<!--

rather, but yes [16:02:00.0000] oh, yeah. that's what I meant [16:02:01.0000] /me hasn't decided what to make of the "can't create cycles of blank nodes" limitation yet [16:04:00.0000] /me goes back to furiously scratching his broken arm. It's only been a day and a half, and this cast is already getting unbearable. [16:05:00.0000] what happened to your arm anyway? [16:05:01.0000] broke it while playing Snakes and Ladders (or, as Americans apparently know it, Shoots and Ladders) [16:06:00.0000] er, that's probably Chutes and Laddres [16:06:01.0000] I don't know what either is, but OK [16:06:02.0000] it's a board game [16:06:03.0000] heh? [16:06:04.0000] were you drunk? [16:06:05.0000] yes. A little. Considering that we had turned it into an awesome drinking game [16:07:00.0000] haha [16:07:01.0000] The rules were simple. If you climb a ladder, everyone else drinks. If you go down a snake, you drink. If you roll a 1, the the person to the left drinks. If a 6, then person to the right drinks. [16:07:02.0000] I hope it heals up Lachy [16:08:00.0000] The dice rule was particularly nasty because the dice was very wonky, and landed on 1 or 6 most of the time [16:09:00.0000] anyway, since it's customary here to wear socks inside when you're a guest at someone's place (even though I hate wearing socks inside), I was wearing socks on a slippery wooden floor. [16:10:00.0000] I then slipped over as I was coming back with the bottle of wine [16:10:01.0000] wacked my left elbow on the floor quite hard [16:10:02.0000] was any of this caught on camera? [16:10:03.0000] no, unfortunately [16:10:04.0000] unfortunate indeed :) [16:10:05.0000] broken elbow? [16:11:00.0000] and the other 3 who were there also didn't see me fall. They looked up when I started screaming [16:11:01.0000] yes [16:11:02.0000] smaug, now what do I do to get this checked in? https://bugzilla.mozilla.org/show_bug.cgi?id=535043 [16:12:00.0000] right/left? [16:12:01.0000] But I didn't know it was broken till yesterday, even though it happened a week ago. I was sure I had just bruised it, but decided to get it checked after it didn't heal. [16:12:02.0000] I said Left [16:12:03.0000] ahhh - I see above now ^ [16:13:00.0000] the only problem is that this kinda ruined my plans to go skiing this weekend :-( [16:26:00.0000] Over here some people make you take off your shoes when you visit, but those people are jerks. [16:27:00.0000] who wears shoes inside anyway? [16:28:00.0000] you know what I always say http://www.surfartposters.com/wp-content/images/surf-posters-SF50.jpg :) [16:28:01.0000] I lived on Kauai for 6 years and there is so much mud it really helps to kick off the shoes at the door [19:37:00.0000] where is the API that javascript uses to access the canvas? [19:39:00.0000] You mean: http://www.whatwg.org/specs/web-apps/current-work/multipage/the-canvas-element.html#canvasrenderingcontext2d [19:40:00.0000] nice [19:41:00.0000] is that javascript or c++ ? [19:42:00.0000] okay forget that question, how can I write an API for another language besides javascript? [19:42:01.0000] ^^ [20:02:00.0000] or.. where can I see the javascript source that implements the canvas functions? [20:14:00.0000] vices: You can read the source for Firefox or Webkit to see the C (or C++? I forget) that underlies the javascript implementation. [20:16:00.0000] nice! [20:16:01.0000] where is that? [20:16:02.0000] ;) [20:19:00.0000] http://mxr.mozilla.org/mozilla-central/source/content/canvas/ is for Firefox [20:20:00.0000] jcranmer: sweet, thank you sir [22:20:00.0000] Hey guys, quick question. With the HTML5 spec, you can use both strict and non-strict tags right? Much like
or
? [22:34:00.0000] jgornick: they're both strict tags now [22:34:01.0000] jgornick: they mean the same thing -- the "/" is ignored but allowed for people who prefer the XML style [22:34:02.0000] Hixie, Great, thanks! [22:36:00.0000] np [23:09:00.0000] is there any way in RDF, given a blank node with various triples having it as the subject, to add a triple that gives the blank node an identifier? [23:09:01.0000] e.g. is there a way, given (_a, X, 1), (_a, Y, 2), (_a, Z, 3), to add some triple(s) that says that _a === A? [23:10:00.0000] or is the only way to go ahead and change all the triples to just use A instead of refering to the blank node? [23:23:00.0000] Hixie: _a = A, or _a A [23:23:01.0000] ah, thanks [23:23:02.0000] so the predicate http://www.w3.org/2002/07/owl#sameAs is some sort of RDF internal magic predicate? [23:26:00.0000] hmm, I don't know if you can call it 'internal' or 'magic', but it does have the = shortcut in Notation3: http://www.w3.org/DesignIssues/Notation3.html [23:27:00.0000] is there any way in RDF to say "any blank node that is the subject of a triple with the predicate B and object C is also the subject of a triple with the predicate X and the object X"? [23:27:01.0000] sure, it can be used to change how triples are processed; but so could knowing that x is transitive in a x b; b x c [23:29:00.0000] that'd be an implication, I guess [23:29:01.0000] i'm trying to work out whether i can write http://www.whatwg.org/specs/web-apps/current-work/multipage/microdata.html#conversion-to-rdf by just giving some RDF that people can just load up and have it Just Work [23:29:02.0000] instead of (as now) requiring that they add extra magic to their code [23:30:00.0000] paragraphs 2, 3, and 4 can use the #sameAs thing [23:30:01.0000] but 1 seems harder [23:59:00.0000] /me looks into the itemref="" loop issue [00:05:00.0000] you know you're making a mistake when you start writing "x is true if the following algorithm does not terminate:" [00:05:01.0000]

itemscope itemprop=x>

[00:06:01.0000] yeah [00:09:00.0000]

[00:10:00.0000] my favourite case is

[00:11:00.0000] A Q graph [00:19:00.0000] oh yay, itemref loops [00:22:00.0000] itemref loops? [00:23:00.0000] Lachy: I don't think Manu has been spreading more "misinformation" about microdata than "we" have about RDFa. all misunderstandings in good faith as far as I can see [00:24:00.0000] I have trouble taking talk about microdata's being unstable in good faith when RDFa 1.1 is planning incompatible changes for an ecosystem that already ignores versioning [00:25:00.0000] i wonder how i ended up having mark_a_morgan blocked on twitter [00:26:00.0000] is there some way to find out who you have blocked on twitter? [00:27:00.0000] Hixie: owl:sameAs if for individuals, I think equivalent predicates should use owl:equivalentProperty (why OWL has several concepts of sameness I don't know) [00:28:00.0000] where "individuals" is nodes in the graph I guess, as opposed to the relations between them [00:33:00.0000] you use owl:sameAs when there really is just one thing you're talking about; ie. it's a strong claim [00:33:01.0000] so is there some way to recast http://www.whatwg.org/specs/web-apps/current-work/multipage/microdata.html#conversion-to-rdf into pure RDF (no prose), and have it mean the same thing? [00:33:02.0000] it's a way of dealing with situations where there are multiple URIs that are really just names for same thing; or mentions of same thing that don't use URIs [00:34:00.0000] /me reads [00:34:01.0000] paragraph 1 being the hard one [00:35:00.0000] 2, 3, and 4 could just be straight-forward sameAs statements [00:35:01.0000] i was about to say, ... [00:35:02.0000] 1st one is the hard one ;) not sure what you're doing there exactly [00:35:03.0000] and the last three, I'd suggest owl:equivalentProperty rather than owl:sameAs [00:37:00.0000] owl:sameAs would mean that everything true of the former is true of the latter; this makes it hard to keep admin metadata, eg. html5:author might have a different properties to ccrel:attributionName (eg. their textual labels, translations etc.) [00:38:00.0000] as long as you can dump it into a triplestore and make queries using ccrel:attributionName all is well [00:38:01.0000] why are the initial urls so odd (ie. embedding / escaping other urls?) [00:39:00.0000] foolip, well you have to watch out for other queries that might give bogus results, eg. if you asked for all properties that had been made by the Creative Commons, you should only find the CC ones, not the whatwg/html5 ones... [00:39:01.0000] owl:equivalentProperty will work there; owl:sameAs will muddle things up [00:39:02.0000] the first one is trying to take the pattern "_ foo X" for a particular value of "foo" and make _ be X [00:39:03.0000] owl:sameAs seems inappropriate from my reading of it, yes [00:39:04.0000] i guess i could say that "foo" is sameAs sameAs [00:40:00.0000] the first one is sameAs, yeah [00:40:01.0000] identity reasoning is a pain in the butt, but unavoidable without a URI God to compel everyone to use exactly same URIs for everything [00:41:00.0000] but how to you capture the condition that if (bla) then x sameAs y ? [00:41:01.0000] (still responding to scrollback) i don't understand why sameAs is wrong for the other three. I want to make them exactly identical, so that it's like nobody ever used the one in the spec, and the other one is used everywhere [00:41:02.0000] the URLs are odd because they're the result of converting the short names used in microdata to the URL form used in RDF [00:42:00.0000] oh, i understand the owl:equivalentProperty thing now [00:42:01.0000] ok [00:42:02.0000] Hixie: is the appending of # and : inspired by some RDFy scheme? [00:42:03.0000] no [00:42:04.0000] I'll just let you reply to that email so everyone can see the answer :) [00:42:05.0000] it's needed to make it impossible to end up with the URLs in another way [00:42:06.0000] (i forget the details) [00:43:00.0000] hixie, if you make them exactly identical, then all properties of the latter become properties of the former; eg. re dc:title [00:43:01.0000] [00:43:02.0000] Title [00:43:03.0000] A name given to the resource. [00:43:04.0000] [00:43:05.0000] 1999-07-02 [00:43:06.0000] 2008-01-14 [00:43:07.0000] [00:43:08.0000] [00:43:09.0000] A second property with the same name as this property has been declared in the dcterms: namespace (http://purl.org/dc/terms/). See the Introduction to the document "DCMI Metadata Terms" (http://dublincore.org/documents/dcmi-terms/) for an explanation. [00:43:10.0000] [00:44:00.0000] ... all that from the dc namespace would also be asserted for http://www.w3.org/1999/xhtml/microdata#http%3A%2F%2Fn.whatwg.org%2Fwork%23%3Atitle [00:44:01.0000] ah [00:44:02.0000] ... and people running queries like 'find me property URIs from Dublin Core issued before 2000' would find your URIs; while we could find wiggleroom to justify this, it would cause general confusion I'm sure [00:44:03.0000] so presumably i also want equivalentProperty sameAs for the other one, not sameAs sameAs? [00:44:04.0000] reading RDF/XML is such a thrill [00:45:00.0000] yes please [00:45:01.0000] k [00:45:02.0000] i'll dump this in when i'm done fixing the itemref issue [00:45:03.0000] and yes, owl:sameAs is over-used a bit. The linked data crowd liked the name but sometimes assert it between similar things... [00:45:04.0000] oh btw - [00:46:00.0000] I started drafting towards an rdfa tutorial. it is far from done, idea was to try to explain basic processing / structuring model without mentioning markup too much [00:46:01.0000] perhaps it would work for microdata too? I haven't studied detailed diffs between the two [00:46:02.0000] -> http://wiki.foaf-project.org/w/RDFaTutorial [00:47:00.0000] i'd like to cut it down to a page that gave the basic idea: that each point in the doc can be 'about' something (= itemscope in your terms?) [00:47:01.0000] and that the attributes either change the focus/scope, or give properties and relationships ... [00:48:00.0000] i think rdfa and microdata are different enough that readers would be better off not trying to learn both at once [00:48:01.0000] we should add owl:equivalentProperty foaf:name and a bunch of other triples to make hcard data more useable to RDF processors [00:48:02.0000] Hixie, perhaps not learn both syntaxes at once, but learn what they have in common? [00:49:00.0000] /me always reminded of http://s173.photobucket.com/albums/w57/spn_imgs/?action=view¤t=blahblah.jpg&newest=1 - old larson cartoon [00:49:01.0000] foolip: i'd rather just make it equivalent to an existing vCard RDF vocabulary, and let the RDF guys worry about how that matches FOAF [00:49:02.0000] the existing vCard RDFs are in flux ... the 90s one is pretty broken and a new one is 'in the works' [00:49:03.0000] oh, I didn't know there was a vCard RDF vocab, that sounds better though [00:50:00.0000] you could add both harmlessly enough [00:50:01.0000] danbri: fundamentally RDFa and microdata present the same underlying triple idea in very different ways [00:51:00.0000] danbri: e.g. in microdata the presentation is more about groups of name-value pairs, where values can be further groups, or strings, or urls (or dates/times) [00:51:01.0000] danbri: whereas in RDFa the presentation is more about graphs [00:51:02.0000] danbri: of URLs [00:52:00.0000] "It provides a simplified version of the basic factual claims that a page makes, stripped of flashy colours, pretty pictures or seductive language." <- surely one can express lies just as well in RDF as in natural language? [00:52:01.0000] danbri: in particular, with itemref="" in microdata you can even declare a bunch of name-value pairs without saying what group they're part of, and multiple groups (itemscope="" items) can import in those name-value pairs [00:53:00.0000] interesting, do you have an example of where that's useful? [00:54:00.0000] it's not really useful for multiple items to use the same properties per se, but the use case for it is the same as the use case in RDFa for having about="" listed in several places [00:54:01.0000] sounds like a kind of inheritance mechanism (but scoped to a single document I guess?) [00:54:02.0000] with the same value [00:54:03.0000] Hixie: that last part about sharing subtrees between several items seems dubious though, certainly it means the model isn't a tree any longer but a directed graph (I think) [00:55:00.0000] e.g. see http://damowmow.com/playground/microdata/004/review-annotated.html [00:55:01.0000] foolip: depends what you consider the nodes [00:55:02.0000] if you can fix itemref loops without breaking that though, fine [00:55:03.0000] I did break it in my solution [00:56:00.0000] danbri: (except, if you look at that solution, pretend that is really an itemref="" attribute next to the itemscope="" attribute) [00:57:00.0000] /me not quite seeing the difference [00:57:01.0000] as what? [00:58:00.0000] ah maybe i misunderstood your 'can import in those name-value pairs' [00:58:01.0000] here's another way of showing the difference [00:58:02.0000] they're linked by a named property/relationship, not copied directly in? [00:58:03.0000] in microdata, if you say

, there are two empty groups of name-value pairs with no properties [00:59:00.0000] in RDFa, if you say

, there is no data present. [00:59:01.0000] (folip, yes, it's important to be able to lie ... I was trying to capture that these property/value language strip out a lot of the human richness from docs, to dumb them down to computer level) [00:59:02.0000] /me nods [01:00:00.0000] now it happens that microdata has a (lossy) mapping to RDF, and RDF has a (lossy, implied) mapping to microdata, but it's not the point of the mechanism [01:01:00.0000] unlike RDFa where the conversion to RDF is lossless and is the point of the mechanism [01:01:01.0000] what's lost when you go from microdata to rdf? [01:02:00.0000] you can't refer to something with no properties in a triple store [01:02:01.0000] there's no triple [01:03:00.0000] you also lose the untyped items [01:03:01.0000] and currently you lose some type information from elements, though that could be fixed [01:04:00.0000] my three legged cat is climbing on top of my bike [01:04:01.0000] this could go very poorly [01:04:02.0000] hmm, I could say that http://danbri.org/foaf.rdf#danbri is a foaf:Person ... [01:04:03.0000] Yeah, he'll never reach the pedals [01:04:04.0000] she's now sitting on the seat [01:04:05.0000] and that it has a http://example.com/fooprop of http://example.org/barvalue [01:04:06.0000] and say nothing more about barvalue [01:05:00.0000] is that what you mean by a thinkg with no properties? do you count "incoming" properties as its properties? [01:05:01.0000] since often the naming/direction of a property is pretty arbitrary [01:05:02.0000] 'made' vs 'maker' etc [01:06:00.0000] /me grins re cat [01:06:01.0000] ok, back. sorry, had to take a picture of the cat doing this or my girlfriend would never believe me. [01:07:00.0000] danbri: consider

-- it's an item, but it has no type, no properties, and no identifier. [01:07:01.0000] /me looks for a numLegs property in http://hyperdata.org/xmlns/pets/hedwig.xml ... fails [01:08:00.0000] danbri: it can exist in microdata (the DOM API can reference it), but I can't see how it can exist in RDF. [01:08:01.0000] ok, it's a placeholder piece of markup? [01:08:02.0000] could be, yeah [01:08:03.0000] oh, got you [01:09:00.0000] so i guess if we distinguish APIs to the markup, from APIs to 'what the markup tells us', we get both [01:09:01.0000] RDF APIs won't tell you the grouping clusters that things came in, in their source graphs [01:09:02.0000] they tend to take pride in protecting people from needing to care about that [01:09:03.0000] yeah [01:09:04.0000] i don't think this is a failure of either model [01:09:05.0000] but markup APIs obviously need that too, eg. to make an editor [01:09:06.0000] /me neither [01:09:07.0000] but it's a difference [01:10:00.0000] in most practical applications, these differences would likely not come up much -- you don't exactly need to express an empty item often [01:10:01.0000] anyway, this is one of the examples of the reason i think it's best not to try to explain microdata and RDFa models together [01:11:00.0000] so this is more a difference between the packet of deliverables the RDFa project and the Microformats project give you; former gives you markup, triples model, and maybe eventually an API to the triples; latter gives you markup, triples model, and an API to the markup [01:11:01.0000] i think it might confuse people more than help, as they might try to apply one model to the other, and get confused [01:12:00.0000] so, in both models, we are saying ... that any bit of markup can be scoped to be 'about' some specific thing of interest; and we might or might not identify that thing specifically. but regardless, it is a grouping construct for a collection of properties/links that describe it. [01:12:01.0000] do you mean "microformats" there or "microdata"? [01:12:02.0000] yeah sorry [01:12:03.0000] s/microformats/microdata/ [01:13:00.0000] ah ok [01:13:01.0000] mini-info, tiny-factoids, etc :) [01:13:02.0000] i wouldn't really say microdata is a triples model [01:13:03.0000] it's more a nestable name-value pair list model [01:13:04.0000] but it does (perhaps grudgingly) give you the triples [01:13:05.0000] it can be converted to triples, yeah [01:13:06.0000] so is any rdf syntax [01:13:07.0000] like rdf can be converted to name-value lists [01:13:08.0000] but it's not a perfect translation [01:14:00.0000] e.g. you can't express an empty list as triples sanely [01:14:01.0000] people didn't make RDF 'cos they loved triples; they made it cos they wanted to get a big pile of name-value metadata systems talking to each other [01:14:02.0000] yup, that's the syntax / model distinction again [01:15:00.0000] also, re your "in both models" statement above, it's possible to make name-value pairs in microdata without scoping it to a subject (item): [01:15:01.0000] in RDF too, kinda [01:15:02.0000] I can write: [01:16:00.0000] e.g.

...

-- q and r could be referenced from other items also [01:16:01.0000] so they're not scoped to A [01:16:02.0000] (actually that's the case even if they _are_ scoped to A) [01:16:03.0000] Ian Hixon3 .... [01:16:04.0000] ... ie. 'there is something, ... owned by hixie, ... with 3 legs' [01:17:00.0000] in the microdata example above, id=q doesn't say "there is something with Q=...", it only says "Q=..." [01:17:01.0000] if nothing references it, then nothing has Q=... [01:17:02.0000] (and there's a conformance error, but that's another problem) [01:17:03.0000] and you're finding that construct useful? [01:18:00.0000] it proved to be a less confusing way of allowing properties of a single item to be scattered across a document than having each one be scoped to an item and for all those items to share the said itemid [01:19:00.0000] but in that case they are properties of some item? [01:19:01.0000] yeah [01:19:02.0000] the one(s) that referenced them [01:19:03.0000] (or item means the grouping construct, not the thing described?) [01:19:04.0000] item means the thing with itemscope="" [01:19:05.0000] the list of name-value pairs [01:20:00.0000] beginning to think we need a testcase repository [01:20:01.0000] e.g. you can do

...blabla... ...blabla... [01:20:02.0000] irc is a bit cramped for this sort of detailed comparison [01:21:00.0000] with the RDFa-style you would do just ..blabla... ...blabla... [01:21:01.0000] for rdf/xml we used http://www.w3.org/2000/10/rdf-tests/rdfcore/ ... basically a collection of rdf/xml documents plus normalised triples extracted, and some working-group metadata [01:21:02.0000] (but when we did the usability tests we found people got far more confused by that than the itemref thing) [01:22:00.0000] (even though it seems simpler) [01:22:01.0000] (it is also harder to expose in the DOM, since there's no longer a single element that maps to an item) [01:22:02.0000] i'm confused by both here! using real examples rather than 'a', 'q', 'foo' etc helps, esp as i don't know your syntax well [01:23:00.0000] care to bash a full example into a wiki somewhere? [01:23:01.0000] both are intending to express the triples X,a,1 and X,b,2 [01:23:02.0000] ...and the human prose wrapped around those two triples is in same bit of the doc, or different bits of the doc? [01:24:00.0000] different [01:24:01.0000] (how do i get an url to this conversation btw? there's a logger bot with #-offsets somehow?) [01:24:02.0000] see /topic [01:26:00.0000] Hixie: were you aware of IE running [07:00:02.0000] Give me an example, so I can look it up in the spec [07:01:00.0000] cyberix: I assume you mean functions that the spec defines, which can be called from anywhere? [07:01:01.0000] See e.g. http://www.whatwg.org/specs/web-apps/current-work/multipage/browsers.html#the-window-object for lots [07:04:00.0000] I just realized we should probably have one player for each frame [07:05:00.0000] which means that, eventually one guy is going to code a drum machine, and another guy is going to code a synth [07:05:01.0000] and they will come to us asking how they can synchronize them [07:05:02.0000] even though they run in different frames [07:08:00.0000] Philip`: Should probably think this trough [07:09:00.0000] Philip`: However now I'm off to write my thesis [07:09:01.0000] have fun [07:09:02.0000] /me should probably do the same, eventually [07:09:03.0000] :-) [07:11:00.0000] Hrm. I'm wondering if it's worth it to figure out how to use Strawberry Perl on my desktop, or if I should just throw the perl up on my server and run it there instead. [07:15:00.0000] What's Strawberry Perl? [07:16:00.0000] apparently a good windows distribution of perl? [07:16:01.0000] Hmm. How about you just use a non-broken OS for your desktop? :) [07:16:02.0000] Company policy. [07:18:00.0000] However, I also can't get this running on my linux server. [07:18:01.0000] Philip`: How I run teh codez? (for font-optimizer) [07:20:00.0000] TabAtkins: First, you download it [07:20:01.0000] TabAtkins: Then, you run it [07:20:02.0000] These instructions are useless lies. [07:20:03.0000] Couldn't be simpler :-) [07:23:00.0000] Do you have a specific problem with it? [07:24:00.0000] Yes. It keeps claiming that it can't find Font/TTF/Font.pm [07:25:00.0000] Which would probably be because there is no Font/TTF in the archive i downloaded. [07:25:01.0000] It should be in the ext directory [07:25:02.0000] Indeed, I have found it buried deep within. [07:26:00.0000] ext/Font-TTF/lib [07:26:01.0000] and subset.pl says: use lib 'ext/Font-TTF/lib'; [07:26:02.0000] so it should find that file [07:28:00.0000] It did not. I was able to drop TTF into the Font folder in one of the spots in my @INC path, though, and it worked. [07:29:00.0000] You shouldn't have to do that [07:30:00.0000] and you shouldn't do that because you'll forget to update it if the code's ever updated [07:30:01.0000] Shrug. I dunno what to tell you. I r noobz at perl. [07:36:00.0000] How do I delete an element from a document using DOM methods? [07:37:00.0000] elem.removeChild(child) [07:37:01.0000] So elem.parentNode.removeChild( elem )? [07:37:02.0000] That seems roundabout. [07:38:00.0000] https://developer.mozilla.org/En/DOM/Node.removeChild [07:38:01.0000] also, ##javascript [07:40:00.0000] AryehGregor: Yeah, somewhat. Although having elem.removeFromParent seems somewhat unnecessary [07:40:01.0000] Why not just elem.remove()? [07:40:02.0000] I would expect that to remove a child [07:40:03.0000] Weird. [07:40:04.0000] Not if it had zero arguments [07:41:00.0000] (in the absence of other clues, like a removeChild method) [07:41:01.0000] How about document.removeNode(node) then? [07:41:02.0000] Hum, no, wouldn't work outside documents [07:41:03.0000] Dashiva: Only works on children of that node [07:41:04.0000] /me realizes that was dumb, and that he should pasy attention [07:42:00.0000] A dumb comment to a dumb suggestions seems par for the course :) [07:42:01.0000] Yes you do look pasty [07:42:02.0000] /me looks blankly at jgraham over his monitor [07:42:03.0000] /me assumes that was a freudian slip [07:42:04.0000] /me slouches back below his monitor [07:42:05.0000] For some loose definiton of "freudian" [07:43:00.0000] That doesn't necessarially imply horses, penis envy, or wanting to sleep with your mother [07:43:01.0000] The "white male with strong opinions" definition? :) [07:43:02.0000] /me shakes head at James... [07:44:00.0000] You would just mention _those_ bits of Freud's work. [07:44:01.0000] What, the famous bits? [07:44:02.0000] Well, yes. [07:44:03.0000] I wasn't aware of the horses [07:45:00.0000] /me gets his education from pop songs [07:45:01.0000] /me gets his education from Wikipedia [07:45:02.0000] (in this case "Judy and the Dream of Horses") [07:46:00.0000] I wonder ponies have horse envy [07:46:01.0000] /me gets his education from IRC [07:48:00.0000] Philip`: How did you learn about IRC? [07:48:01.0000] ah, freud, probably the most prolific and influential case of experimental bias ever :) [07:51:00.0000] Dashiva: I don't remember [07:51:01.0000] That experience is lost in the mists of time [07:51:02.0000] Maybe you were born here [07:51:03.0000] For all I know, I could have learnt it from IRC [07:52:00.0000] Maybe his future self will learn about time travel from IRC and teach his past self about IRC [08:05:00.0000] whoa! Since when has Apple been shipping a DejaVu derivative font called Menlo? [08:05:01.0000] new in Snow Leopard? [08:11:00.0000] hsivonen: http://arstechnica.com/apple/news/2009/06/font-changes-coming-to-mac-os-x-snow-leopard.ars [08:14:00.0000] daedb: thanks [08:16:00.0000] daedb: thanks [08:17:00.0000] Hixie: the new characters for ⟨ and ⟩ have a much worse real-world font situation than the old ones [12:44:00.0000] Philip`: I don't think we need binary buffers for sound [12:45:00.0000] Philip`: I just ate lunch with the main Kunquat developer, and we discussed the matter [12:45:01.0000] Even while coding C it is just typical to use float arrays to represent sound [12:47:00.0000] cyberix: I think the problem is that in JS (in some (all?) implementations), a float is a garbage-collected heap-allocated object [12:48:00.0000] which is a lot of overhead, when you have a hundred thousand of them per second [12:49:00.0000] (compared to a specialised array with a 4-byte-per-element backing store) [12:56:00.0000] Philip`: It'd the be job for the js engine writers to improve it [12:57:00.0000] then [12:57:01.0000] I don't think it is going to block anything [12:57:02.0000] it is just going to consume lots of resources [12:57:03.0000] that is ok [12:57:04.0000] there is always two options for that kind of problem [12:57:05.0000] one is to buy a more expensive computer [12:58:00.0000] and the other is to improve js [12:58:01.0000] :-) [12:58:02.0000] I think canvas uses this funny type because it can then access graphics acceleration [12:58:03.0000] standards like opengl define operations on graphics [12:59:00.0000] it is then easy to take that set of operations and reveal it to the use [12:59:01.0000] user [12:59:02.0000] in sound such a standard set of operations has not been defined [13:00:00.0000] and trying to do it on demand would probably not lead to very good results [13:03:00.0000] cyberix: Yeah, this probably isn't an important point [13:04:00.0000] The best approach is likely to be to list and describe use cases, and come up with an initial design, and get interest from implementors, and then iterate on the design [13:08:00.0000] If a page has frames [13:08:01.0000] do those frames have a Windows object? [13:08:02.0000] -s [13:09:00.0000] Every browsing context has its own Window object, and every page and frame and [08:26:00.0000] So how is that not parsing as a URL? It's just a URL that doesn't use the hierarchy syntax. [08:26:01.0000] RFC 3986 allows those as URIs. [08:26:02.0000] Basically just a leading "something:" following by anything you like, and then a fragment. [08:26:03.0000] AryehGregor: Because we'd have to separate it up into segments, and would only apy attention to path [08:28:00.0000] Well, you do. The path is just one segment, namely, the whole URL other than the initial "data:" and the trailing fragment. [08:28:01.0000] And query part [08:28:02.0000] Yeah, that too. [08:28:03.0000] Was about to say. [08:28:04.0000] AryehGregor: No, we don't. [08:28:05.0000] Look at that example above [08:28:06.0000] AryehGregor: The whole thing is in the host [08:28:07.0000] What about it? The path is "//foobar\ndocument.write('a')". [08:28:08.0000] No, there's no host. [08:29:00.0000] How not? [08:29:01.0000] A path can't start with // [08:29:02.0000] Because an authority starts with // [08:29:03.0000] URI = scheme ":" hier-part [ "?" query ] [ "#" fragment ] [08:29:04.0000] /me thinks AryehGregor has got his ABNF wrong [08:29:05.0000] hier-part = "//" authority path-abempty [08:29:06.0000] / path-absolute [08:29:07.0000] / path-rootless [08:29:08.0000] / path-empty [08:29:09.0000] path-rootless = segment-nz *( "/" segment ) [08:29:10.0000] It doesn't match path-absolute, path-rootless, or path-empty [08:29:11.0000] segment-nz = 1*pchar [08:29:12.0000] It matches path-rootless. [08:29:13.0000] pchar doesn't contain / [08:30:00.0000] Hmm. [08:30:01.0000] Otherwise that would be ambiguous for cases like //gsnedders.com [08:30:02.0000] Touche. [08:30:03.0000] javascript: URLs are just an abomination, we shouldn't try to treat them otherwise [08:30:04.0000] The only alternation there that can start with // is "//" authority path-abempty [08:30:05.0000] Okay, you win. [08:31:00.0000] data: and javascript: don't follow URI syntax at all, so no reason to quibble over the fragment or query parts. [09:57:00.0000] Whoops [09:57:01.0000] I think I broke the spec [09:58:00.0000] nice goin! [10:02:00.0000] Oh, it's okay now [10:02:01.0000] /me has moved the multipage thing to a new server and hopes it won't break much [10:03:00.0000]