| 00:29 | <mpilgrim> | boy, webkit's indexeddb implementation does not appreciate when you use a 64KB string as a version identifier |
| 00:29 | <mpilgrim> | this is all Philip`'s fault |
| 00:29 | <mpilgrim> | (the string has one of each possible 2-byte character) |
| 00:30 | <Philip`> | Does it appreciate that particular string less than any other 64KB string? |
| 00:30 | <mpilgrim> | doesn't seem to like anything that length |
| 00:31 | <mpilgrim> | also, keys can't be \u0000 or \ufffe, but names, version strings, and values can |
| 00:31 | <mpilgrim> | i ran an exhaustive test, and those were the only two values that failed |
| 00:31 | <mpilgrim> | er, only two keys |
| 00:32 | <mpilgrim> | so kudos for recommending them when i asked for problematic strings |
| 00:32 | <mpilgrim> | they were! |
| 05:11 | <llrcombs> | question: do mouseover/mouseout events bubble by default in the current spec? |
| 07:19 | <llrcombs> | question: do mouseover/mouseout events bubble by default in the current spec? |
| 07:45 | <hsivonen> | so it looks like people really want Dublin Core stuff to be registered but can't manage to register that stuff properly |
| 07:46 | <hsivonen> | I supposed I should take the time to register the DC stuff instead of just doing BOFHy wiki management explaining why the attempts so far failed |
| 07:50 | <zcorpan> | with hardcoded prefix? |
| 07:51 | <hsivonen> | zcorpan: I was planning on registering the dc. stuff. I've never seen a *meta element*-related spec for the dcterms. stuff |
| 07:51 | <hsivonen> | I wished crowdsourcing could fix this |
| 07:52 | <hsivonen> | but it seems our very low-bar bogofilter is too high bar |
| 07:52 | <hsivonen> | the bogofilter being that the person wanting to use a meta keyword should be bothered to find a spec that defines that keyword *for use as a meta keyword* |
| 07:53 | <hsivonen> | s/the person/the first person/ |
| 07:54 | <hsivonen> | at least the person registering the porn meta keyword got the registration correct enough |
| 07:55 | <hsivonen> | I guess porn publishers have incentives to give the appearance of self-regulation to avoid actual regulation |
| 08:00 | <hsivonen> | I still find it interesting that people rather file bugs or blog than look up the spec URL and edit a wiki |
| 08:02 | <zcorpan> | what's teh porn keyword? |
| 08:03 | <hsivonen> | zcorpan: rating |
| 08:04 | <zcorpan> | ah |
| 08:10 | <hsivonen> | what's the mediawiki incantation for hiding an inline HTML <h2> from the table of contents? |
| 08:12 | <Ms2ger> | Use <h2> instead of ==, I think |
| 08:16 | <hsivonen> | Ms2ger: the heading I want to hide already uses <h2> instead of == |
| 08:19 | <hsivonen> | I used <b> |
| 08:19 | <hsivonen> | problem solved |
| 08:37 | <zcorpan> | hsivonen: the semanticists will haunt you and your children for this |
| 08:51 | <hsivonen> | zcorpan: the semanticists could start by registering their keywords! |
| 08:55 | <hsivonen> | doh. I made the mistake of writing a proper response to a bug that was filed by contributor@ |
| 08:58 | <hsivonen> | and the comment form on the reporter's blogspot blog is broken |
| 08:59 | hsivonen | wonders if MikeSmith would like to redeploy the W3C HTML5 validator with zcorpan's new UI text about the registries |
| 09:01 | <zcorpan> | hsivonen: btw i have changed the text for meta-name too |
| 09:03 | <hsivonen> | zcorpan: thanks. now deployed |
| 09:03 | <hsivonen> | It seems that figuring out what all the DC keywords are is going to take enough time that I'm going to wait for crowdsourcing to work for a while longer |
| 09:04 | <hsivonen> | for reference: http://dublincore.org/documents/dc-html/ and http://tools.ietf.org/html/rfc5791 |
| 09:18 | <hsivonen> | Hixie: should p3pv1 be Ratified when it is defined normatively in a W3C REC that predates HTML5 and the registry system? |
| 09:19 | <foolip> | danbri, that works for me |
| 09:19 | <foolip> | zcorpan, that would require server-side stuff... |
| 09:19 | danbri | tries to remember something that happened more than 5 mins ago; fails |
| 09:19 | <foolip> | danbri, about http://schema.org/TVSeries |
| 09:20 | <danbri> | ah :) |
| 09:20 | <foolip> | you said it broke http://foolip.org/microdatajs/live/ in some way |
| 09:20 | <danbri> | i'll try again. i'm in osx chrome fwiw. |
| 09:20 | <foolip> | ok, I can try it in Chromium here |
| 09:20 | <danbri> | btw @edsu has made or integrated a python microdata parser into rdflib |
| 09:21 | <danbri> | so that gives 3 tools to compare/test now at least; yours, perl from tobyink and python |
| 09:21 | <foolip> | the more the merrier :) |
| 09:21 | <foolip> | as you can deduce from the mailing list we're implementing it natively in Opera too |
| 09:21 | <Ms2ger> | foolip, you can't say that ;) |
| 09:21 | <foolip> | anyway, I tried that in Chromium and it still works fine |
| 09:22 | <foolip> | Ms2ger, it tends to give more weight to the feedback when you confess to implementing something :) |
| 09:25 | <zcorpan> | foolip: what? |
| 09:25 | <foolip> | looks like everyone has poor long-term memory today :) |
| 09:25 | <zcorpan> | ah, short url |
| 09:26 | <danbri> | does natively still mean 'in c'? |
| 09:26 | <zcorpan> | sure but i'm sure you're talented enough to figure it out :) |
| 09:26 | <danbri> | (or just 'bundled'...) |
| 09:27 | <zcorpan> | danbri: natively means in IE on Windows, duh |
| 09:27 | <foolip> | danbri, it happens to mean "in C++", but as long as the API is there for scripts to use I don't think that's important |
| 09:27 | <danbri> | :) |
| 09:31 | <jgraham> | foolip: I thought it was supposed to be a kind of intelligence test. People are supposed to deduce from us providing detailed algorithm-level feedback that maybe, just maybe, we aren't just reading the spec for fun |
| 09:33 | <foolip> | In that case I want "are they implementing" as a turing test for spambots :) |
| 09:33 | <hsivonen> | foolip: would that be like Anne's "well-formed XML" spam filter? :-) |
| 09:34 | <foolip> | hehe, I usually fail that on the first attempt :) |
| 09:34 | <zcorpan> | bozo! |
| 09:35 | <foolip> | why Tim Bray no love me? :'( |
| 09:37 | <jgraham> | Tim bray love you long time!</spam> |
| 09:38 | jgraham | wonders why people think that inconsistencies between the full spec and the author only view are more likely to be "deeper problems" than simple mechanical issues in the redaction process, or human errors in the markers |
| 09:41 | <Ephemera> | any body know about uncaching oneself using offline application cache |
| 09:42 | <hsivonen> | Ephemera: I don't know. I only fix browser bugs that go the other way round. |
| 09:42 | <hsivonen> | Ephemera: but I'm guessing you could serve a manifest file that is empty |
| 09:42 | <hsivonen> | maybe |
| 09:42 | <hsivonen> | just a guess |
| 09:42 | <hsivonen> | (don't try omitting the manifest attribute. that won't do the trick) |
| 09:43 | <hsivonen> | Ephemera: empty as in the new version of an old manifest has no entries |
| 09:44 | <Ephemera> | yes, that's right. but it doesnt work for oneself file |
| 09:45 | <danbri> | what's the use case for microdata's properties being ordered? is it more around editing than around the semantic content of what the properties *say*? (eg. consider repeated author property -- does being '1st' mean something useful?) |
| 09:45 | <jgraham> | danbri: I assume the vocabulary defines that |
| 09:45 | <danbri> | in English / natural language? |
| 09:46 | <danbri> | at the vocabulary URI? |
| 09:46 | <jgraham> | However the vocabulary chooses to define such things |
| 09:46 | danbri | nods |
| 10:06 | danbri | gets http://code.google.com/p/json-template/ from foolip on chinese html list (I only read it for the urls :) |
| 10:10 | <foolip> | danbri, I just found right now, thought it looked like a sane approach (he asked for how to turn JSON into HTML, since the microdata API can do the reverse) |
| 10:12 | danbri | hadn't seen it, will pass it to rdfa folk see if it works with their api too |
| 10:22 | <foolip> | danbri, json-template doesn't seem to have anything to do with microdata as such, I think |
| 10:23 | <foolip> | And there's no mapping from RDFa to JSON, I presume? (it could only be a list of triples or something) |
| 10:23 | <danbri> | i guess the bridge is assuming the microdata API gives a json-ny view of the properties |
| 10:23 | <foolip> | right |
| 10:23 | <foolip> | although probably not pretty enough to want to use directly |
| 10:23 | <foolip> | since you have these "items" and "properties" indirections in the JSON |
| 10:23 | <danbri> | yeah there are a few things for rdf(a)<->json; some ugly pragmatic triple dumps and a few attempts to do something prettier, eg. http://json-ld.org/ |
| 10:24 | <danbri> | yeah it ends up being indirect in some representations. json-ld maybe is better that way? except it has prefixing... |
| 10:25 | <foolip> | I'm not sure, it looks from A Simple Example like one object talks about a single subject |
| 10:26 | <foolip> | but perhaps you're allowed to nest the object when there's no loops in the graph |
| 10:28 | <danbri> | there's a microdata example here btw, http://json-ld.org/spec/latest/#microdata |
| 10:28 | <danbri> | and yup it allows nesting |
| 10:29 | <foolip> | funny how things look more official and trustworthy with the W3C stylesheet :) |
| 10:29 | <foolip> | looks like the microdata extracted as RDF and then encoded as JSON-LD |
| 10:29 | <foolip> | I guess there's a reason he didn't pick an example that's not designed to be exported to RDF |
| 10:30 | <foolip> | people don't seem amused by URIs like http://www.w3.org/1999/xhtml/microdata#http%3A%2F%2Fmicroformats.org%2Fprofile%2Fhcard%23%3Afn |
| 10:31 | <danbri> | heh, yeah the stylesheet took a second to load too, so i saw the naked version |
| 10:31 | <danbri> | yeah, they result in 'Sorry, Insufficient Access Privileges' at w3.org |
| 10:32 | <foolip> | ah, does it only work if you're logged in as a member? |
| 10:32 | <foolip> | (which I am) |
| 10:32 | danbri | is too (at least i can see /Member ) |
| 10:34 | <danbri> | I get same in OSX and Safari |
| 10:35 | <foolip> | danbri, did you manage to reproduce the issue with http://schema.org/TVSeries |
| 10:37 | <danbri> | let's try. Ok, in OSX Chrome I visit that + http://foolip.org/microdatajs/live/ ... I view src on TVSeries, ...select all, copy, paste into your form. Now I see the tab fill out with the semi-rendered page (ie. image urls break) in Preview tab. |
| 10:37 | <foolip> | select all, not just the markup? |
| 10:37 | <danbri> | ...at this point, i wait 30 seconds, it's still there, ... but if i click and type a space char into the textarea, whole thing goes to white page |
| 10:38 | <danbri> | yes, doctype and all |
| 10:38 | <foolip> | do you're pasting the source of http://schema.org/TVSeries ? |
| 10:38 | <foolip> | s/do/so/ |
| 10:39 | <danbri> | yes |
| 10:39 | <danbri> | oh, when i type that space, i see this in dev console: Uncaught TypeError: Object [object Object] has no method 'tabs' |
| 10:40 | <foolip> | I must say I don't understand why you'd paste the source instead of the example |
| 10:40 | <foolip> | but I tried it, it screws up Opera as well |
| 10:40 | <danbri> | but some errors before that, screenshot coming up... http://i.imgur.com/gzsgV.png |
| 10:41 | <danbri> | because i thought the source actually contained microdata :) guess it doesn't? |
| 10:41 | <danbri> | i was after their schema |
| 10:41 | <danbri> | ...which is available (in escaped form in full) at http://schema.org/docs/full_md.html |
| 10:42 | <foolip> | the problem is probably that I put the HTML in the same document as the tool itself, so they can interfere |
| 10:42 | <foolip> | run scripts and what not |
| 10:42 | danbri | nods |
| 10:42 | <foolip> | it also makes all relative URLs really ugly |
| 10:42 | <foolip> | even more so if you've followed a permalink |
| 10:43 | <danbri> | full_md.html schema was parsed with tobyink's perl parser to generate -> http://ar.no.de/schema.org/ (click 'origin of the world' ...) |
| 10:43 | <danbri> | ...doing it in pure js would make some sense :) |
| 10:44 | <foolip> | I must say having types is a bit odd, are those actually used as microdata properties somewhere? |
| 10:44 | <danbri> | which ones? |
| 10:45 | <foolip> | http://schema.org/DataType |
| 10:46 | <foolip> | seems like it's just for "Expected Type", perhaps for validation |
| 10:46 | <danbri> | hang on, i'll post the turtlified schema |
| 10:47 | <danbri> | https://raw.github.com/gist/1018567/f24ddc39be31a6850a49bf9349384e6dff13a420/gistfile1.txt |
| 10:48 | <danbri> | so so:Date gets used with so:birthDate, so:datePublished, so:deathDate, so:endDate, ... so:expires, so:foundingDate, so:priceValidUntil etc |
| 10:48 | <danbri> | boolean seems unused |
| 10:49 | <danbri> | also Number, Text. I guess they were added for pseudo-completeness? |
| 10:53 | <foolip> | maybe |
| 10:54 | <foolip> | I do wonder why they made such a schema right up front |
| 10:55 | <foolip> | it's note like they're going to have specialized search results for http://schema.org/WebPage |
| 10:55 | <foolip> | it's just cruft that people will add as SEO incantations |
| 10:57 | <foolip> | with http://schema.org/SiteNavigationElement they also seems to assume that the microdata is somehow connected to the elements where they're declared, while the spec says "It's important to note that there is no relationship between the microdata and the content of the document where the microdata is marked up." |
| 10:57 | <Lachy> | how did they develop these schemas? Were any based on existing microformats, or is there any evidence that they actually based them on real use cases? |
| 10:58 | <foolip> | dunno |
| 10:58 | <foolip> | but it looks like 90% of has very questionable utility |
| 10:58 | <foolip> | (which is of course true of most vocabularies) |
| 10:59 | <danbri> | any schema that has 'Boolean', 'Optician', 'cholesterolCount', 'UserPageVisits', and 'Volcano' presents certain, erm, maintainance challenges :) |
| 10:59 | <danbri> | I expect they're based on some sense of most common query topics |
| 11:00 | <foolip> | still, there are parts that I hope will succeed, like the music-related terms |
| 11:00 | <foolip> | specialized results for MusicBrainz (if it used the schema) would be actually useful |
| 11:01 | <danbri> | i was about to say yeah, getting music schema right is hard, ... it took MusicBrainz years to migrate to their improved schema - http://musicbrainz.org/ ...which treats classical music better |
| 11:01 | <danbri> | http://blog.musicbrainz.org/?p=398 |
| 11:02 | <danbri> | oh pretty but scary diag in http://wiki.musicbrainz.org/Next_Generation_Schema |
| 11:02 | <foolip> | I know, but schema.org actually looks surprisingly sane |
| 11:03 | <foolip> | to be clear, MusicBrainz is of course the canonical schema for music, and by definition the sanest! |
| 11:03 | <danbri> | no searching for mp3s tho ;) |
| 11:04 | <danbri> | y'all saw http://dallemang.typepad.com/my_weblog/2011/06/stop-press-microdata-in-topbraid.html btw? |
| 11:05 | <foolip> | nope, thanks |
| 11:05 | <hsivonen> | foolip: it should only take a bit of OWL to deal with the unamusing URLs :-) |
| 11:06 | <foolip> | hsivonen, quite a bit actually, since you'd have to create sameAs links for each vocabulary term |
| 11:07 | <danbri> | or we could not screw things up in the first place, how's that for a crazy idea? |
| 11:07 | danbri | not a big believer in OWL as a data repair tool |
| 11:07 | <foolip> | anyway, even if it were simple people aren't going to be amused, because they like to look at their RDF and see nice prefixes and such |
| 11:10 | <foolip> | huh, looks like it's the same holger that complained about ugly microdata RDF and that implemented it for TopBraid :) |
| 11:10 | <danbri> | we're pragmatists ;) |
| 11:11 | <foolip> | danbri, you're involved with TopBraid? |
| 11:11 | <danbri> | no tho I have friends working there, and i've collab'd w/ Holger lightly in his previous work |
| 11:12 | <danbri> | i suggested some integration (sparql/protege) and within a day he'd built it ... http://permalink.gmane.org/gmane.comp.misc.ontology.protege.owl/12269 |
| 11:23 | <hsivonen> | foolip: Hixie had a theoretically sensible reason for doing the URL mapping the way he did |
| 11:24 | <hsivonen> | foolip: though I'm pretty sure almost everyone actually seeking to use the RDF mapping is aesthetically unhappy about it |
| 11:24 | <foolip> | hsivonen, yes, I'm fully aware of that, and have pointed it out to others who have complained |
| 11:25 | <hsivonen> | in other news, it seems that there there are Social Media experts who plagiarize tantek's tweets |
| 11:25 | <foolip> | though I suspect the number of people whould *actually* use RDF reasoners to do anything useful is vanishingly small |
| 11:30 | <smaug____> | does a_element.itemValue = "http://www.example.org"; end up changing href attribute? |
| 11:32 | <smaug____> | apparently so |
| 11:33 | <smaug____> | strange API |
| 14:26 | <karlcow> | this is a strange wikipedia page http://en.wikipedia.org/wiki/Comparison_of_layout_engines_(Non-standard_HTML) |
| 14:30 | <gsnedders> | hah, I'm cited on that page |
| 14:32 | <Ms2ger> | gsnedders, looks like hsivonen added that :) |
| 14:33 | <gsnedders> | Ms2ger: Indeed. Especially amusing if you follow the citation. |
| 14:33 | <Ms2ger> | Heh |
| 15:10 | <_bga> | hm |
| 15:10 | <_bga> | global ns bigger and bigger |
| 15:10 | <_bga> | its bad |
| 15:10 | <_bga> | php way |
| 15:11 | <_bga> | may be time for namespace all "classes" ? |
| 15:11 | <_bga> | WebGL.Shader |
| 15:11 | <_bga> | SVG.SetElement |
| 15:12 | <_bga> | ? |
| 15:16 | <gsnedders> | _bga: http://wiki.ecmascript.org/doku.php?id=harmony:modules |
| 15:18 | <_bga> | i dont believe that es6 will be success and vendors will implement it but ok |
| 15:19 | <gsnedders> | _bga: SpiderMonkey is already implementing parts of it. |
| 15:19 | <_bga> | gsnedders because Eich make spidermonkey |
| 15:20 | <gsnedders> | _bga: Brenden is scarcely involved in day-to-day development of it, AFAIK |
| 15:20 | <_bga> | gsnedders i guess coffeescript or other translang will win |
| 15:21 | <gavin> | brendan's still pretty involved in day-to-day development |
| 15:24 | <_bga> | btw |
| 15:24 | <_bga> | Float64Array in webkit |
| 15:25 | <_bga> | but typed arrays is currently slower than [] |
| 15:25 | <_bga> | :( |
| 15:25 | <gavin> | for what testcase? |
| 15:26 | <gavin> | spidermonkey also implements typed arrays (including Float64Array) |
| 15:26 | <mpilgrim> | lol. if you call webkitIndexedDB.open() with no parameters, it creates a database with the 9-character name "undefined" |
| 15:26 | <gavin> | heh |
| 15:26 | <mpilgrim> | is that an IDL failing? |
| 15:27 | <mpilgrim> | is there some magic IDL parameter to say "treat undefined as null" or something? |
| 15:27 | <jgraham> | Thyat sounds like expected behaviour in js at least |
| 15:27 | <mpilgrim> | (we already handle open(null) properly) |
| 15:27 | <jgraham> | Yeah I think there is some IDL magic one can wave |
| 15:27 | <jgraham> | mpilgrim: Do you plan to contribute your tests to the WG btw? |
| 15:28 | <mpilgrim> | as time permits |
| 15:28 | <jgraham> | You should make time permit :) |
| 15:28 | <mpilgrim> | my manager is very webkit-focused |
| 15:30 | <jgraham> | I thought everyone at google believed that what's good for the web is good for Google |
| 15:30 | <mpilgrim> | my objectives this quarter have been very webkit-centric |
| 15:31 | <jgraham> | You should charge the time to the other departments who won't have to spend their time working around bugs in other browsers that your tests would have caught |
| 15:31 | <mpilgrim> | i'll push for some time to port these tests in Q3 |
| 15:31 | <jgraham> | Great |
| 15:32 | <mpilgrim> | i've already been wrapping webkit-specific prefixes on variable names and such |
| 15:32 | <mpilgrim> | has the testing WG settled on a framework yet? |
| 15:32 | <mpilgrim> | i've paid 0 attention |
| 15:33 | <jgraham> | For javascript tests we seem to be using testharness.js |
| 15:33 | <mpilgrim> | ok |
| 15:33 | <jgraham> | Which you used for some video tests iirc |
| 15:33 | <mpilgrim> | yes |
| 15:33 | <mpilgrim> | i ported those to webkit, btw |
| 15:33 | <mpilgrim> | wrapped 'em like we wrapped philip's canvas tests |
| 15:33 | <mpilgrim> | to make them work as LayoutTests |
| 15:34 | <jgraham> | In some generic way? |
| 15:34 | <mpilgrim> | yes, just had a change the script tags |
| 15:34 | <jgraham> | Which script tags? |
| 15:34 | <mpilgrim> | it was a generic testharness.js-to-LayoutTests wrapper |
| 15:35 | <jgraham> | There is a not-very-well-enforced requirement to add a <script src="/resources/testharnessreport.js"></script> in testharness.js files |
| 15:35 | <mpilgrim> | the ones that pointed to ../../../../testharness.js |
| 15:35 | <jgraham> | So that vendors can put whatever random code they need to integrate with testharness.js in testharnessreport.js |
| 15:37 | <mpilgrim> | [Undefined=Null, Null=Null] attribute DOMString owner; |
| 15:37 | <mpilgrim> | looks promising |
| 15:38 | <jgraham> | mpilgrim: BTW if you have an webkit followup on http://www.w3.org/mid/4DE4C870.7020408⊙oc it would be very interesting |
| 15:40 | <mpilgrim> | looks like webkit needs that IDL magic on basically every non-optional non-nullable argument on every method |
| 15:40 | <mpilgrim> | bleah |
| 15:41 | <mpilgrim> | JAVASCRIPT Y U NO SANE?!? |
| 15:41 | <gavin> | that seems like an odd default behavior for IDL |
| 15:41 | <jgraham> | Speak to heycam|away |
| 15:41 | <mpilgrim> | http://www.w3.org/TR/2008/WD-WebIDL-20081219/#Undefined has a very clear example that perfectly illustrates the behavior i'm seeing in webkit |
| 15:41 | <gsnedders> | mpilgrim: JS is sane! Just, uh, not host objects… |
| 15:42 | <mpilgrim> | shit, i'm looking at an old copy of the spec |
| 15:42 | <mpilgrim> | it's now http://www.w3.org/TR/WebIDL/#TreatUndefinedAs |
| 15:42 | <gavin> | w3c'ed |
| 15:43 | <jgraham> | Hmm, the one on /TR/ shouldn't be the newest one |
| 15:43 | <jgraham> | You want the one on dev.w3.org |
| 15:43 | <mpilgrim> | damn it |
| 15:44 | <mpilgrim> | ok, no change: http://dev.w3.org/2006/webapi/WebIDL/#TreatUndefinedAs |
| 15:45 | <mpilgrim> | ok, now i have two problems |
| 15:45 | <mpilgrim> | [TreatUndefinedAs=Null] can only be used on DOMString attributes |
| 15:45 | jgraham | notes again how awesome the W3C dated URLs are |
| 15:46 | <mpilgrim> | but i have "any" attributes that also need this behavior |
| 15:46 | mpilgrim | wonders if webkit enforces that rule anyway |
| 15:47 | karlcow | wonders what systems jgraham would take for unique identifiers |
| 15:49 | <jgraham> | karlcow: I would make that URL http://specs.w3.org/WebIDL |
| 15:49 | <jgraham> | Or something |
| 15:49 | <mpilgrim> | no, i take it back, the "any" parameters are behaving themselves already |
| 15:49 | <karlcow> | jgraham: what about the future? |
| 15:50 | <AryehGregor> | karlcow, put the latest version of WebIDL at that URL forever. |
| 15:50 | <AryehGregor> | Why would anyone want an obsolete version anyway? |
| 15:51 | <karlcow> | well AryehGregor it is what /TR/shortname is somehow. with a different notion of last version. |
| 15:51 | <karlcow> | AryehGregor: I'm interested by older versions. |
| 15:51 | <jgraham> | karlcow: http://sepcs.w3.org/WebIDL/{revison-id} |
| 15:51 | <AryehGregor> | karlcow, then you can spend slightly more effort to find them. |
| 15:52 | <AryehGregor> | Way too many people wind up looking at useless, obsolete versions of specs in the current W3C setup. |
| 15:52 | <karlcow> | two different issues. |
| 15:52 | <AryehGregor> | But yeah, you can always just use a path. |
| 15:52 | <karlcow> | :) |
| 15:52 | <AryehGregor> | After the revision. |
| 15:52 | <karlcow> | the pattern for naming the document |
| 15:52 | <karlcow> | the findability of the document |
| 15:53 | <AryehGregor> | Like just have it take a date or revision id and show you what the version was at that time. |
| 15:53 | <AryehGregor> | Or, really, just set up hgweb or ViewVC or whatever so people can look at the changelog. |
| 15:53 | <karlcow> | Maybe when a specification has been replaced by a newer version, there should be a robots.txt for removing it from search engines crawling |
| 15:54 | <llrcombs> | question: do mouseover/mouseout events bubble by default in the current spec? |
| 15:55 | <llrcombs> | also: this line often sets a channel into a spiral of fun that simply can't be stopped: http://rodgercombs.webhop.net/maps/freenode/whatwg/current/ |
| 15:57 | <karlcow> | jgraham: http://specs.w3.org/WebIDL/{revison-id} would be the last commit? |
| 15:58 | <karlcow> | let me rephrase that |
| 15:58 | <karlcow> | jgraham: http://specs.w3.org/WebIDL/ would be the last commit |
| 15:58 | <karlcow> | jgraham: http://specs.w3.org/WebIDL/{revison-id} would be the revision-id commit? |
| 16:07 | <jgraham> | karlcow: Yes |
| 16:09 | <jgraham> | karlcow: (and it could even cope with multiple versions and things because you could have e.g. spec.w3.org/html/5/ which would be a tag or branch in the underlying repository) |
| 16:10 | <jgraham> | (depends on 5 not being a valid rev. id of course) |
| 16:10 | <jgraham> | (but one could make up a name somehow) |
| 16:16 | <mpilgrim> | http://lists.w3.org/Archives/Public/www-archive/2007May/0068.html |
| 16:23 | <karlcow> | holy cow. interesting list. I would love to see other engines too. |
| 16:25 | <Ms2ger> | https://developer.mozilla.org/en/XPIDL |
| 16:25 | <Ms2ger> | But mostly out-of-date, unfortunately |
| 17:39 | <othermaciej> | mpilgrim: pretty sure that is considerably out of date now |
| 18:43 | <mpilgrim> | othermaciej: yt? |
| 18:43 | <mpilgrim> | whom would i talk to about webkit's IDL implementation? |
| 18:43 | <mpilgrim> | i need a way to map undefined DOMStrings to null, or at least to something more useful than the 9-character string "undefined" |
| 18:44 | <othermaciej> | mpilgrim: you could talk to me, or maybe weinig |
| 18:44 | <othermaciej> | mpilgrim: in my case I'm busy for the next few hours |
| 18:44 | <othermaciej> | weinig knows more than me but may not be fresh on it |
| 18:45 | <mpilgrim> | it can wait |
| 18:45 | <mpilgrim> | most of the required method arguments in IndexedDB need this IDL magic |
| 18:46 | <mpilgrim> | ooh, i might have just found it |
| 18:47 | <mpilgrim> | ConvertUndefinedOrNullToNullString |
| 18:47 | <zcorpan> | what's wrong with "undefined"? |
| 18:50 | <jgraham> | zcorpan: It's too obvious. It should convert it to "null" instead :) |
| 18:51 | <TabAtkins> | NullString is a null string, not "null". |
| 18:52 | <mpilgrim> | TabAtkins: we had that problem too, open(null) would open a database with the 4-character name "null" |
| 18:52 | <jgraham> | TabAtkins: Try again with a much lower humor threshold |
| 18:52 | <mpilgrim> | i fixed that a few weeks ago, IIRC |
| 18:52 | <TabAtkins> | jgraham: Error lowering humor threshold - subject already giggles at poop jokes. Perhaps try improving humor? |
| 18:53 | jgraham | claims that it is only suficiently inconsistent with the web platform if openDatabase(null) -> "undefined" and openDatabase(undefined) -> "null" |
| 18:53 | <jgraham> | s/with the/to be part of the/ |
| 18:53 | <zcorpan> | speaking of null, have somebody figured out yet which methods and properties want null to be "null" and which want it to be ""? |
| 18:53 | <jgraham> | TabAtkins: NOT_SUPPORTED_ERR |
| 18:53 | <Ms2ger> | Worse is better, make them all "" |
| 18:54 | <mpilgrim> | can't. empty strings are allowed in places where null is not. |
| 18:54 | <mpilgrim> | so not kidding |
| 18:54 | <jgraham> | Well that would make sense |
| 18:55 | mpilgrim | knows waaay too much about IndexedDB already |
| 18:55 | <jgraham> | I mean if javascript had any type safety at all |
| 18:55 | <mpilgrim> | and it's only been a few weeks |
| 18:55 | <jgraham> | Which would, you know, be nice |
| 18:55 | jgraham | glares meaningfully in Brendan Eich's direcetion |
| 18:56 | <TabAtkins> | Fuck type safety. You just want type consistency. |
| 18:56 | <TabAtkins> | Which Javascript, admittedly, also doesn't have. |
| 18:56 | <mpilgrim> | fuck this, let's talk about microdata |
| 18:57 | <TabAtkins> | Shoot. |
| 18:57 | <mpilgrim> | TabAtkins: do you know of anywhere that schema.org claims to be either open or a standard? |
| 18:57 | <jgraham> | TabAtkins: The internets don't hlp me wok out what you mean by "type consistency" |
| 18:57 | <TabAtkins> | Aw, I thought we were going to talk about Microdata. |
| 18:57 | <jgraham> | Do you just mean "strong dynamic typing"? |
| 18:58 | <TabAtkins> | jgraham: More like "treat null the same everywhere, goddammit" |
| 18:58 | Ms2ger | wonders if jgraham likes wok |
| 18:58 | <mpilgrim> | schema.org is microdata. microdata is schema.org. freedom is slavery. |
| 18:58 | <mpilgrim> | i can feel the impending yellow highlighter already |
| 18:58 | Ms2ger | puts mpilgrim in charge of minitruth |
| 18:59 | <gsnedders> | TabAtkins: Just type it at runtime! :P |
| 18:59 | <TabAtkins> | BRILLIANT |
| 18:59 | <mpilgrim> | in retrospect, i probably shouldn't have said that microformats implementers had "backed the wrong horse" |
| 18:59 | <jgraham> | The first rule of yellow highlighter is... no fuck it that flight club thing was done to death years ago |
| 18:59 | <zcorpan> | so has anyone tested the search engines' impl of microdata? |
| 18:59 | <Ms2ger> | jgraham++ |
| 19:00 | <TabAtkins> | First rule of flight club is, no fucking in the bathroom. |
| 19:00 | <jgraham> | I am not in control of my fingers |
| 19:00 | <zcorpan> | are you in the bathroom? |
| 19:02 | <jgraham> | So... microdata |
| 19:02 | <mpilgrim> | holy crap, the IDL magic actually worked |
| 19:03 | <mpilgrim> | like all semantic markup, i care very little about microdata, yet feel oddly compelled to discuss it at great length |
| 19:04 | <TabAtkins> | At least, if you're doing semantic markup, Microdata has little evil in it. |
| 19:04 | <TabAtkins> | It's a decent syntax for a problem with very limited utility in the first place. |
| 19:05 | <scor> | TabAtkins: little evil in comparison to what? or what do you mean? |
| 19:05 | <zcorpan> | lang="en-x-not-hixie" - that doesn't follow the syntax rules does it? should be en-x-not-x-hixie right? (or en-x-nothixie or some such) |
| 19:06 | <Ms2ger> | I suggest filing an ISSUE. |
| 19:06 | <mpilgrim> | i think TabAtkins means "relatively little extraneous markup required" |
| 19:06 | <Ms2ger> | And threaten with an FO |
| 19:06 | <TabAtkins> | scor: Compared to the Microformats family of syntaxes (underspecified and specific to each vocab) or RDFa (overcomplicated). |
| 19:06 | <scor> | ah, you mean lightweight in other words? |
| 19:06 | <mpilgrim> | since the microdata value extraction algorithm works with the semantics of well-written HTML |
| 19:06 | <TabAtkins> | (Nothing against the Microformats vocabs - I think they're pretty decent in general. But the syntax sucks.) |
| 19:07 | <TabAtkins> | scor: More to it than just lightweightness, but that's a significant component. |
| 19:07 | <zcorpan> | Ms2ger: need to file a bug first |
| 19:07 | <Ms2ger> | Pff |
| 19:07 | <Ms2ger> | You don't need to follow the process if you want to abuse it |
| 19:07 | <zcorpan> | but i could file a FORMAL COMPLAINT |
| 19:08 | <gsnedders> | A FORMAL COMPLAINT and not a FORMAL OBJECTION? |
| 19:08 | <scor> | I see |
| 19:08 | <zcorpan> | yah |
| 19:08 | <TabAtkins> | Hixie: How did you come up with the selectors for setting the third nesting level of lists in the Rendering section? |
| 19:08 | <TabAtkins> | Hixie: It seems... inconsistent. |
| 19:09 | <TabAtkins> | Hixie: Wait, nevermind, I was misreading. It's consistent. It's just weird that <dl> is in the list. |
| 19:10 | <AryehGregor> | zcorpan, you don't need to file a bug to make a Formal Objection. I'm pretty sure that if you come out of nowhere and file a Formal Objection that HTML5 does not have a <cheese> element, with no supporting reasoning whatsoever, and refuse to withdraw it, the chairs have to record it and forward it to the Director. |
| 19:11 | <AryehGregor> | Who will of course be slightly amused and completely ignore it. |
| 19:11 | <TabAtkins> | AryehGregor: That is correct. It would obviously be rejected, but yeah. |
| 19:11 | <AryehGregor> | But everyone has the unalienable right to be personally ignored by the Director. |
| 19:11 | <zcorpan> | mmm cheese |
| 19:11 | <zcorpan> | i have plenty of reason to want cheese |
| 19:11 | <TabAtkins> | Plus, since it's TimBL, he can ignore you 50% faster than a normal human! |
| 19:11 | AryehGregor | likes referring to people by their titles when he could just as well refer to them by their names, for some reason |
| 19:11 | <TabAtkins> | (Dude's hyperactive.) |
| 19:12 | AryehGregor | also often refers to Hixie as "the editor" in HTMLWG discussions |
| 19:12 | <TabAtkins> | AryehGregor: Should I just refer to you as "The Intern", then? |
| 19:12 | <AryehGregor> | No, because I'm a contractor, not an intern. |
| 19:12 | <AryehGregor> | Or I'm a vendor, technically, as far as I understand it. |
| 19:12 | <TabAtkins> | Ah, kk. |
| 19:12 | <AryehGregor> | I just happen to be vending services rather than goods. |
| 19:12 | <gsnedders> | AryehGregor: So not The Knight? |
| 19:12 | <mpilgrim> | "The Editor" sounds like a really bad John Grisham novel |
| 19:12 | <jgraham> | AryehGregor: Surely his title is Sir Berners-Lee |
| 19:12 | <AryehGregor> | gsnedders, if that's a Batman reference, I don't get it. |
| 19:13 | <AryehGregor> | If it's a reference to something else, I also don't get it. |
| 19:13 | <Ms2ger> | Sir Tim, surely |
| 19:13 | <zcorpan> | Sir! |
| 19:13 | <zcorpan> | or maybe Mr. T. |
| 19:13 | <AryehGregor> | jgraham, knowing the British, I bet it's The Right Honorable Sir Timothy or something. |
| 19:13 | <AryehGregor> | I'm pretty sure that titles of nobility are supposed to go with your first name, not your last name. |
| 19:13 | <jgraham> | Sir TimBL has a nice ring to it. If you pronounce TimBL to rhyme with "thimble" |
| 19:14 | <gsnedders> | AryehGregor: He has a knighthood, thus the "Sir". |
| 19:14 | <AryehGregor> | I'm aware. |
| 19:14 | <AryehGregor> | Crazy Brits. |
| 19:14 | <jgraham> | Pretty sure they don't go with your first name |
| 19:14 | <AryehGregor> | Like Queen Elizabeth or Prince Charles? |
| 19:14 | <jgraham> | At least Lord Rees of Ludlow is Lord Rees, not Lord Martin |
| 19:15 | <gsnedders> | Apart from for monarchs, titles go with either surname or full name. |
| 19:15 | <jgraham> | Oh properly royal ones are different |
| 19:15 | <gsnedders> | s/monarchs/royalty/ |
| 19:15 | <jgraham> | They want to hide their German heritage |
| 19:15 | <AryehGregor> | Or they go with the plot of land it's associated with, right? If it is associated with one. |
| 19:15 | <gsnedders> | AryehGregor: The Earl of X |
| 19:15 | <gsnedders> | Earl Y of X |
| 19:15 | <AryehGregor> | I concede that Martin Rees is Baron Rees of Ludlow, according to Wikipedia. |
| 19:15 | <jgraham> | Wikipedia suggests "Sir Timothy John "Tim" Berners-Lee, OM, KBE, FRS, FREng, FRSA" |
| 19:16 | <AryehGregor> | Maybe it was different in the old days, and associating the titles with last names is a newfangled thing? |
| 19:16 | <jgraham> | Oh is he Baron? |
| 19:16 | <mpilgrim> | jeez, and i thought the web platform was crusty and complicated |
| 19:16 | <jgraham> | You would have though I would know after working in the same building for four years |
| 19:17 | <jgraham> | But all I learnt was that he got married in his lunch break |
| 19:17 | <mpilgrim> | did he get knighted while running out for a smoke? |
| 19:18 | <jgraham> | I also have an arguably-better photo than the one on wikipedia |
| 19:20 | <AryehGregor> | Is it freely licensed? |
| 19:20 | <AryehGregor> | Commons now has a user-friendly upload interface: http://commons.wikimedia.org/wiki/Special:UploadWizard |
| 19:20 | <jgraham> | It is CC-BY-NC but I could happily make it any other license |
| 19:21 | <AryehGregor> | Apparently, the user-friendly part involves a comic-book style interface in the form of a giant <img> with all text hardwired in and no alt text. |
| 19:21 | <AryehGregor> | But hey, baby steps, right? |
| 19:24 | jgraham | gives up at the "create an account" step |
| 19:25 | <jgraham> | I can't read the captcha |
| 19:25 | <zcorpan> | isn't there a button with an image of a wheelchair that you can try? |
| 19:28 | <jgraham> | No, there's a page that says something like "sucks to be you" |
| 19:29 | <jgraham> | Well it says "Unfortunately this may inconvenience users with limited vision or using text-based or speech-based browsers. At the moment we do not have an audio alternative available. Please contact the site administrators for assistance if this is unexpectedly preventing you from making legitimate posts at the Commons:Help desk" |
| 19:30 | <jgraham> | Dunno how it would unexpectedly prevent you from doing anything; it seems quite expected that an inability to see would prevent you reading the image |
| 19:30 | <AryehGregor> | You really can't read the captcha? It seems pretty readable to me. |
| 19:31 | <AryehGregor> | (although yes, it looks like no one has bothered streamlining the signup process) |
| 19:31 | <AryehGregor> | (registration used to be on the same page as login, you could just type your password a second time in an extra field, but that was removed a few years ago :( ) |
| 19:33 | <jgraham> | Well the first time my desired username was already taken |
| 19:33 | <jgraham> | Then the second time I had to enter a captcha again and I must have made a mistake |
| 19:34 | <jgraham> | Then the third time I couldn't read it |
| 19:34 | <jgraham> | Then I gave up |
| 19:34 | <zewt> | heh, recaptcha recently became totally unreadable, i regularly have to refresh them 3-5x |
| 19:34 | <AryehGregor> | Presumably the bots are getting smarter. |
| 19:34 | <zewt> | everyone knew captchas were a losing battle to begin with, anyway |
| 19:35 | <AryehGregor> | jgraham, oh well. The usability project has focused on improving editing usability, but maybe not improving signup usability. |
| 19:35 | <AryehGregor> | There's lots of low-hanging fruit there, it seems. |
| 19:35 | <AryehGregor> | zewt, anyone who runs a small wiki knows that captchas kill an awful lot of the spambots. |
| 19:35 | <AryehGregor> | At least they do in my experience. |
| 19:35 | <AryehGregor> | Although maybe not enough. |
| 19:35 | <TabAtkins> | zewt: CAPTCHAs were a *very clever* hack to get people to work harder on OCR. |
| 19:36 | <TabAtkins> | And it succeeded! |
| 19:36 | <zcorpan> | what do we want people to work on next? |
| 19:36 | <TabAtkins> | I think XKCD answered that. |
| 19:36 | <zewt> | TabAtkins: and in the case of recaptcha, a hack to turn people into involuntary OCR monkeys |
| 19:37 | <zcorpan> | TabAtkins: pointer? |
| 19:37 | <TabAtkins> | http://xkcd.com/810/ |
| 19:37 | <TabAtkins> | zcorpan: Started looking for it immediately. Just took me a bit to find it. ^_^ |
| 19:37 | <zewt> | AryehGregor: they do now, because presumably the robots that spam things like that aren't all that smart, but evidently smarter ones are becoming available |
| 19:37 | <Ms2ger> | zcorpan, 0xdefd67fa |
| 19:38 | <AryehGregor> | Ms2ger, what, you're still on 32-bit? |
| 19:38 | <Ms2ger> | Yes |
| 19:39 | <AryehGregor> | :( |
| 19:39 | AryehGregor | is too, but not for long |
| 19:39 | <zcorpan> | Your search - 0xdefd67fa - did not match any documents. |
| 19:39 | <zewt> | for most cases there's nothing wrong with 32-bit userland, anyway--very few individual applications need more |
| 19:40 | <jgraham> | It should have matched http://xkcd.com/138/ |
| 19:41 | <AryehGregor> | zcorpan, int main() { printf("%d\n", *0xdefd67fa); } |
| 19:41 | <TabAtkins> | zewt: Chrome uses some special allocator options that specifically vend high addresses preferentially, specifically to help us flush out 64-bit bugs. |
| 19:41 | <AryehGregor> | zewt, the extra address space is useful for ASLR, and the larger registers can be a big speedup for some applications (e.g., crypto). |
| 19:42 | <AryehGregor> | TabAtkins, for production builds or debug? |
| 19:42 | <TabAtkins> | AryehGregor: Not entirely sure. Probably prod, though. |
| 19:42 | <zewt> | sure, it helps--it's just not a big deal |
| 19:42 | <AryehGregor> | No, certainly not a big deal for most applications. |
| 19:42 | <AryehGregor> | It bloats executable size a bit, too. |
| 19:42 | <zewt> | that is, unless you're in something that needs it--I think things like video encoding can be helped a lot, too |
| 19:42 | <gsnedders> | Well, it's a bigger deal on x86 than most arches. |
| 19:43 | <jgraham> | AryehGregor: That is just a side-effect of X86_64 being less sucky that x86 though |
| 19:43 | <jgraham> | *than |
| 19:43 | <gsnedders> | Having far more GPRs is a help, and being able to rely upon everything up to SSE2 being part of the instruction set by definition helps. |
| 19:43 | <zewt> | gsnedders: heh, it's pretty ridiculous the level of hacks in the x86, though, to make the small number of GPRs not matter nearly as much as it used to |
| 19:44 | <AryehGregor> | jgraham, yeah, sure, but it's a fact anyway. |
| 19:44 | <AryehGregor> | We're talking about x86, after all. |
| 19:45 | <zewt> | ("the x86"? not sure what half-edit resulted in that) |
| 19:55 | <mpilgrim> | ok, back to the topic of unruly strings |
| 19:55 | <mpilgrim> | what's the strangest string that is a valid javascript identifier? |
| 19:55 | <mpilgrim> | in ES3 |
| 19:56 | <zewt> | look at some obfuscation contests? heh |
| 19:56 | <gsnedders> | mpilgrim: Depends on what version of Unicode the impl supports. |
| 19:57 | <mpilgrim> | excellent answer |
| 19:57 | <mpilgrim> | zewt: no, looking to test IndexedDB keyPaths, which can contain javascript identifiers separated by periods |
| 19:58 | <gsnedders> | mpilgrim: With ES5 you at least have the guarantee that every character within certain general categories in Unicode 3.0 is supported, though they can also support further characters from later Unicode versions. |
| 19:59 | <gsnedders> | mpilgrim: How about some non-ASCII letter and as many combining characters (Mn/Mc) as possible? |
| 19:59 | <mpilgrim> | love it |
| 20:00 | <AryehGregor> | Z̡ͭ̒͑ͣ̀͘͏̠̟̯̣̠̟̲̬̻̙͓̤̝̤͎̼ͅA̙̠̗̯̪̤̲̻̗͙̦̦̭͆ͨ̔͊̇̀͡͝L̢͍̠̣͈̩̠̯̆̈́ͯ̔ͦ̅̂̇̀̊̏ͦ̂͆ͦ̈́͝G̵̣̣̝͔̖̼̟̤̝͇̰̦͓̱̗̳̃͆͗ͯ͂̎̓̑̄̒̑̏͛́̀͟͞͞ͅO̷̴̧̳̭̻̤͔̔͊̈ͮ̋̄͘!̸̶̣̗͓̺̺̬̝͖̯͎͙͎̪̥͍͇̃̃̒̐ͩ̾̆ͧ́͘̕ |
| 20:00 | <AryehGregor> | Zalgo generator for your convenience: http://www.eeemo.net/ |
| 20:00 | mpilgrim | wonders if he can even copy/paste that |
| 20:00 | <gsnedders> | Or some non-ASCII letters with U+200C in the middle. |
| 20:01 | <TabAtkins> | Yay zwnj! |
| 20:01 | <gsnedders> | It's a valid identifier character (well, except for the first char)! |
| 20:01 | <AryehGregor> | Nice. |
| 20:02 | <mpilgrim> | "Unexpected token: ILLEGAL" on the zalgo text |
| 20:03 | <gsnedders> | http://es5.github.com/#x7.6 |
| 20:04 | <AryehGregor> | :( |
| 20:06 | <Hixie> | Philip`: https://bugs.webkit.org/show_bug.cgi?id=48291 might be of interest to you |
| 20:15 | <Hixie> | dude that zalgo string keeps making me think my screen has glitched |
| 20:15 | <AryehGregor> | It doesn't display right in my fonts. |
| 20:15 | <AryehGregor> | But Zalgo is pretty awesome, yeah. |
| 21:12 | <Hixie> | zcorpan: was it you who asked for TrackList to expose objects instead of having .getKind() .getLabel() etc? |
| 21:12 | <Hixie> | i can't find wherever i had that discussion |
| 21:12 | <Hixie> | (i want to admit being wrong and fix it) |
| 21:14 | <clair> | Hixie: I noticed you mentioned in the mailing list you were considering a dialogue element - I'm interested in getting into the working group stuff so is there anything I can do to help? (I'm not an implementor or anything, just a developer) |
| 21:15 | <Hixie> | aw man, if you want to help with that in paricular that would be awesome |
| 21:15 | <zcorpan> | Hixie: don't recall |
| 21:15 | <zcorpan> | Hixie: ask foolip |
| 21:15 | <Hixie> | clair: the main thing i need for that is to go through all the current libraries like jquery, etc, and see how they expose inline dialogs |
| 21:15 | <Hixie> | clair: to see what the current set of features that people expect are |
| 21:15 | <clair> | Yeah, modal stuff annoys me so it sounds a good introduction :) |
| 21:16 | <Hixie> | zcorpan: he filed a bug that i haven't responsed to, but i could have sworn i'd discussed this before with someone |
| 21:16 | <Hixie> | maybe nessy? |
| 21:16 | <Hixie> | can't find a bug about it |
| 21:16 | <clair> | And sure, I can do that, is the wiki the best place to put it? |
| 21:16 | <Hixie> | clair: the best place to start would just be to make a wiki page on wiki.whatwg.org and just list all the libraries that support this kind of thing |
| 21:16 | <eggsby> | Hello, can anyone recommend a good lib for working with canvas? |
| 21:17 | <zewt> | canvas is a pretty good library for using canvas |
| 21:17 | <clair> | Hixie: Will work on that, thanks :) |
| 21:18 | <Hixie> | clair: don't hesitate to ask people here for advice :-) |
| 21:18 | <Hixie> | clair: any help would really be truly awesome |
| 21:18 | <eggsby> | use the js api directly? D: |
| 21:19 | <clair> | Cool, I'll lurk around and get a feel for the channel - and will ping you if I come up with anything |
| 21:19 | <clair> | Happy to help :) |
| 21:19 | <Hixie> | :-) |
| 21:20 | <Hixie> | eggsby: i find http://software.hixie.ch/utilities/js/canvas/ helpful for playing with the api |
| 21:21 | <eggsby> | Hixie: but my crutch! :p Guess I should just pull up my sleeves and put my nose to the grindstone |
| 21:21 | <eggsby> | thanks for all your hard work btw Hixie |
| 21:22 | <zcorpan> | eggsby: http://blog.nihilogic.dk/2009/02/html5-canvas-cheat-sheet.html |
| 21:22 | <Hixie> | eggsby: my pleasure |
| 21:22 | <Hixie> | eggsby: depending on what you want to do, the canvas api isn't half bad |
| 21:22 | <Hixie> | eggsby: if you want to do some complex stuff, it makes sense to get a library... but then it depends what you want to do |
| 21:23 | <zcorpan> | eggsby: the 2d canvas api lacks many convenience things so you'll probably find yourself wanting to factor out stuff and thus writing your own canvas library |
| 21:23 | <eggsby> | yea, I wanted to use it for a little 2.5d browser game |
| 21:23 | <Hixie> | eggsby: e.g. there are (i believe) 2d physics libraries, but you wouldn't want to use that if you wanted to write a sprite platformer :-) |
| 21:23 | <zcorpan> | there are probably plenty of them around, but i haven't used any so can't recommend any |
| 21:23 | <Hixie> | yeah me either |
| 21:23 | <eggsby> | but yeah I might just end up rolling my own out of convenience methods |
| 21:24 | <eggsby> | maybe w/ backbone or something... alright well thanks for the suggestions :) |
| 21:25 | <jgraham> | (it's backbone more about MVC for data-driven apps? Not sure how it would help with canvas) |
| 21:25 | <jgraham> | (not that I have used it) |
| 21:26 | <zcorpan> | i wonder if the sheet cheat is outdated |
| 21:26 | <eggsby> | jgraham: backbone just provides a utility belt |
| 21:26 | <zcorpan> | it doesn't have the drawFocusRing and stuff |
| 21:26 | <eggsby> | things like events and models and such |
| 21:27 | <jgraham> | Right, it's not clear to me why moels would be releavnt to canvas drawing |
| 21:27 | <jgraham> | Or relevant to the kinds of things that you would be doing |
| 21:27 | <jgraham> | But I am speaking from a position of pure ignorance |
| 21:28 | <zcorpan> | Hixie: does the spec support updating the inband text tracks mid-stream? |
| 21:29 | <eggsby> | jgraham: yeah it would have been better to say build a game engine w/ backbone, not a canvas lib |
| 21:29 | <Hixie> | zcorpan: i think it supports the in-band text tracks changing, but doesn't let anyone know they did, or something |
| 21:29 | <Hixie> | zcorpan: i forget the exact details |
| 21:31 | <zcorpan> | ok |
| 22:01 | <zcorpan> | Hixie: http://www.jenitennison.com/blog/node/156 |
| 22:17 | <AryehGregor> | Weird, www.gov.il is a website. |
| 22:18 | <AryehGregor> | Which gov.il redirects to. |
| 22:23 | <AryehGregor> | Not as weird as "to." having an A record. |
| 22:35 | <_bga> | is it possible to allow newless creation of dom object in standard? |
| 22:36 | <TabAtkins> | Huh? You make DOM objects without a new already. |
| 22:36 | <TabAtkins> | document.createElement("foo"); |
| 22:36 | <_bga> | no |
| 22:36 | <_bga> | i want {Image()} |
| 22:36 | <_bga> | w/o new |
| 22:36 | <TabAtkins> | ...why? |
| 22:37 | <_bga> | new keyword is bit outdated. |
| 22:37 | <TabAtkins> | ...why? |
| 22:37 | <_bga> | just semantic |
| 22:37 | <TabAtkins> | ...what? |
| 22:37 | <gsnedders> | It completely changes the meaning… |
| 22:38 | <TabAtkins> | (I am actually quite confused here, not being facetious.) |
| 22:38 | <gsnedders> | It's a reference to [[Construct]] and not [[Call]]. |
| 22:38 | <_bga> | new langs allows create object w/o "new" |
| 22:38 | <_bga> | gsnedders i know |
| 22:38 | <gsnedders> | _bga: But they have a difference between classes and functions mostly. |
| 22:38 | <TabAtkins> | New languages arent' javascript, and don't have the mixing of functions and classes and constructors. |
| 22:38 | <gsnedders> | _bga: JS doesn't have classes, so you need some different way to separate out a function call and an object constructor. |
| 22:39 | <TabAtkins> | gsnedders: And even in the Classes proposal that tc39 accepted at the last meeting, you still new them up as normal. (It's just syntax sugar for the existing prototype stuff.) |
| 22:39 | <_bga> | gsnedders imho there nothing differences between create object and cast to class |
| 22:40 | <zcorpan> | as it happens most (or all) dom constructors can be called as functions with the same effect |
| 22:40 | <gsnedders> | TabAtkins: Don't even make me think about how far behind I am on es-discuss. |
| 22:40 | <_bga> | Foo(1, 2) and Foo(bar) |
| 22:41 | <TabAtkins> | Anyway, 'new' does something very specific - it sets "this" to a clone of the prototype. Just calling the function doesn't do that; "this" is set as normal. |
| 22:41 | <_bga> | gsnedders i can propose also Class.prototype = Class as standard :) |
| 22:41 | <TabAtkins> | You can make new-less calls still construct, but you ahve to do it manually. |
| 22:41 | <_bga> | TabAtkins i do it |
| 22:41 | <_bga> | my class sugar |
| 22:42 | <TabAtkins> | That's nice. It's weird, and not idiomatic javascript to do so. Outside of the oddness of some DOM constructors, it's not common for new-less constructors to work. |
| 22:43 | <jgraham> | TabAtkins: Well and some of the stdlib :) |
| 22:43 | <TabAtkins> | Yeah, some of it. |
| 22:44 | <_bga> | ok TabAtkins. idea to remove 'new' keyword is not still popular |
| 22:44 | <jgraham> | It's weird in that the syntax doesn't really support it but normal in that it happens all over the place in typical js |
| 22:45 | <TabAtkins> | It's not that it's unpopular. It's that there's no *reason* for it. "new" has a purpose, and omitting it means the dev has to do more work. It also makes the code less clear, imo, because you can no longer tell apart object construction from function calls. |
| 22:46 | <_bga> | TabAtkins i can differ function and Class |
| 22:46 | <_bga> | first letter |
| 22:46 | <_bga> | lower and upper cased |
| 22:46 | <jgraham> | TabAtkins: I'm not sure I agree with your second point |
| 22:47 | <zewt> | ... i'd say that while it works great in Python, trying to turn Javascript into Python is not going to work. heh |
| 22:47 | <jgraham> | a = func_call() can set a to a new object if func_call returns it |
| 22:48 | <_bga> | jgraham imho 'new' keyword is just semanic |
| 22:48 | <_bga> | in js - not |
| 22:48 | <_bga> | ok |
| 22:49 | <_bga> | not if you abstract from some lang |
| 22:49 | <TabAtkins> | It's not semantic. There is a functional difference. I explained it. Without "new", the dev has to do additional work in their function to clone the prototype themselves. |
| 22:49 | <jgraham> | _bga: As usual when people start using the word "semantic" I have no idea what you mean |
| 22:49 | <_bga> | TabAtkins in js. yes |
| 22:50 | <TabAtkins> | ...yes, in js. Because that's how "this" works. If you're programming in a different language with different semantics for "this", or some other concept altogether (like real classes with dedicated constructor functions), then the situation is obviously different. |
| 22:53 | <zewt> | (fwiw, as far as syntax sugar goes, being able to omit "new" is about at the bottom of what I'd bother spending time worrying about) |
| 23:05 | <AryehGregor> | "Personally, i refer and work from the official specification as it stands, w3c, because i know that it is the spec which will be implemented and, while it is still in flux, it is the one i can rely on the most to be the closest to what will eventually become standard in the years to come." |
| 23:05 | <AryehGregor> | lol? |
| 23:05 | <AryehGregor> | orly? |
| 23:06 | <TabAtkins> | It's funny when clueless people try to work up a righteous indignation. |
| 23:07 | <jgraham> | It's funny when non-implementors make gross assumptions about the needs of implementors |
| 23:08 | <jgraham> | Well I say "funny" |
| 23:08 | <jgraham> | What I actually mean is "reponsible for lots of problems" |
| 23:09 | <jgraham> | Yes, /TR/. I'm looking at you |
| 23:17 | <Hixie> | anyone know where ms2ger is these days? |
| 23:17 | <Hixie> | i don't remember seeing him recently |
| 23:18 | <TabAtkins> | He was around earlier today. |
| 23:20 | <Hixie> | k |
| 23:21 | <Hixie> | ms2ger, if you read this later, see http://www.w3.org/Bugs/Public/show_bug.cgi?id=11191 |
| 23:41 | jgraham | is increasingly convinced that createContextualFragment was a better design than innerHTML but a worse API |
| 23:41 | <TabAtkins> | The worse API being "the DOM"? |
| 23:42 | <Hixie> | jgraham: the right design imho was E4X without comments |
| 23:42 | <TabAtkins> | Yus. |
| 23:42 | <Hixie> | but not enough other people seem to agree :-( |
| 23:43 | <Hixie> | createContextualFragment() has the same major flaw as innerHTML (no compile-time checking) |
| 23:43 | <paul_irish> | XMLSerializer.serializeToString() in gecko predated elem.innerHTML in trident, right? |
| 23:44 | <Hixie> | didn't innerHTML in trident predate gecko as a whole? |
| 23:44 | <paul_irish> | more generally i had a though that innerHTML was the first thing to serialize DOM into HTML but the more i think about it, i think it wasnt first |
| 23:44 | <Hixie> | i thought innerHTML was a netscape 4.x thing from the mid 90s |
| 23:44 | <Hixie> | maybe even 3.x |
| 23:45 | <smaug____> | innerHTML is IEism |
| 23:45 | <Hixie> | ah |
| 23:45 | <paul_irish> | hm.. well i can tell you that innerHTML was ie7, so sept 97... gecko/netscape/moz didnt have it until 99 |
| 23:45 | <paul_irish> | ie4* |
| 23:45 | <Hixie> | gecko didn't exist until like 98/99 |
| 23:45 | <paul_irish> | a |
| 23:45 | <paul_irish> | hah |
| 23:47 | <paul_irish> | looks like XMLSerializer and XMLHttpRequest landed simultaneously in gecko. interesting. |
| 23:50 | <paul_irish> | Hixie: whats the compile-time checking flaw about? |
| 23:51 | <Hixie> | code like this shouldn't compile: foo.innerHTML = '<p><span title="x">bla</b>'; |
| 23:51 | <TabAtkins> | Yes. |
| 23:51 | <Hixie> | it's a syntax error, so it can trivially be caught at compile time |
| 23:52 | <Hixie> | so it should be caught at compile time |
| 23:52 | <Hixie> | mind you, for the same reason i'm a firm believe in strong typing, which js doesn't have either |
| 23:52 | <Hixie> | believer, even |
| 23:53 | <paul_irish> | so if it had compile time checking it'd throw an exception indicating the input isn't good enough? |
| 23:53 | <Hixie> | it just wouldn't compile |
| 23:53 | <Hixie> | same as function x() { foo(} ); |
| 23:54 | <paul_irish> | k |
| 23:54 | <Hixie> | e4x works like that |
| 23:54 | <Hixie> | var foo = <x></x>; |
| 23:55 | <Hixie> | var foo = <x></y>; // this line makes the script not compile |
| 23:55 | <_bga> | Hixie some new translangs supports embedded xml |
| 23:55 | <Hixie> | that's what e4x is, yeah |
| 23:55 | <Hixie> | gecko has (had?) support for it |
| 23:58 | <smaug____> | has |
