| 01:57 | <Hixie> | man, why is utf-8 so hard for libraries to get right |
| 10:13 | <annevk> | I thinks this API for events is the way to go: https://gist.github.com/4463430 |
| 10:13 | <annevk> | with the second and third argument reversed |
| 10:13 | <annevk> | and for now just make everything bubble and ignore capture altogether |
| 10:24 | <jgraham> | Is matchesSelector suppoed to be a built-in function on window, or something that a user could do in js? |
| 10:24 | <Ms2ger> | Builtin on Element, no? |
| 10:25 | <jgraham> | No, the one in the proposal is different |
| 10:43 | <hsivonen> | gsnedders: I put the <main> tests in a separate file and checked with jgraham and annevk first |
| 10:43 | <annevk> | jgraham: you would do it in JS |
| 10:44 | <annevk> | jgraham: as in the proposal |
| 10:45 | <jgraham> | annevk: Well it was marked as "syntactic sugar" so I thought you were proposing a built-in equavalent to the following JS |
| 10:45 | <jgraham> | Since sugar is usually something that you don't have to implement yourself |
| 10:46 | <jgraham> | hsivonen: I think what you did was fine fwiw |
| 10:47 | <annevk> | jgraham: ah yeah |
| 10:47 | <annevk> | jgraham: I think for Element.on we could overload and let the third argument be a string which would be interpreted as that sugar |
| 10:47 | <odinho> | Getting back a stop() function is a bit strange though? |
| 10:47 | <annevk> | odinho: why? |
| 10:47 | <odinho> | No, I just haven't seen it. s/strange/unfamiliar/ |
| 10:48 | <annevk> | I quite like it, seems much more lightweight than returning an object with a method |
| 10:48 | <odinho> | More used to getting back object I can do moar stuff on. |
| 10:48 | <darobin> | well, it can also be an object :) |
| 10:48 | <odinho> | And if we'd even want to do something else with the registered listener than stopping it? |
| 10:50 | <odinho> | Though I guess the jqueryists would actually expect on to return eventTarget itself, so you could chain. |
| 10:51 | <Ms2ger> | Boo |
| 10:51 | <odinho> | It'd be nice not having to write addEventListener all the time. |
| 10:51 | <annevk> | I want to return a function/object as the other pattern of unregistering sucks |
| 10:52 | <odinho> | annevk: True dat. |
| 10:52 | <odinho> | (I'm very for the .on it no matter how it's done) |
| 10:52 | <annevk> | Yeah, I don't want to make addEventListener more complicated |
| 10:53 | <annevk> | The only other thing I thought of was making the methods exclusive but I think that's too much inconvenience |
| 10:54 | <darobin> | note that you could return an object and allow for chaining |
| 10:55 | <darobin> | all it takes would be for the returned object to also support .on() and know what event target to send it to (which it ought to) |
| 10:55 | <darobin> | of course, I'm not saying that's necessarily a good idea :) |
| 10:57 | <annevk> | I thought it was just element.matches() these days btw? |
| 10:57 | <annevk> | darobin: sounds like a bit too much magic |
| 11:00 | <darobin> | annevk: what bothers me more than the magic is that it actually only chains .on() |
| 11:00 | <darobin> | whereas in chaining I'd expect to be able to call whatever is callable on the original event target |
| 11:00 | <darobin> | $el.on(...).css(...).appendTo(...) etc. |
| 11:01 | <Ms2ger> | Eww |
| 11:04 | <annevk> | My new hobby: finding useful return values that break chaining |
| 11:04 | <jgraham> | Isn't that the normal case? |
| 11:05 | <jgraham> | I mean typically you expect a function (or method) to return a value |
| 11:05 | <jgraham> | Returning the object that you first thought of is a rather strange idea |
| 11:11 | <annevk> | It was a joke :) |
| 11:15 | <annevk> | Minutes are so bad... http://www.w3.org/2012/05/02-webapps-minutes.html#item04 "[ Bike shedding about name ]" |
| 11:15 | <annevk> | Pretty sure we actually made a decision there |
| 11:19 | <hsivonen> | btw, has anyone tested if IE10 has an “IE9 XML” mode in addition to an “IE10 XML” mode? |
| 11:19 | <hsivonen> | I guess that’s something I need to test |
| 11:20 | <hsivonen> | Also: Modes in mobile edition of IE9 and IE10 |
| 11:21 | <hsivonen> | hmm. Google Maps for desktop renders in the IE10 Standard mode, so they don’t even have the excuse of that mode not existing on mobile |
| 11:23 | <annevk> | http://programmingisterrible.com/post/39812339693/are-16-bits-providing-at-most-65-536-distinct beautiful |
| 11:24 | <hsivonen> | hah |
| 11:24 | <darobin> | beautiful |
| 11:24 | darobin | u2665 chaining |
| 11:31 | <annevk> | also from that blog I learned about https://www.facebook.com/notes/facebook-engineering/xhp-a-new-way-to-write-php/294003943919 |
| 11:49 | <darobin> | "The correct solution is to pass this input through htmlspecialchars() before displaying it to the user." |
| 11:49 | <darobin> | dude, the correct solution is to not use PHP in the first place |
| 12:01 | <annevk> | http://programmingisterrible.com/post/39590933826/the-ghost-of-http |
| 12:01 | <annevk> | this blog is getting better and better |
| 12:02 | <annevk> | darobin: doesn't matter, every language has the same problem whenever you string concat HTML |
| 12:02 | <darobin> | annevk: it does matter |
| 12:02 | <darobin> | if you build a templating language specifically for the web and you don't make it so that it escapes HTML by default, don't be surprised that there are so many exploits later |
| 12:03 | <annevk> | fair enough I suppose, but I don't really see PHP as a templating language, it's just another language |
| 12:13 | <hsivonen> | annevk: points to the blogger for including Atom on the list of failures. We should have publishes an RSS5 instead. |
| 12:47 | <odinho> | Hmm. HTML6 is XML. Sneaky. Not even backwards compat afics. http://html6spec.com/ |
| 12:49 | <odinho> | ( http://lists.w3.org/Archives/Public/public-html/2013Jan/0016.html ) |
| 12:52 | <hsivonen> | sigh. constant vigilance needed. I thought we agreed not to do stuff like http://html5.org/tools/web-apps-tracker?from=7621&to=7622 back in 2006 after burying the “significant inline content” concept |
| 12:55 | <MikeSmith> | hsivonen: Hixie seems to be saying he intended it that way all along, for the title element at least |
| 12:56 | <hsivonen> | I wonder if he pinged glazou first |
| 12:57 | <MikeSmith> | I don't think he pinged anybody |
| 12:58 | <MikeSmith> | I think he made that change as a result of a recent IRC discussion |
| 12:58 | <MikeSmith> | I'll try to find a pointer to the logs |
| 13:19 | <MikeSmith> | hsivonen: http://logbot.glob.com.au/?c=freenode%23whatwg&s=20+Dec+2012&e=21+Dec+2012#c756715 and http://logbot.glob.com.au/?c=freenode%23whatwg&s=21+Dec+2012&e=22+Dec+2012#c757484 |
| 13:25 | <hsivonen> | MikeSmith: thanks |
| 13:27 | <hsivonen> | I think I will end up sending email to the list :-( |
| 13:29 | <hsivonen> | do we have a wiki page that explains why conneg is bad and you should not try to solve things by saying “I know! Let’s use conneg!”S |
| 13:29 | <hsivonen> | s/S/?/ |
| 13:31 | <annevk> | slightlyoff: not sure if you're on IRC, but the Event class I pointed to is the same as the DOM has plus WebKit proprietary extensions |
| 13:31 | <annevk> | slightlyoff: not actually changed for Dart |
| 13:32 | <annevk> | hsivonen: someone from Google is pushing it again: http://www.igvita.com/2012/12/18/deploying-new-image-formats-on-the-web/ |
| 13:32 | <annevk> | hsivonen: 386'ing conneg is not for me |
| 13:32 | <hsivonen> | boo. Dart’s “cleaned up” DOM still has live child collections |
| 13:33 | <hsivonen> | firstChild/nextSibling FTW |
| 13:33 | <annevk> | Part of it is "cleaned up", lots of it seems a straight copy of the existing classes/interfaces |
| 13:34 | <annevk> | oh, Eric Lawrence points out some problems with conneg in the comments |
| 13:37 | <jgraham> | Dart hasn't died yet? How sad |
| 13:38 | <annevk> | It's not like Google's bleeding money over it |
| 13:39 | <jgraham> | Well no, but it is keeping bright people from working on better things |
| 13:43 | <hsivonen> | I ended up writing a short one-off comment about Accept: https://bugzilla.mozilla.org/show_bug.cgi?id=240493#c89 |
| 13:44 | <annevk> | does webarch say anything about conneg? |
| 13:45 | <annevk> | it does |
| 14:45 | <annevk> | https://gist.github.com/4475457 |
| 14:45 | <annevk> | That includes all the syntax sugar libraries have... |
| 14:46 | <annevk> | heycam|away: are you there? I wonder if the above overloading thingie actually works... Maybe Ms2ger knows? |
| 14:46 | <Ms2ger> | dictionary<EventCallback>? |
| 14:47 | <annevk> | {click: callback1, resize: callback2} |
| 14:48 | <Ms2ger> | Hrm |
| 14:48 | <Ms2ger> | A dictionary with an open-ended set of keys? |
| 14:48 | <Ms2ger> | Because that doesn't exist |
| 14:48 | <annevk> | Yes, with string keys and values of type between < and > |
| 14:49 | <annevk> | Ms2ger: well there's a bug for that |
| 14:49 | <darobin> | Ms2ger: that doesn't exist but there's a bug for it |
| 14:49 | <Ms2ger> | Oh? |
| 14:49 | <Ms2ger> | Has someone specced iteration order? |
| 14:50 | <annevk> | Ms2ger: https://mail.mozilla.org/pipermail/es-discuss/2013-January/027842.html |
| 14:50 | <annevk> | Ms2ger: anything else? :) |
| 14:51 | <Ms2ger> | But most likely, those wouldn't be distinguishable, no |
| 14:57 | <zewt> | annevk: don't know what the point of the weird dictionary thing is; not sure what "EventFilter" is but if it's the "filter based on keys for keypress" or whatever, like I said before that seems like a complex optimization for something having no need of optimization |
| 14:57 | <annevk> | zewt: it's so you can implement your own delegation |
| 14:58 | <annevk> | zewt: e.g. based on a selector, or something else |
| 14:58 | <zewt> | you can already do it manually now within the function if you want to do it yourself |
| 14:59 | <annevk> | sure but that's less reusable |
| 15:00 | <zewt> | if(!event.target.matchesSelector("whatever")) return; |
| 15:02 | <zewt> | also, an options dictionary seems natural, instead of an infrequently-used optional parameter |
| 15:03 | <hsivonen> | OK. I started the wiki page: http://wiki.whatwg.org/wiki/Why_not_conneg |
| 15:03 | <annevk> | hsivonen: heh |
| 15:04 | <annevk> | zewt: I believe it's quite frequently used actually |
| 15:04 | <zewt> | not to channel a hixie, but are there use cases for making up your own filters other than selectors? in my experience selectors are a pretty good fit |
| 15:06 | <zewt> | elem.addEventListener("click", this.onClick, { filter: ".button" }); seems nice and clean and clear to me |
| 15:06 | <darobin> | to back up zewt's point, is there a library that filters on something other than selectors? |
| 15:08 | <zewt> | can always start with string selectors, and make it a (DOMString or EventFilter) later if wanted |
| 15:08 | <annevk> | well selectors don't make sense for non-RootNode nodes |
| 15:09 | <zewt> | what's a RootNode? |
| 15:09 | <annevk> | see DOM |
| 15:09 | <zewt> | if that means "document" then that's wrong |
| 15:09 | <annevk> | why? if you register on a document you want to filter on its elemetn descendants, same for documentfragment |
| 15:10 | <zewt> | if you register on an element you want to filter on its descendants too, why wouldn't that make sense? that's how it's used |
| 15:10 | <darobin> | yeah, that's how it's used |
| 15:10 | <annevk> | sure, RootNode includes Element |
| 15:10 | <annevk> | again, see DOM |
| 15:11 | <zewt> | then that's not what I said :P and that's a pretty broken name |
| 15:11 | <zewt> | gotta head to work, later |
| 15:11 | <annevk> | well it means Document |
| 15:11 | <annevk> | just as much as it means Element |
| 15:11 | <annevk> | and DocumentFragment |
| 15:12 | <annevk> | those can all be roots |
| 15:12 | <zewt> | it doesn't mean the root node of a document, which is what it sounds like--it doesn't make sense for every node in a tree to be called a "root node", heh |
| 15:12 | <darobin> | yeah it's not the best name ever |
| 15:12 | <annevk> | it can be renamed if you have a suggestion |
| 15:12 | <darobin> | TreeNode? |
| 15:12 | <annevk> | it doesn't really matter |
| 15:12 | <zewt> | but anyway selectors inside a document are the overwhelming majority of use cases, makes sense to me to optimize for that |
| 15:12 | <annevk> | Comment is a TreeNode too |
| 15:12 | <darobin> | not really, no, but it's very confusing |
| 15:13 | <annevk> | zewt: the proposal optimizes for that |
| 15:13 | <zewt> | but afk work for real |
| 15:13 | <darobin> | comments have children? |
| 15:14 | <annevk> | a comment is part of a Tree |
| 15:14 | <darobin> | sure, everything is :) |
| 15:14 | <darobin> | BranchNode |
| 15:14 | <darobin> | ? |
| 15:14 | <annevk> | right, that's why it's super confusing |
| 15:14 | <darobin> | NonLeafNode? |
| 15:14 | <darobin> | TrunkNode |
| 15:14 | <Ms2ger> | BlahNode |
| 15:15 | <darobin> | ContainerNode |
| 15:15 | <annevk> | ah |
| 15:15 | <darobin> | yeah, what Ms2ger said |
| 15:15 | <darobin> | anything that doesn't imply it's a root when it's not :) |
| 15:15 | <annevk> | Wikipedia suggests InternalNode and ExternalNode |
| 15:15 | <Ms2ger> | darobin, heya |
| 15:15 | <Ms2ger> | Feel like reviewing some tests? :) |
| 15:15 | <annevk> | http://en.wikipedia.org/wiki/Tree_(data_structure)#Terminology |
| 15:15 | <darobin> | Ms2ger: I will |
| 15:15 | <annevk> | oh but that doesn't work |
| 15:15 | <darobin> | annevk: don't use wikipedia for that, you'll get some computer science terminology out of it |
| 15:16 | <annevk> | yeah and it'd be wrong anyway |
| 15:17 | <annevk> | it should indicate it has the potential to have child nodes |
| 15:17 | <darobin> | MotherNode |
| 15:17 | <annevk> | uhuh |
| 15:17 | <darobin> | ParentNode |
| 15:17 | <annevk> | I guess that could work |
| 15:17 | <annevk> | the other one is named ChildNode at the moment |
| 15:17 | <annevk> | and element is in both |
| 15:18 | darobin | likes MotherNode better, but hey :) |
| 15:18 | <darobin> | "I found it and it's the MotherNode!" |
| 15:19 | <annevk> | looks like your first contribution to the DOM spec |
| 15:19 | <annevk> | at least from this generation |
| 15:22 | <annevk> | committed |
| 15:25 | <annevk> | oh hey, GitHub -> Twitter works again... |
| 15:25 | <annevk> | https://twitter.com/thedomstandard/status/288304791833161728 |
| 15:35 | <odinho> | darobin: MotherNode sounds like something cool and nongeneric, should find some other use for it. :] |
| 15:36 | <darobin> | odinho: definitely :) |
| 16:23 | gsnedders | wonders how many HTTP devices there are where gzip is significant in overhead |
| 16:29 | <darobin> | gsnedders: I'm guessing mostly old feature phones and the zigbee devices |
| 16:29 | <darobin> | (and anything like the latter) |
| 16:31 | <MikeSmith> | what's zigbee devices? |
| 16:35 | <darobin> | MikeSmith: it's IoT stuff |
| 16:36 | <darobin> | like X10, but designed this decade |
| 16:36 | <annevk> | I changed https://gist.github.com/4475457 around btw |
| 16:36 | <darobin> | they use HTTP on some ridiculously small devices |
| 16:38 | <gsnedders> | darobin: How many feature phones run full HTTP clients? |
| 16:38 | <gsnedders> | Opera Mini will at times use HTTP, but normally not. |
| 16:38 | <gsnedders> | And it's rare to see a full browser nowadays on a feature phone. |
| 16:39 | <darobin> | gsnedders: well many *have* an HTTP client, few would actually see much actual usage |
| 16:39 | <annevk> | oh yes |
| 16:39 | <annevk> | GitHub -> Twitter works again fully |
| 16:39 | <annevk> | teehee |
| 16:39 | <darobin> | gsnedders: there are quite a lot of very old Nokias out there, they all have browsers |
| 16:40 | <darobin> | but browsing the web with those is unlikely to be a very happy experience :) |
| 16:40 | <annevk> | lots of phones have browsers, but are they used? |
| 16:41 | <gsnedders> | Most phones with browsers that get used have the power that gzip isn't massively noteworthy. |
| 16:51 | <darobin> | gsnedders: agreed, but that's not the case for zigbee devices |
| 16:51 | <darobin> | that said, I'm not sure what your context is — it's quite likely that those don't matter in your case |
| 16:51 | <darobin> | on really dumb devices gzip does have an impact |
| 16:52 | <darobin> | in fact IIRC zigbee uses EXI for those |
| 17:02 | <dglazkov> | good morning, Whatwg! |
| 17:02 | <gsnedders> | darobin: My context is hsivonen's claim HTTP would've been better without the negociation for gzip |
| 17:03 | <darobin> | mmm, possibly |
| 17:05 | <darobin> | HTTP would have been better with no authoritative metadata whatsoever |
| 17:05 | <darobin> | so you drop content type and content encoding, and only use magic numbers (falling back to sniffing) |
| 17:06 | <darobin> | authoritative metadata is just a nightmare design antipattern |
| 17:09 | <annevk> | Hmm, Twitter suggests people maybe want to encode to other encodings... http://cache.gyazo.com/2ad55482c4cbcc983ec49306dc3fa47d.png |
| 17:10 | <Ms2ger> | darobin, I bet you like Content-MD5 |
| 17:10 | <darobin> | annevk: that's a lie |
| 17:10 | gsnedders | wonders how much Content-MD5 would help if everyone actually encforced it |
| 17:10 | <darobin> | Ms2ger: one of my faves! |
| 17:10 | <annevk> | darobin: Ms2ger is lying or I am? |
| 17:11 | annevk | is confused |
| 17:11 | <Ms2ger> | Eh? |
| 17:11 | <darobin> | annevk: Ms2ger always is, but it's a lie that people want to encode to anything other than utf8 |
| 17:11 | <darobin> | don't listen to the voices! |
| 17:11 | <annevk> | ooh |
| 17:11 | <gsnedders> | Yay VCS! |
| 17:11 | <annevk> | I wasn't planning on changing the API based on some rambling on twitter |
| 17:11 | <gsnedders> | bzr failing to clone because of some UnicodeDecodeError. |
| 17:14 | <annevk> | I wonder how https://bugzilla.mozilla.org/show_bug.cgi?id=757726 helps if you can still do "in" |
| 17:15 | <gsnedders> | Or just have an <object> and see if the fallback content is used. |
| 17:16 | <annevk> | all it does it seems is create a weird object |
| 17:21 | <Pomax> | is this a reasonable place for html5lib questions, or do I have to use the mailing list for that? |
| 17:22 | <Ms2ger> | Ask away |
| 17:22 | <Pomax> | I'm just trying the basic documentation example, with an import html5lib, f = open("test.html"), doc = html5lib.parse(f), but already the content of doc is <None> |
| 17:23 | <Pomax> | using the smallest test html I could come up with that's a full document with some content, http://pastebin.mozilla.org/2043951 |
| 17:23 | <gsnedders> | You almost certainly want to pass a treebuilder to it, so html5lib.parse(f, treebuilder="etree") gives you an ElementTree |
| 17:23 | gsnedders | hopes he's got the argument format right |
| 17:23 | <gsnedders> | Oh wow, I did. |
| 17:23 | <Pomax> | hm. in that case the docs definitely need a minor tweak, since the first example claims to already return a simpletree |
| 17:23 | <Pomax> | heh, nice |
| 17:24 | <Pomax> | alright, let me try that. |
| 17:24 | <gsnedders> | You basically don't want to use simpletree. We should probably just make etree the default. |
| 17:24 | <Pomax> | I was using someone else's library (bleach) that relies on html5lib but things started going tremendously wrong once I used html docs with svg elements, so I figured I'd try to track down where it goes wrong. |
| 17:24 | <jgraham> | <None> is the unhelpful representation of the root node in simpletree |
| 17:24 | gsnedders | is busy trying to fix 3to2 to get it converting Python 3 html5lib |
| 17:25 | <jgraham> | So it actually did work |
| 17:25 | <jgraham> | But didn't look like it had |
| 17:25 | <gsnedders> | jgraham: ... |
| 17:25 | <gsnedders> | I've contributed to this project for how long and didn't know this! |
| 17:25 | <jgraham> | Well None != <None> |
| 17:25 | <jgraham> | The angle brackets are everything |
| 17:25 | <gsnedders> | Well, yes, but still. Looks like some container containing just None. |
| 17:26 | <jgraham> | Well it is a side effect of the implementation of repr being "<%s>"%element.tagname and the root node having tagname of None |
| 17:26 | <jgraham> | Should just override it in that case |
| 17:27 | <gsnedders> | File a bug? |
| 17:28 | <jgraham> | It would be faster to fix the bug |
| 17:28 | <gsnedders> | Can you manually port the fix to Py3 then? |
| 17:29 | gsnedders | is sufficently far along with this he wants to just get it done without doing Py2->Py3 again |
| 17:30 | <MikeSmith> | gsnedders: about your tweet saying that hold html5lib runs much faster under latest PyPy for parsing the HTML5 spec |
| 17:30 | <MikeSmith> | how fast is that in absolute terms? |
| 17:30 | <gsnedders> | 6s locally, which is 2x speed. |
| 17:30 | <MikeSmith> | I mean, does it make it actually usable for docs of that size? |
| 17:30 | <MikeSmith> | ok |
| 17:31 | <gsnedders> | When of the ugly bits of html5lib exhibits really bad perf in PyPy though, so it should be easy enough to make that quicker |
| 17:31 | <Ms2ger> | When? |
| 17:31 | <MikeSmith> | gsnedders: OK |
| 17:32 | <gsnedders> | s/When/One/ |
| 17:32 | <MikeSmith> | without PyPy it takes a lot longer than 12s even |
| 17:32 | <gsnedders> | Then I guess I have a quicker computer than you :P |
| 17:33 | <gsnedders> | $ time python parse.py -b etree --no-html ../testdata/sites/web-apps-new.htm |
| 17:33 | <gsnedders> | real 0m13.742s |
| 17:33 | <gsnedders> | That appears to be pretty much it. |
| 17:33 | <MikeSmith> | hmm lemme try it now |
| 17:34 | <gsnedders> | web-apps-new.htm isn't in the repo, it's just http://www.whatwg.org/c yesterday evening |
| 17:35 | <MikeSmith> | real 0m11.481s |
| 17:37 | <gsnedders> | Interestingly, it's far slower with lxml |
| 17:37 | <MikeSmith> | I get 0m15.771s if I actually generate output from parse.py |
| 17:40 | <annevk> | hsivonen: fwiw, www-international is now "debating" "Why not conneg" |
| 17:43 | <MikeSmith> | annevk: "He can’t be serious, can he?" |
| 17:43 | <MikeSmith> | gotta love that |
| 17:43 | <Ms2ger> | Bwahahahahahahahahaha |
| 17:43 | <MikeSmith> | "Well, you expect shocking remarks out of WhatWG members." is good too |
| 17:43 | <annevk> | "Well, you expect shocking remarks out of WhatWG members." (Is this missing a "can"?) |
| 17:44 | <annevk> | I guess it works this way too |
| 18:23 | <TabAtkins_> | Morning, whatwg! |
| 18:23 | <TabAtkins_> | ^_^ |
| 18:48 | <annevk> | Ms2ger: step 9 in http://dom.spec.whatwg.org/#concept-event-dispatch is redundant with step 4.5 in invoke or not? |
| 18:48 | <annevk> | Ms2ger: (not step 9's substeps, those are needed) |
| 18:50 | <Ms2ger> | annevk, eh, why? |
| 18:51 | <annevk> | Ms2ger: oops |
| 18:51 | <annevk> | Ms2ger: hmm so yeah, bubbles needs to be checked either way :) |
| 18:58 | <Pomax> | hm, another html5lib question; if I want to parse an html5 doc that has embedded SVG with namespaced attributes like xlink:href, which creative mix of html5lib settings do I need to use to preserve those namespaces during parse? =) |
| 19:05 | <jgraham> | How much "like" xlink:href? |
| 19:06 | <jgraham> | There are a few special cases where namespaces are preserved, but generally HTML (and SVG-in-HTML) doesn't have namespaces |
| 19:06 | <jgraham> | (xlink:href is one such special case) |
| 19:06 | <Pomax> | "things like" being used to mean "things including" =) |
| 19:07 | <Pomax> | good point, the only namespaced attribute I found in the SVG spec attribute list for all the elements we need to support is the xlink:href one |
| 19:07 | <Pomax> | although then that still leaves the question of how to preserve it when running the html doc through html5lib. with "dom" the 'xlink: |
| 19:07 | <Pomax> | ' bit is removed |
| 19:08 | <Pomax> | at least when I print out the parse result using the "Serialization of Streams" example |
| 19:08 | <Hixie> | hsivonen: what's wrong with http://html5.org/tools/web-apps-tracker?from=7621&to=7622 ? just s/must/should/ for <title> and <option>? and why so grumpy :-P |
| 19:11 | <vcarbune> | can anyone here help me with a doubt about cross-origin video text tracks processing? |
| 19:13 | <Ms2ger> | Did people here look at http://wiki.ecmascript.org/doku.php?id=harmony:binary_data, btw? |
| 19:17 | <Hixie> | bz's feedback is orders of magnitude more complicated to deal with than almost anyone else's |
| 19:17 | <Ms2ger> | His brain is orders of magnitude more complex, too |
| 19:21 | <Hixie> | http://w3cmemes.tumblr.com/post/39943615914/image-inline-image-1 |
| 19:21 | <Hixie> | not sure what's up with the text |
| 19:26 | <annevk> | Ms2ger: shouldn't that binary_data include of the TypedArray stuff? |
| 19:26 | <Ms2ger> | I dunno |
| 19:26 | <annevk> | Ms2ger: I thought that was the idea actually |
| 19:27 | <Ms2ger> | I don't know what the idea was :) |
| 19:28 | <annevk> | I like how there's just as many opinions on es-discuss on how to design JS APIs as elsewhere |
| 19:32 | <JonathanNeal> | Is there a study available on the classnames or IDs that websites tend to use to label certain areas, like "main" for content, or the initial "wrapper"? |
| 19:33 | <Ms2ger> | Google had some of that, I think |
| 19:33 | <Ms2ger> | Maybe MOMA too |
| 19:33 | <Ms2ger> | (Or MAMA?) |
| 19:33 | <annevk> | JonathanNeal: Steve Faulkner did one recently to look for justification for the <main> element, don't have a pointer handy |
| 19:45 | <Ms2ger> | Oh, wow, navigator.plugins is in installation-time order? |
| 19:47 | <JonathanNeal> | annevk: thanks, yea, hopefully someone will find that link. |
| 19:49 | <annevk> | Ms2ger: sounds whacky |
| 19:49 | <Ms2ger> | That it does indeed |
| 19:53 | <tantek> | JonathanNeal, annevk, you might be thinking of: http://webdevdata.org/ |
| 19:53 | <tantek> | It's even a community group: http://www.w3.org/community/webdevdata/ |
| 19:58 | <stevefaulkner> | jonathanNeal http://triin.net/2006/06/12/CSS#figure-34 http://westciv.typepad.com/dog_or_higher/2005/11/real_world_sema.html http://dev.opera.com/articles/view/mama-common-attributes/#id http://lists.w3.org/Archives/Public/public-html/2012Oct/0109.html |
| 19:59 | <JonathanNeal> | Thanks stevefaulkner! |
| 19:59 | <stevefaulkner> | np |
| 19:59 | <JonathanNeal> | and thanks tantek! |
| 20:00 | <annevk> | well that was easy http://dvcs.w3.org/hg/d4e/raw-file/tip/source_respec.htm |
| 20:00 | <JonathanNeal> | container is much more popular than wrapper. |
| 20:00 | <annevk> | thanks Jacob / Travis (if you read this :)) |
| 20:00 | <JonathanNeal> | I hardly see any "inset" either, and that's my goto name. |
| 20:05 | <annevk> | Which specification uses VoidFunction from IDL? |
| 20:11 | <Hixie> | anyone got IE10 handy? |
| 20:31 | <Hixie> | you really have to love the way we use IDL interfaces to define classes, not interfaces, and then implement them as prototypes, not classes. |
| 20:33 | <TabAtkins> | To be fair, JS prototypes are the equivalent of classes. |
| 20:33 | <TabAtkins> | But the interface/class thing is silly, yes. ^_^ |
| 20:33 | <TabAtkins> | (And the "class" keyword coming in ES6 just desugars to prototype stuff.) |
| 20:38 | <dglazkov> | annevk: do you know what's the latest thinking on the prototype-swizzling story around adoptNode, document.open? |
| 20:39 | <annevk> | dglazkov: yes, I'm waiting for input from non-Gecko vendors on what they're planning on doing |
| 20:39 | <annevk> | dglazkov: plus maybe an explanation of what web components is doing in this area that has bz concerned |
| 20:39 | <annevk> | dglazkov: https://www.w3.org/Bugs/Public/show_bug.cgi?id=20567 is the tracking bug |
| 20:40 | <annevk> | we could rename interface to class... |
| 20:40 | <annevk> | doesn't really matter |
| 20:40 | <dglazkov> | annevk: custom elements don't do anything in this area. |
| 20:40 | <annevk> | it seems bz thought otherwise |
| 20:41 | <dglazkov> | annevk: I was actually hopeful that if we had a precedent for this, I don't have to do the whole renameNode-style thing for custom elements |
| 20:42 | <annevk> | does this come down to the same thing though? not exactly... |
| 20:42 | <dglazkov> | annevk: I need to look at the bugs he filed to see what his thoughts were. |
| 20:43 | <annevk> | that bug has pointers to the mailing list discussion, which is where bz provided feedback |
| 20:43 | <annevk> | but e.g. there's not going to be precedent for doing this if e.g. WebKit is not going to implement this or Opera/IE are not interested... |
| 20:43 | <annevk> | all that's somewhat clear is what Gecko does and why it does that |
| 20:45 | <dglazkov> | annevk: bz doesn't actually list any specific concerns with web components in the thread. But he did file a bunch of bugs that might have more specifics. |
| 20:45 | <annevk> | k |
| 21:24 | <Steve^> | Hi, can anyone recommend a resource that simply lists the attributes and methods of <video>? |
| 21:25 | <tantek> | Steve^: https://developer.mozilla.org/en-US/docs/HTML/Element/video |
| 21:34 | <Steve^> | tantek, it doesn't seem to list the methods |
| 21:34 | <tantek> | they're linked from that page |
| 21:34 | <tantek> | from the TOC: https://developer.mozilla.org/en-US/docs/HTML/Element/video#DOM_interface |
| 21:35 | <Steve^> | Aha! https://developer.mozilla.org/en-US/docs/DOM/HTMLMediaElement |
| 21:35 | <Steve^> | thanks |
| 21:35 | <tantek> | and currently across two pages, for <video> and in general media elements: https://developer.mozilla.org/en-US/docs/DOM/HTMLVideoElement and https://developer.mozilla.org/en-US/docs/DOM/HTMLMediaElement |
| 21:35 | <tantek> | right |
| 21:35 | <tantek> | :) |
| 21:36 | <Steve^> | I thought the world was crazy, was just me |
| 21:36 | <tantek> | you being crazy does not falsify the possibility of the world also being crazy ;) |
| 22:21 | <Hixie> | TabAtkins: i think there are qualitative differences between what i would consider a "class" and the js style oop |
| 22:26 | <TabAtkins> | I submit that the definition of "class" is very very loose, unless you're purposely just using a single language's definition of the term. |
| 22:26 | <TabAtkins> | And within the possibility-space of "things that languages call 'classes' which are reasonable impls of some sort of OO thing", I claim that JS's prototype-based stuff is a class. |
| 22:26 | <TabAtkins> | Proof by example: |
| 22:27 | <gsnedders> | TabAtkins: classes aren't just syntactical sugar, because super. |
| 22:27 | <TabAtkins> | class foo { constructor(a) { this.a = a; return this; } bar(b) { print(b); } } |
| 22:27 | <TabAtkins> | === |
| 22:27 | <TabAtkins> | function foo(a) { this.a = a; return this; } foo.prototype.bar = function(b) { print(b); } |
| 22:28 | <TabAtkins> | Modulo a few tiny bits like super which can't be strictly desugared. |
| 22:28 | <Hixie> | yeah i'm not quite sure how to put into words what the difference is |
| 22:28 | <TabAtkins> | Cause there isn't one. ^_^ |
| 22:28 | <Hixie> | there's definitely a difference between what JS does and what I think of as class-based-oop |
| 22:28 | <Hixie> | but it might be unrelated to the classness |
| 22:29 | <TabAtkins> | There are indeed differences, but every OOP implementation has differences, because OOP is a loosely-connected set of concepts. |
| 22:29 | <Hixie> | yes |
| 22:31 | <Hixie> | on another note, did we ever resolve http://lists.w3.org/Archives/Public/public-script-coord/2012OctDec/0363.html ? |
| 22:36 | <Hixie> | does webidl define what should happen here? http://software.hixie.ch/utilities/js/live-dom-viewer/saved/2061 |
| 22:36 | <Hixie> | firefox, safari, chrome: throw (various exceptions) |
| 22:37 | <Hixie> | opera: actually runs the method |
| 22:37 | <annevk> | Hixie: we did not resolve that |
| 22:38 | <annevk> | Hixie: my impression is that unless you file a bug, heycam might miss it |
| 22:38 | <Hixie> | opera doesn't even throw if you make the method 'select', which button doesn't have |
| 22:39 | <Hixie> | window.alert.call(document.body) in opera does response "WRONG THIS" though |
| 22:39 | <heycam> | annevk, Hixie, I think that should throw |
| 22:39 | <heycam> | iirc |
| 22:40 | <Hixie> | heycam: does the reason it should throw cover cases like http://lists.w3.org/Archives/Public/public-script-coord/2012OctDec/0363.html ? |
| 22:41 | <heycam> | Hixie, (will read in a bit, email exceeds my "reading simultaenous with attending meeting" bandwidth...) |
| 22:41 | <Hixie> | that e-mail exceeded my bandwidth even when i was doing nothing else, so, understood |
| 22:42 | <heycam> | :p |
| 22:43 | <SimonSapin> | It’s out! http://packages.python.org/cairocffi/ (sorry for the spam) |
| 22:46 | <annevk> | SimonSapin: doesn't explain why I'd want this drop-in replacement |
| 22:47 | <annevk> | SimonSapin: or should I know what "CFFI-based" means for that? :) |
| 22:50 | <SimonSapin> | annevk: pycairo does not install with the usual python tools, is different on python 2 and 3, exposing new cairo stuff means requiring new cairo version, it’s a pain to maintain (C code doing manual refcounting of Python objects, yay!), and nobody seems to be really maintaining it |
| 22:50 | <SimonSapin> | annevk: CFFI is an implementation detail, you don’t really need to care about it |
| 22:50 | <SimonSapin> | but it’s what makes it different from pycairo |
| 23:12 | <heycam> | Hixie, the reason I thought that example threw was because HTMLInputElement.prototype.checkValidity can't be called on an HTMLButtonElement (and vice versa) |
| 23:13 | <Hixie> | webidl says that, i assume? |
| 23:13 | <heycam> | Hixie, yes, there's a check if the object implements the interface that the operation is declared on |
| 23:13 | <Hixie> | k |
| 23:13 | <heycam> | Hixie, the origin issue is something that would need to be checked in the same place, but isn't currently |
| 23:14 | <Hixie> | k |
| 23:14 | <Hixie> | do i need to do anything to allow you to fix that one too? |
| 23:15 | <heycam> | Hixie, you might yes -- if anything it'll probably be a hook in Web IDL that defines a "security check" to be done at that point |
| 23:16 | <heycam> | Hixie, or maybe just define the origin of objects, and then I just check origins myself in that part of Web IDL |
| 23:16 | heycam | will reply on that thread |
| 23:16 | <Hixie> | k, thanks |
| 23:35 | <annevk> | bz will not be happy with an additional check there |
| 23:44 | <heycam> | why is the additional check needed? is there a way to get the value of the property despite the check that is done? (before a navigation or something?) |
| 23:45 | <Hixie> | you can get the property from one Document |
| 23:45 | <Hixie> | and apply it to another |
| 23:45 | <heycam> | oh right, I understand |
| 23:45 | <Hixie> | or take it from one Location, navigate such that the origin has changed, then apply it |
| 23:49 | <annevk> | heycam: why is it unreasonable for a subclass to supersede a method of a parent class? |
| 23:49 | <annevk> | heycam: is that some particular kind of OO people don't like? |
| 23:50 | <heycam> | unreasonable if it makes the ancestor one invalid to be called on the object |
| 23:50 | <annevk> | but why would you do that? |
| 23:50 | <heycam> | well why indeed :) |
| 23:50 | <heycam> | so I don't want to disallow overriding |
| 23:50 | <annevk> | sorry, I meant why would you call |
| 23:51 | <Hixie> | the case i most commonly run into is where the superclass implementation is "throw EAbstractMethod" |
| 23:51 | <heycam> | oh, who knows why you would -- but for languages where it's possible I think it makes sense that it still be valid to call the ancestor method |
| 23:51 | <annevk> | how would you do it in JS? |
| 23:51 | <heycam> | if you can get at the ancestor method that throws EAbstractMethod and you call it on an object that overrides that method, then I think it should still throw EAbstractMethod |
| 23:51 | <Hixie> | annevk: grab something from HTMLCollection and apply it to HTMLAllCollection |
| 23:52 | <Hixie> | heycam: and your app will crash, right |
| 23:52 | <heycam> | in some languages, yes |
| 23:52 | <Hixie> | heycam: i'm not saying it's the wrong thing to do, i'm just saying that on the web we can't let people crash the browser |
| 23:52 | <heycam> | but it's not crashing because it has an unexpected object type it's been called on |
| 23:52 | <annevk> | Hixie: and why does the this check not make that throw here? |
| 23:52 | <annevk> | Hixie: because it's a parent class? |
| 23:52 | <heycam> | I'm not saying that we should allow the browser to crash... |
| 23:53 | <Hixie> | annevk: well if there's no overridden subclass method, you don't want it to throw, right? |
| 23:53 | <annevk> | the whole inheritance thing on most classes in DOM-land is basically a shortcut for defining stuff... |
| 23:53 | <heycam> | if you define `interface A { void f(); };` then you need to write your f() prose such that it makes sense for any object that implements A, right? |
| 23:54 | <annevk> | Hixie: oops yeah |
| 23:54 | <Hixie> | heycam: in the case of the web apis, the problem is that we sometimes think in terms of interfaces and sometimes in terms of classes |
| 23:54 | <annevk> | maybe we should s/interface/class/ to get that out of the way? |
| 23:54 | <heycam> | s/implements A/is of a class that inherits from A/ then |
| 23:55 | <Hixie> | heycam: so e.g. HTMLElement is both an interface (HTMLElement.onblur is a property that means different things on HTMLAnchorElement and HTMLBodyElement), and a class (HTMLElement.click() is defined for HTMLElement and implemented identically for HTMLAnchorElement and HTMLBodyElement) |
| 23:56 | <heycam> | I don't think that's the relevant distinction, or at least, onblur ought to be defined so that its getter/setter makes sense when applied to both kinds of element |
| 23:56 | <heycam> | it's just that its behaviour is defined in terms of looking at what object "this" is when called |
| 23:56 | <Hixie> | onblur on HTMLAnchorElement is a way to set an event listener on the element, and on HTMLBodyElement is a way to set an event listener on the Window. |
| 23:57 | <annevk> | onblur you could easily define generically no? |
| 23:57 | <heycam> | right |
| 23:57 | <annevk> | just checking if it's body and then forwarding |
| 23:57 | <Hixie> | they're essentially completely different. |
| 23:57 | <Hixie> | annevk: sure. that's what i'm saying we should do for all these cases. |
| 23:57 | <heycam> | completely? |
| 23:57 | <Hixie> | annevk: either check the this, and throw if it's not the right one, or, check the this, and run the right code. |
| 23:57 | <annevk> | Hixie: heycam is saying the same, but that you do it in prose :) |
| 23:58 | <Hixie> | i don't really care where it's done |
| 23:58 | <heycam> | as long as we have well defined behaviour for when HTMLElement.onblur's setter is applied to an HTMLBodyElement, that is fine |
| 23:58 | <Hixie> | if it's prose, it's gonna require you (anne) and I to edit our specs; if it's done in WebIDL's definition, it's just heycam |
| 23:59 | <annevk> | do you need anything more than if onblur is set, these steps must be run: 1. if the context object is <body>, set a listener on Window. 2 otherwise set a listener on the context object |
| 23:59 | <annevk> | ? |
| 23:59 | <Hixie> | heycam: for that one, there's only one in the idl. for onerror, there's two, and they have different types. |
| 23:59 | <Hixie> | heycam: i don't know that we really say what happens when you call HTMLElement.onerror's setter or getter on an HTMLBodyElement. |
