| 00:58 | <Hixie> | wow there is really no freaking interop between browsers on Location |
| 01:02 | <Hixie> | wow, IE actually lets you detect which properties are on another origin's Location object |
| 01:03 | <heycam> | that sounds bad |
| 01:04 | <Hixie> | parent.location instanceof Location => false in IE, true in Gecko, false in Opera, and there's no Location object in WebKit. |
| 01:04 | <heycam> | should be false in Gecko once Location is implemented with new Web IDL bindings |
| 01:05 | <Hixie> | Location.test = 1; w(parent.location instanceof Location); w(Location.test); w(parent.location.test) => true, 1, undefined in Gecko |
| 01:05 | <Hixie> | wtf Gecko |
| 01:06 | <Hixie> | wait, i'm on crack |
| 01:06 | <Hixie> | i should have been doing Location.prototype.test = 1 |
| 01:06 | <Hixie> | and gecko throws if i do that... |
| 01:07 | <heycam> | hmm |
| 01:50 | <Hixie> | abarth: https://bugs.webkit.org/show_bug.cgi?id=83493#c20 |
| 01:51 | <abarth> | Hixie: hi |
| 01:52 | <Hixie> | abarth: haven't yet paged in the details here but that bug has a question for you |
| 01:52 | <abarth> | Hixie: answered |
| 01:52 | <Hixie> | cool |
| 01:52 | <Hixie> | thanks |
| 01:53 | <Hixie> | anything to report on ancestorOrigins btw? did it work out as expected? |
| 01:53 | <abarth> | yes, the people for whom it solves a problem are very happy with it |
| 01:54 | <abarth> | it certainly doesn't solve every problem though |
| 01:54 | <abarth> | the main people who like it are people who write widgets |
| 01:54 | <abarth> | and want to know something about their embedded context |
| 01:54 | <Hixie> | cool |
| 01:54 | <Hixie> | do you know of any implementation in other UAs? |
| 01:54 | <Hixie> | annevk: you around? |
| 01:55 | <abarth> | I don't know of any |
| 01:55 | <abarth> | in other user agents, you can try to access things and catch security exceptions |
| 01:55 | <abarth> | so there's less of a need |
| 01:55 | <abarth> | but that doesn't help with the clickjacking use cases |
| 01:55 | <Hixie> | oh, why doesn't that work in webkit? |
| 01:56 | <abarth> | WebKit doesn't throw exceptions for security errors |
| 01:56 | <abarth> | in some cases |
| 01:56 | <abarth> | and it logs to the console when there is a security error |
| 01:56 | <abarth> | the widget authors don't like the console spam |
| 01:56 | <abarth> | because people who embed the widgets complain about it |
| 01:58 | <Hixie> | oh |
| 01:58 | <Hixie> | why not fire an exception? |
| 01:58 | <abarth> | two reasons: |
| 01:59 | <abarth> | 1) we're worried about breaking content that doesn't expect the exception |
| 01:59 | <abarth> | 2) othermaciej is concerned about making it too easy to probe whether security checks will pass or fail |
| 01:59 | <abarth> | (i'm not really sold on (2), so I might not be explaining it the most clearly) |
| 02:00 | <abarth> | i've got to run |
| 02:00 | <Hixie> | laer |
| 02:00 | <abarth> | l8r |
| 02:01 | <TabAtkins> | Damn punk kids and your txt spk |
| 02:36 | <othermaciej> | abarth: I am not all that concerned about it, certainly not enough to outweigh any actual pragmatic benefit in a given case if there is one |
| 02:49 | <Hixie> | abarth: ok so looks like ancestorOrigins in webkit uses a DOMStringList, and isn't blocked by sandboxing, is that right? |
| 02:51 | <abarth> | othermaciej: oh, maybe I misunderstood the earlier discussion. sounds like (1) is the main issue then |
| 02:51 | <abarth> | Hixie: yes |
| 02:51 | <othermaciej> | abarth: I don't know what you mean by "the earlier discussion" |
| 02:52 | <othermaciej> | abarth: I personally do think it is a bad idea to return informative exceptions to the caller when a security violation occurs, but not enough to fight the weight of other implementations, or what have you |
| 02:52 | <abarth> | its from a message you wrote on Aug 13, 2010 |
| 02:53 | <abarth> | let me find the link |
| 02:54 | <abarth> | http://lists.webkit.org/pipermail/webkit-dev/2010-August/013883.html |
| 02:54 | <abarth> | sounds like you weren't overly concerned then either |
| 02:55 | <abarth> | but just raising it as a possible issue |
| 05:42 | <Hixie> | in webkit, does securityOrigin()->toString() return a unicode origin or an ascii origin? |
| 05:44 | <Hixie> | (abarth?) |
| 06:58 | <MikeSmith> | Hixie: I looked at the sources myself to try to figure out but can't tell from that either |
| 06:59 | <MikeSmith> | http://trac.webkit.org/browser/trunk/WebCore/platform/SecurityOrigin.h?rev=30184 |
| 06:59 | <MikeSmith> | I can't tell from that at least |
| 07:13 | <MikeSmith> | Hixie: question about the outline algorithm behavior for the case of <hgroup>foo <h1>bar</h1> <h2>baz</h2></hgroup> |
| 07:14 | <MikeSmith> | it's not clear to me what the spec says should be the heading that goes into the outline for that |
| 07:15 | <MikeSmith> | all the implementations of the outline algorithm I've tried use just "bar" |
| 07:16 | <MikeSmith> | but I don't see where the spec requires that instead of using "foo" |
| 07:18 | <Hixie> | MikeSmith: the heading is "foo" with subheading "bar" |
| 07:19 | <MikeSmith> | aha |
| 07:19 | <Hixie> | oh wait |
| 07:19 | <Hixie> | i misread your example |
| 07:19 | <Hixie> | it's bar with subheading baz |
| 07:19 | <Hixie> | the foo is bogus |
| 07:19 | <MikeSmith> | ok |
| 07:19 | <Hixie> | hopefully the spec actually agrees |
| 07:19 | <MikeSmith> | I can't tell that it does actually |
| 07:20 | <MikeSmith> | will file a bug |
| 07:21 | <Hixie> | outline algorithm says the hgroup is the heading, right? |
| 07:21 | <MikeSmith> | yeah |
| 07:22 | <MikeSmith> | that part's clear |
| 07:23 | <JonathanNeal> | hgroup can't contain anything that isn't a <h1-6>. |
| 07:23 | <JonathanNeal> | That's why us <subline>rs were rabble rousing. |
| 07:24 | <MikeSmith> | JonathanNeal: it can contain them in the DOM |
| 07:24 | <MikeSmith> | I realize it's not valid |
| 07:24 | <Hixie> | MikeSmith: ok so if hgroup is the heading, then the question is just what does the hgroup represent |
| 07:25 | <Hixie> | ...which i guess is indeed rather vague |
| 07:25 | <MikeSmith> | ah damn |
| 07:25 | <MikeSmith> | I think it might be clear |
| 07:25 | <MikeSmith> | http://www.whatwg.org/specs/web-apps/current-work/multipage/sections.html#headings-and-sections |
| 07:26 | <MikeSmith> | "The first element of heading content in an element of sectioning content represents the heading for that section." |
| 07:26 | <MikeSmith> | oh no sorry |
| 07:26 | <MikeSmith> | not that |
| 07:26 | <Hixie> | right that's just saying the hgroup is the heading |
| 07:26 | <MikeSmith> | yup |
| 07:27 | <JonathanNeal> | rah rah <hgroup> poorly thought out rah rah <subline> solves all problems rah rah rahn paul rah |
| 07:31 | <JonathanNeal> | rah rah clear up :heading(n) confusion, reduce ambiguity of headings, spend more time solving issue of multiple subheadings of different weights. |
| 07:32 | <Hixie> | (in other news, i hate document.domain) |
| 07:36 | <MikeSmith> | Hixie: recent changes and subsequent bz bugs? |
| 07:36 | <Hixie> | everything around the effective script origin crap |
| 07:36 | <Hixie> | worst security design ever |
| 07:51 | <Hixie> | someone really should review the HTML spec's security stuff, I'm sure I have all kinds of bugs where I talk about the origin (should it be effective origin) of scripts (first script? calling script?) |
| 08:00 | <pottersson> | Does anyone know where I can download a copy of the WHATWG spec for HTML5? |
| 08:02 | <Hixie> | wow, he didn't stick around long |
| 08:06 | <zcorpan> | annevk: do you know what ie does with % in host? |
| 08:07 | <Hixie> | zcorpan: any idea where sigbjoern got the idea of commas in CORE-35269? |
| 08:08 | <Hixie> | zcorpan: it seems to contradict the data in https://www.w3.org/Bugs/Public/show_bug.cgi?id=18228 |
| 08:11 | <zcorpan> | Hixie: i guess he looked at the URL in the bug, saw that it worked in firefox and webkit but not in opera, didn't test IE and didn't research other content |
| 08:12 | <Hixie> | the url doesn't use a comma |
| 08:12 | <Hixie> | and never did, per the comments |
| 08:12 | <zcorpan> | CORE-40846 does though |
| 08:14 | <Hixie> | i can't get to that url |
| 08:14 | <Hixie> | so can't check |
| 08:15 | <zcorpan> | url for that is https://bespin.mozillalabs.com/ |
| 08:15 | <zcorpan> | http://code.google.com/codesearch#search/&q=%3Cmeta%5Cs%2Bhttp%5C-equiv%5C=%5C%22refresh%5C%22%5Cs%2Bcontent%5C=%5C%22%5Cd%2B%5C,&type=cs |
| 08:17 | <Hixie> | zcorpan: right, that url doesn't work for me |
| 08:17 | <zcorpan> | ah. |
| 08:18 | <zcorpan> | i don't have a login either |
| 08:19 | <Hixie> | that second link is good though, i've reopened the bug for it |
| 08:21 | <Hixie> | ok, bed time. nn. |
| 08:23 | <zcorpan> | nn |
| 08:43 | <zcorpan> | i guess the html spec is full of things that can be changed to webidl enums |
| 08:45 | <Ms2ger> | There are a number in canvas, at least |
| 09:08 | <MikeSmith> | so http://qa-dev.w3.org:8888/ no has an experimental "Show outline" feature |
| 09:09 | <MikeSmith> | implemented in JavaScript, using h5o |
| 09:10 | <MikeSmith> | so you can do, e.g., http://qa-dev.w3.org:8888/?doc=http://quirks.spec.whatwg.org/&showoutline=yes |
| 09:10 | <MikeSmith> | or http://qa-dev.w3.org:8888/?doc=http://quirks.spec.whatwg.org/&showoutline=yes#outline to go straight to the outline |
| 09:19 | <MikeSmith> | and a separate feature at http://qa-dev.w3.org:8888/outline/?doc=http://quirks.spec.whatwg.org/ that generates the outline on the server side |
| 09:21 | <zcorpan> | MikeSmith: why isn't the server-side thing used for the validator? |
| 09:22 | <zcorpan> | MikeSmith: can you flip the order of the outline and source checkboxes so they match the output? |
| 09:23 | <MikeSmith> | yeah can flip that |
| 09:24 | <MikeSmith> | and the I'm not using the server-side one directly yet is that it's harder to integrate on the server side |
| 09:24 | <zcorpan> | ok |
| 09:24 | <zcorpan> | looks nice! |
| 09:25 | <MikeSmith> | thanks |
| 09:25 | <MikeSmith> | also I'm more suspicious of the server-side code |
| 09:25 | <MikeSmith> | I didn't write it |
| 09:25 | <MikeSmith> | well, not most of it |
| 09:25 | <MikeSmith> | but I may have already spent more time rewriting it than it would have taken me to write it myself from scratch |
| 09:30 | <jgraham> | "01:36 < MikeSmith> Hixie: recent changes and subsequent bz bugs?" - I literally have no idea if that's bz as-in "Bugzilla" or "Boris" |
| 09:31 | <abarth> | Hixie: I think ascii, but you should check |
| 09:35 | <zcorpan> | jgraham: i haven't seen bugzilla abbreviated to bz |
| 09:46 | <jgraham> | zcorpan: They talk about the bzapi |
| 09:47 | <jgraham> | Which I don't think is for controlling Boris |
| 09:47 | <zcorpan> | oh |
| 09:47 | <zcorpan> | well you never know. I'm not convinced Boris is human. |
| 09:47 | <jgraham> | Indeed |
| 09:50 | Ms2ger | denies everything |
| 09:55 | <jgraham> | In unrelated news, tobie added something to testharness.js that required the use of ancestorOrigins in WebKit to avoid meaningless console spam. I don't think "happy with it" is quite accurate; it was horrible to have to make this workaround and the API wasn't a great fit. (See https://github.com/w3c/testharness.js/commit/e1918cae6e743247ad58842aae58196b229101c6#L0R2029 ) |
| 09:57 | <annevk> | zcorpan: no |
| 10:01 | <zcorpan> | if ancestorOrigins exists because webkit doesn't throw on cross-origin access, i see little reason for other browsers to implement it |
| 10:04 | <odinho> | jgraham: All that strange code was added to testharness, -- is there some timeout where we can remove it again if it's not being used for these hypotetichal web services that'll gather results? |
| 10:19 | <jgraham> | I thought there was some use case that wasn't tied to the deficiencies of WebKit |
| 10:21 | <zcorpan> | there are use cases for getting your origin, but that's addressed by http://url.spec.whatwg.org/#dom-url-origin AFAICT |
| 10:22 | <zcorpan> | if there's a use case for knowing that you're in a cross-origin sandbox, i've probably missed it |
| 10:39 | <MikeSmith> | me sees lots of heycam|away code arriving for style@scoped |
| 10:40 | <MikeSmith> | https://bugzilla.mozilla.org/show_bug.cgi?id=508725#attachment_table |
| 10:47 | <jgraham> | zcorpan: I think the use case was something like widgets wanting to behave differently in same-origin vs different origin cases? |
| 10:48 | <jgraham> | In unrelated news, it seems that the postMessage change to testharness.js broke stuff |
| 10:49 | <zcorpan> | heycam|away: ideas for tests: style scoped as root element, style scoped with svg parent, in head, as child of the root, in quirks mode, changing scopedness and textContent in a timeout (not sure if onload happens after layout) |
| 10:50 | <zcorpan> | heycam|away: check CSSOM stuff together with scoped |
| 10:51 | <zcorpan> | heycam|away: interaction with !important |
| 10:52 | <zcorpan> | oh there were a couple of tests with !important |
| 10:54 | <darobin> | jgraham: you mean tobie's patch? |
| 10:54 | <zcorpan> | heycam|away: moving a style scoped element to another document |
| 10:54 | <jgraham> | darobin: Yes |
| 10:55 | <jgraham> | In particular tests that postMessage results from one frame/window to another will now get confused by extra messages coming from the harness |
| 10:55 | <jgraham> | As I said on #testing, I think the right solution is probably just not to postMessage on the document with testharness included but only on ancestors |
| 10:58 | <annevk> | I guess today I should tackle percent-encoded bytes :/ |
| 11:00 | <darobin> | jgraham: mmmm, I reckon that actually makes sense yes |
| 11:07 | <zcorpan> | annevk: option 1 without the weirdness seems like the least insane option, though i'd be happy with 2 |
| 11:08 | <annevk> | zcorpan: I think we need 2, and that means either deferring to Unicode for valid utf-8 sequence or defining it in Encoding (either way is pretty simple though) |
| 11:08 | <zcorpan> | annevk: ok |
| 11:09 | <zcorpan> | annevk: why do we need 2? |
| 11:09 | <annevk> | zcorpan: in particular 2 is likely needed for UI anyway, and if we need it there we might as well use it for a slightly better result |
| 11:09 | <zcorpan> | ok |
| 11:11 | <zcorpan> | annevk: isn't valid utf-8 sequence implicitly defined in Encoding as part of the decode algorithm? |
| 11:11 | <annevk> | although I guess the result is not that much better... if you really tried to put bytes higher than 0x7F over DNS you could not use this system anyway |
| 11:11 | <annevk> | zcorpan: yeah it is |
| 11:11 | <annevk> | zcorpan: seems nicer to give something to pattern match against though, but maybe you'd use the decoder for that anyway |
| 11:12 | <annevk> | maybe I should go with what Chrome has first and see if people complain... |
| 12:01 | <SimonSapin> | was colspan="0" removed in html? |
| 12:03 | <annevk> | "For the purposes of the CSS table model, the colspan and rowspan attributes on td and th elements are expected to provide the special knowledge regarding cells spanning rows and columns." |
| 12:04 | <annevk> | SimonSapin: and within the HTML table model it seems 0 and 1 are equivalent |
| 12:05 | <zcorpan> | i thought 0 was supported in one of rowspan or colspan |
| 12:06 | <annevk> | SimonSapin: oh, and it must be non-zero as conformance requirement |
| 12:06 | <annevk> | zcorpan: prolly rowspan then |
| 12:09 | <odinho> | I remember hearing implementations differ on it :-) |
| 12:10 | <zcorpan> | yeah. and gecko does (or did?) different things in quirks and non-quirks, iirc |
| 12:21 | <SimonSapin> | MDN documents colspan=0 as "extends until the end of the <colgroup>" |
| 12:22 | <SimonSapin> | which is the definition in html4 |
| 12:22 | <SimonSapin> | (and is also what I want to do) |
| 12:22 | <annevk> | SimonSapin: what do implementations do? |
| 12:23 | <SimonSapin> | don’t know |
| 12:23 | <SimonSapin> | let me try |
| 12:24 | <SimonSapin> | 0 is the same as 1 in Opera and Chromium; as in HTML4 in Firefox |
| 12:25 | <annevk> | and IE? |
| 12:26 | <SimonSapin> | I don’t have it available right now |
| 12:26 | <annevk> | be sure to test soon, HTML WG might declare consensus based on Opera/Chromium |
| 12:26 | <annevk> | and then... well nothing really |
| 12:27 | <annevk> | it's not like previous W3C RECs withstood the test of time |
| 12:28 | <annevk> | I kinda wish IDL had a participate box |
| 12:30 | <SimonSapin> | Here is a test case, if someone wants to test in IE : http://dabblet.com/gist/4117646 |
| 12:50 | <annevk> | SimonSapin: IE9 I get the same as in Opera |
| 12:50 | <annevk> | SimonSapin: using http://netrenderer.com/ anyway |
| 12:54 | <SimonSapin> | annevk: treating colspan="0" like 1 is what is specified in HTML now, so no problem here |
| 12:55 | <SimonSapin> | But I’d still like to achieve the functionality |
| 12:55 | <SimonSapin> | huge values like colspan="999999" are to be ignored sometimes, colspan="99" seems to be ok |
| 12:56 | <Ms2ger> | I believe anything over a thousand or so is set to 1 |
| 12:56 | <SimonSapin> | apparently |
| 12:57 | <annevk> | SimonSapin: do you want this for hand authoring tables? |
| 12:57 | <SimonSapin> | yes |
| 12:57 | <annevk> | http://encoding.spec.whatwg.org/#concept-encoding-get uses rowspan, but it's just computed in Python |
| 12:57 | <annevk> | I guess it would be kinda nice if I didn't have to do that |
| 12:58 | <SimonSapin> | I know the table currently has 5 columns so I can use colspan="5", but it’s been 4 and then 6 in the past hour before I changed my mind |
| 12:59 | <SimonSapin> | rowspan="0" is still supported |
| 12:59 | <SimonSapin> | at least in the spec |
| 13:00 | <annevk> | does that do what I need? |
| 13:00 | <SimonSapin> | yes |
| 13:00 | <annevk> | oh |
| 13:00 | <annevk> | :) |
| 13:00 | <annevk> | well maybe I'll replace that len(...) with 0 at some point |
| 13:01 | <SimonSapin> | and add some <tbody>s |
| 13:01 | <annevk> | oh you need <tbody>? |
| 13:01 | <annevk> | it uses <tbody> already for something |
| 13:02 | <SimonSapin> | "if its value is set to 0, it extends until the end of the table section (<thead>, <tbody>, <tfoot>, even if implicitly defined), that the cell belongs to." |
| 13:02 | <annevk> | ah so it wouldn't work |
| 13:09 | <marcosc_> | does anyone know if there is any practical difference between HTML's algorithm for parsing floats and ECMAScript's parseFloat()? |
| 13:13 | <jgraham> | Don't they disagree about scientific notation? |
| 13:13 | <jgraham> | 1.2e7 and so on? |
| 13:13 | jgraham | didn't check |
| 13:18 | <marcosc_> | jgraham: yeah, that's what I'm wondering... but for most common cases I guess they are functionally the same (i.e., for parsing the value some attribute) |
| 13:26 | <Ms2ger> | marcosc_, yep, agree in the common case and disagree if you try |
| 13:29 | marcosc_ | lazy, just wants to use parseFloat() for prototyping in a small parser |
| 13:30 | <darobin> | marcosc_: don't be a baby and go write a JS version of the HTML5 algo |
| 13:31 | <darobin> | then we can run coverage tools on that to see if we're testing it properly ;) |
| 13:31 | <marcosc_> | tools. will. save. |
| 13:31 | <annevk> | wasn't it the case that implementations used the JS algo? |
| 13:31 | annevk | forgot the details |
| 13:32 | <Ms2ger> | No |
| 13:38 | <annevk> | Ms2ger: what was the thing AryehGregor pointed out then? |
| 13:38 | <annevk> | Ms2ger: had something to do with number parsing too I believe |
| 13:39 | <Ms2ger> | Hmm? I think Aryeh suggested changing spec and implementations to use the JS algorithm instead |
| 13:39 | <annevk> | there were at least some implementations on his side |
| 13:39 | <annevk> | iirc |
| 13:39 | <Ms2ger> | Hmm |
| 13:40 | <marcosc_> | would be nice to have a single algo instead of two |
| 13:40 | <Ms2ger> | Was that about negative zero, maybe? |
| 13:40 | <annevk> | maybe it was floating point precision? |
| 13:40 | <annevk> | ooh, and maybe JS whitespace versus HTML whitespace |
| 13:40 | <annevk> | stuff like that |
| 13:41 | marcosc_ | checks if he implemented the JS one |
| 13:42 | <marcosc_> | hmm... nope |
| 13:54 | <annevk> | heh |
| 13:55 | <annevk> | open DragonFly, > document.location.host="x" |
| 13:55 | <annevk> | hit cmd+back |
| 13:55 | <annevk> | crash |
| 13:56 | <annevk> | in Chrome if you do that a few times you end up at data:text/html,chromewebdata |
| 13:56 | <annevk> | Gecko just throws |
| 13:56 | <annevk> | Film at 11? |
| 13:56 | <annevk> | maybe there's something to be said for making Location members both Unforgeable and readonly |
| 14:11 | <marcosc_> | Hixie: I think there might be a mistake in the img@srcset parsing algorithm. In the "Descriptor parser:", where it says "If width is *not* absent, then let error be yes". But if a value is not absent and is valid, then it's not in error. When I follow the algorithm, I can't build the entry list with valid candidate entries. Note this also applies to height, and density. Or am I missing something? |
| 14:14 | <annevk> | width can only be not absent at that point if it's double |
| 14:15 | <annevk> | duplicated* |
| 14:15 | <annevk> | marcosc_: so you're missing something I think |
| 14:21 | <marcosc_> | annevk: to me that reads "if the width exists, then error = yes " |
| 14:22 | <annevk> | marcosc_: you're not reading it correctly then |
| 14:22 | <annevk> | marcosc_: it explicitly says to split the list on spaces, then to set various variables, including width to absent |
| 14:22 | <annevk> | marcosc_: then it says to process what you just split |
| 14:23 | <annevk> | marcosc_: if you start with the first token in what you just split, width is obviously still absent |
| 14:23 | <annevk> | marcosc_: simple English... |
| 14:24 | <marcosc_> | if it was simple English, we would not be having this discussion (i.e., please refrain from passably insulting me :) ) |
| 14:24 | <annevk> | I'm saying it simple English because people have a tendency to read stuff into specifications that are not there |
| 14:24 | <marcosc_> | annevk: let me show you a 1 to 1 with code |
| 14:26 | <marcosc_> | annevk: https://github.com/ResponsiveImagesCG/picture-refimp/blob/master/srcsetfill.js#L369 |
| 14:27 | <annevk> | yeah you're code is not doing what the English says |
| 14:27 | <annevk> | you need if(!width) error = 'yes'; |
| 14:28 | <annevk> | or if (width == undefined) if you want to be more careful |
| 14:28 | <annevk> | your code, oh god |
| 14:28 | <marcosc_> | what's wrong with it? |
| 14:28 | <annevk> | I just told you? |
| 14:28 | <marcosc_> | right, but then you said "oh god" |
| 14:29 | <annevk> | because I made a typo |
| 14:29 | <annevk> | in spelling "your" |
| 14:32 | <annevk> | marcosc_: code looks well documented :) |
| 14:33 | <marcosc_> | annevk: I tried to keep the code 1.1 with the spec |
| 14:35 | <marcosc_> | annevk: but yes, I f'ed up it seems by using undefined instead of "absent" |
| 14:36 | <annevk> | euh no |
| 14:36 | <annevk> | using undefined is fine, you just didn't check for it |
| 14:50 | <zcorpan> | marcosc_: typo? ulr: rawCandidates[i].url, |
| 14:50 | <marcosc_> | zcorpan: thanks |
| 14:50 | marcosc_ | debugging at the moment, first time I've managed to run the code |
| 14:52 | <zcorpan> | marcosc_: i hear from people who know js better than me that delete should be avoided |
| 14:53 | <marcosc_> | zcorpan: what's a good alternative? |
| 14:53 | <marcosc_> | just copy into new array? |
| 14:53 | <zewt> | github's javascript syntax coloring is the worst |
| 14:54 | <zewt> | grey on white is not okay |
| 14:54 | <marcosc_> | zewt: yeah, agree |
| 14:54 | <zewt> | zcorpan: well, presumably only if it's a hot code path; if not, that seems like premature optimization |
| 14:56 | <jgraham> | Right, I think I might make candidates a map and have a bool indicating whether it should be considered or not, but that might not be better |
| 14:56 | <zewt> | (not necessarily in the "making code more complicated by optimizing it" sense, but in the "thinking about performance in places where it doesn't matter" sense) |
| 14:57 | <jgraham> | Oh, so it is sort of a map already |
| 14:57 | <jgraham> | But yeah, this seems like it could be premature optimisation |
| 14:59 | <annevk> | marcosc_: so I would make error a boolean and use undefined or some such for the others... |
| 14:59 | <marcosc_> | annevk: ok, I'll add that |
| 14:59 | <annevk> | marcosc_: though I suppose that only matters if you want to make this a little faster |
| 15:00 | <zewt> | later, work |
| 15:00 | <marcosc_> | annevk: right now, I just want to make sure it's running as spec'ed... then I can optimize it. There is _a lot_ of redundancy in the spec (which is good for legibility and all that) |
| 15:00 | <annevk> | marcosc_: also, you check height after you set it... |
| 15:01 | <annevk> | marcosc_: which doesn't match the spec |
| 15:01 | <marcosc_> | annevk: you are looking at the latest version? |
| 15:02 | marcosc_ | just checked in an update |
| 15:02 | <zcorpan> | this isn't about optimization. this is about not having to worry that farre will find your code, haunt you down and kill you slower than delete is slow. |
| 15:03 | <marcosc_> | :) |
| 15:03 | <Ms2ger> | Pseudonymity is useful in such cases :) |
| 15:03 | <annevk> | marcosc_: yes? |
| 15:03 | <jgraham> | Ms2ger: He might just kill everyone that can't prove they aren't you |
| 15:04 | <marcosc_> | that makes sense, I don't want to die. |
| 15:04 | <marcosc_> | annevk, ok, fixing |
| 15:04 | <annevk> | marcosc_: unless your latest commit is not "fixed typo and bug" and you fixed that |
| 15:04 | <Ms2ger> | jgraham, I'd just ask him if he can prove he isn't me :) |
| 15:06 | <marcosc_> | annevk: thanks, well spotted. |
| 15:07 | <jgraham> | Ms2ger: Well now he has a good way to tell if it is you or not |
| 15:07 | <Ms2ger> | So if he wouldn't ask himself... I give up |
| 15:19 | <Ms2ger> | Is http://dev.w3.org/html5/webvtt/ the canonical? |
| 15:21 | <annevk> | think so |
| 15:58 | <annevk> | http://annevankesteren.nl/2012/11/github-whatwg |
| 15:58 | <annevk> | wasn't quite sure what else to write down |
| 15:59 | <annevk> | in the process of writing that I found http://annevankesteren.nl/2004/11/whatwg-is-alive again which is kinda amusing |
| 15:59 | <annevk> | "There could be a place for what you call HTML 5.0 but it needs to be submitted to a standards body and approved by them once it is finished before we can start using it." |
| 15:59 | <marcosc_> | annevk: are you allowing people to file bugs on GH? |
| 15:59 | <marcosc_> | (i see that is currently enabled) |
| 16:00 | <annevk> | marcosc_: it's only enabled for whatwg/javascript I think |
| 16:00 | <marcosc_> | annevk: https://github.com/whatwg/xref/issues |
| 16:01 | <darobin> | "you can't rely on HTML alone these days" |
| 16:01 | <darobin> | shorter annevk: "in the Netherlands, we don't need your stinky document conformance criteria" |
| 16:02 | <marcosc_> | "Pardon me for not seeing the point of HTML5.0. I see XHTML as the rightful heir of HTML." |
| 16:02 | <marcosc_> | heh |
| 16:03 | <annevk> | marcosc_: I guess I haven't disabled it for that one, but that should prolly stay there as there's no alternative |
| 16:03 | <annevk> | marcosc_: though I'll close the wiki, thanks |
| 16:03 | <jgraham> | In related news, I see myself as the rightful heir of Arthur, legendary king of England |
| 16:05 | <darobin> | annevk: reckon HTML might move to GH at some point? |
| 16:05 | <annevk> | darobin: my magic eight ball says no |
| 16:05 | <jgraham> | darobin: Good luck convincing Hixie of that |
| 16:05 | <darobin> | oh, I wasn't planning on trying to convince anyone — just curious |
| 16:06 | <annevk> | darobin: only thing I considered was setting up a mirror, but wasn't sure what the use would be |
| 16:06 | <darobin> | the W3C repo has a git-happy copy anyway |
| 16:06 | <darobin> | annevk: a copy would only make sense if you plan on taking pull requests there |
| 16:06 | <darobin> | and I'm guessing that's not really the spirit |
| 16:07 | <annevk> | darobin: it is the spirit for many of the stuff WHATWG writes these days, but Hixie has a different preference |
| 16:08 | <darobin> | I know, I know, as I said I was just curious |
| 16:09 | <annevk> | fair enough, I thought of actually mentioning it in the blog post, but explaining that took too much text, but now I wrote all these lines maybe I should have done it :) |
| 16:11 | <darobin> | hahaha |
| 16:16 | <annevk> | darobin: so https://github.com/w3c/html/commits/feature/whatwg is your WHATWG copy right? |
| 16:17 | <annevk> | seems to match the commits in any case |
| 16:21 | <darobin> | annevk: yeah, that should be it |
| 16:21 | <darobin> | there are further branches that are subsets of that |
| 16:22 | <darobin> | I don't know how often it's updated, I think hourly or so |
| 16:22 | <darobin> | at any rate it's never far behind |
| 17:10 | <TabAtkins> | heycam|away: Re <style scoped>, the effects on the cascade are now well-defined in the Cascade spec. Please either match it or file bugs on us with arguments to change. ^_^ |
| 17:23 | <GPHemsley> | Argh, Apache. Why you gotta be so frustrating? |
| 17:29 | <GPHemsley> | There is apparently no way to turn off Apache's type-detection-by-file-extension behavior. |
| 17:30 | <GPHemsley> | oh! maybe I spoke right too soon! |
| 17:32 | <GPHemsley> | ahh... :) |
| 17:32 | <GPHemsley> | so simple: RemoveType |
| 17:32 | <GPHemsley> | (but off in another part of the docs) |
| 17:34 | <annevk> | don't use a file extension? |
| 17:35 | <GPHemsley> | annevk: I'm testing whether browsers assume anything from it. |
| 17:38 | <matjas> | annevk: can haz link to the javascript spec on http://www.whatwg.org/specs/? |
| 17:39 | <annevk> | matjas: Hixie can arrange that |
| 17:42 | <GPHemsley> | oh, hmm... |
| 17:42 | <GPHemsley> | I'm realizing now that I'm not testing any mismatches |
| 17:43 | <_divya> | matjas: are you an editor these days?! |
| 17:44 | <annevk> | _divya: uhuh, when do you start? |
| 17:44 | <matjas> | _divya: i’ve been working on that spec occasionally since August or so |
| 17:45 | <matjas> | _divya: anyone can become an editor! http://annevankesteren.nl/2012/11/github-whatwg :) |
| 17:45 | <_divya> | annevk: i only gossip about specs. |
| 17:45 | <_divya> | who wants to do the hard work |
| 17:45 | <matjas> | haha |
| 17:45 | <_divya> | matjas: very nice!! I want to say congratulations but unsure of what the official stance of whatwg/w3c is |
| 17:45 | <annevk> | heh, I wish we had weekly gossip magazines about specs, that'd be great |
| 17:46 | <_divya> | isnt that w3cmemes annevk? |
| 17:46 | <annevk> | only @mattur is left I think |
| 17:46 | <_divya> | ahahaha |
| 17:46 | <_divya> | yes |
| 17:46 | <annevk> | ah yeah @w3cmemes is there |
| 17:46 | <_divya> | and steve faulkner |
| 17:46 | <annevk> | no more mr last week |
| 17:46 | <_divya> | that was a bit weird |
| 17:47 | <annevk> | and there's the occasional culture clash |
| 18:42 | <annevk> | TabAtkins: I think TC39 reads public-script-coord, but we'll see |
| 18:42 | <[tm]> | speaking of culture clashes, i normally disagree with pretty much everything Great Leader Roy says, but in pretty much every discussion about DNT, dude is speaking the gospel truth |
| 18:42 | <annevk> | TabAtkins: that was sort of my idea behind posting it there |
| 18:43 | <[tm]> | e.g., http://lists.w3.org/Archives/Public/public-tracking/2012Nov/0290.html |
| 18:44 | <Ms2ger> | [tm], damn gray, black&white is much easier |
| 18:45 | <[tm]> | Heh |
| 18:45 | <[tm]> | yeah |
| 18:47 | <TabAtkins> | annevk: kk |
| 18:48 | <TabAtkins> | annevk: Alex Russell makes the point that we might not need to really do the stringifying eagerly in this case - we can wait and do it upon requesting things instead. Thoughts? |
| 18:48 | <TabAtkins> | (This isn't a general argument - there are definitely cases where you need the values to be eagerly checked/converted.) |
| 18:48 | <annevk> | that doesn't work, you directly manipulate the URL |
| 18:48 | <TabAtkins> | Ah, yeah. |
| 18:49 | <annevk> | and even if you did set a string, it needs to be further converted |
| 18:49 | <TabAtkins> | Okay, that's what I thought. |
| 18:49 | <annevk> | e.g. lone surrogates need to be replaced with U+FFFD, "&" needs to be escaped, etc. |
| 18:49 | <TabAtkins> | Cool, thanks, I'll argue appropriately. |
| 19:00 | <annevk> | TabAtkins: seems you might also have to argue with TC39 |
| 19:00 | <annevk> | TabAtkins: also, I'm not sure I want to wait for ES7 before getting this in browsers; I was actually hoping to have this specced within one or two months |
| 19:04 | <TabAtkins> | annevk: Yes, I'm willing to do that arguing. And also to get this resolved in a short period of time. ^_^ |
| 19:04 | <annevk> | cool |
| 19:04 | <annevk> | I guess my writing won't be impacted by much of it, I'll just keep in mind it needs to tie into something at some point |
| 19:04 | <Velmont> | Upgrading my Piwik. Find it funny that a new fix disables DNT header and still tracks, if the user agent shows that it's IE10. Because they set DNT on to standard, Piwik won't respect it :-) |
| 19:05 | <annevk> | Apache did that too, I think that's one of the bad things Roy did <- [tm] |
| 19:06 | <Velmont> | Nah, I think Microsoft was totally in the wrong. |
| 19:06 | <Velmont> | How would it even be useful when it's on by default? |
| 19:06 | <Velmont> | Then you don't actually get only the crazies that care about it, -- but even your father and everyone else who doesn't really care. |
| 19:22 | <annevk> | huh? nobody wants to be tracked |
| 19:28 | <Velmont> | annevk: you crazy :D |
| 19:34 | <annevk> | I think I might be |
| 19:34 | <annevk> | I somehow ended up on http://www.koalastothemax.com/ and moved the mouse around until I saw a dog, then reloaded and did it again, only to see an owl |
| 19:35 | <TabAtkins> | That's the best site. |
| 19:51 | <annevk> | Hixie: btw, these longer diff lines suck on the iPhone |
| 19:51 | <annevk> | Hixie: but maybe not on the iPhone 5... hmm |
| 19:57 | <Hixie> | annevk: dude, who reads diffs on a phone -_- |
| 19:58 | <annevk> | Hixie: who edits the HTML spec with the ultimate goal of controlling his model train set? |
| 20:04 | <Hixie> | annevk: details |
| 20:10 | <annevk> | btw, I finally found where that " " makes the host invalid comes from: http://tools.ietf.org/html/rfc3454#appendix-C |
| 20:10 | <annevk> | of course, per IDNA2008 that's no longer used |
| 20:20 | <gsnedders> | Hixie: Has your train set got out of a box yet? |
| 20:21 | <gsnedders> | Or are you still working on the HTML side of things in the first instance? :) |
| 20:32 | <Hixie> | "train set" |
| 22:10 | <heycam> | TabAtkins, thanks for the pointer to css3-cascade. I think my patches do match what you've got in there. |
| 22:11 | <TabAtkins> | heycam: Cool. |