, but I can *understand* them, and that's really useful to have. |
| 21:55 | <TabAtkins> | Bah, back to writing spec proposals. |
| 22:34 | jgraham | wonders which real browsers implement the XHTML access element |