| 01:10 | <TabAtkins> | Hixie: Typo in the parser section "Preprocessing the input stream" -> "explicit ignored" should be "explicitly ignored". |
| 03:30 | <Hixie> | TabAtkins: unless it's an area i'm in the middle of editing, it's best to send that kind of feedback to the bug system by just pasting the offending text and saying "typo" in the feedback box |
| 03:30 | <Hixie> | TabAtkins: just as quick as IRC, but I won't lose track of it :-) |
| 05:36 | <Von_Davidicus> | Hello, all. |
| 05:47 | <annevk> | oh hey, it's Friday the 13th |
| 05:48 | <Von_Davidicus> | Hello, annevk. May I get your opinion as to the accuracy of a statement I wanted to put into an essay? |
| 05:49 | <jwalden> | friday, friday, gotta get down on friday |
| 05:49 | <annevk> | where is Neil deGrasse Tyson to point out Thursday the 12th is just as rare |
| 05:49 | <annevk> | jwalden: heh |
| 05:49 | <Von_Davidicus> | The statement is: "XHTML 2.0 was virtually booed off the web." |
| 05:50 | <annevk> | Von_Davidicus: XHTML 2.0 got some attention, but was mostly not adopted because it was not backwards compatible, and the breaking changes didn't bring enough benefit to be worth the enormous transition cost |
| 05:50 | <Von_Davidicus> | Oh, okay. From what I read, it was virtually rejected en masse. |
| 05:50 | <Von_Davidicus> | So basically, my statement is way off base? |
| 05:51 | <annevk> | in the end people were pretty opposed, in the beginning even Hixie cheered that effort on |
| 05:52 | <annevk> | see e.g. http://ln.hixie.ch/?start=1042630901&count=1 |
| 05:53 | <Hixie> | Von_Davidicus: what do you mean by "virtually" in that sentence? |
| 05:54 | <Hixie> | Von_Davidicus: really it wasn't that xhtml2 was rejected, so much as it had no adoption |
| 05:54 | <Hixie> | annevk: i was supportive of the xhtml2 wg even as it was being shut down |
| 05:54 | <annevk> | :) |
| 05:54 | <Hixie> | annevk: i just didn't think what they were doing was being adopted (or had much chance of being adopted) :-) |
| 05:55 | <Von_Davidicus> | I know there was a lot of what XHTML2 that I was looking forward to. <li><a...><img...> condensed into <li>? Yes please. |
| 05:57 | <Von_Davidicus> | And what I meant by "virtually" was that there were some dissenting voices, the concept remained, but the majority of the developers gave xHTML 2.0 the thumbs-down. |
| 05:57 | <annevk> | oh, Chrome will be the second browser with <input type=date> |
| 05:57 | <jwalden> | yeah, see, that's not *really* a tangible benefit :-) a little cleaner looking, but it doesn't give any new functionality |
| 05:57 | <annevk> | only six years after Opera |
| 05:57 | <Hixie> | screw type=date |
| 05:57 | <Hixie> | where's type=color! |
| 05:57 | <Hixie> | and when will firefox do type=range! |
| 05:58 | <Hixie> | also i really want more browsers to do <meter> |
| 05:58 | <jwalden> | people adopt stuff that gives them functionality, but a bit of change half for change's sake just doesn't do much for people when it comes to brass tacks |
| 05:58 | <annevk> | Gecko is doing B2G instead of WF2 |
| 05:58 | <Hixie> | also, why are browsers being so slow to implement <dialog> |
| 05:58 | <Hixie> | it's been in the spec for DAYS |
| 05:58 | <annevk> | ok now you're getting silly :p |
| 05:58 | <Hixie> | B2G? |
| 05:58 | <jwalden> | clearly dialog needs to be talked up more |
| 05:58 | <annevk> | https://wiki.mozilla.org/B2G |
| 05:59 | <Hixie> | <dialog> is gonna be awesome |
| 05:59 | jwalden | isn't sure what the current state for web forms is these days |
| 05:59 | <jwalden> | (in Gecko) |
| 05:59 | <annevk> | https://wiki.mozilla.org/B2G/FAQ is better actually |
| 05:59 | <Hixie> | annevk: ah, right |
| 06:00 | <annevk> | searching for your old XHTML 2.0 post I also found this one: http://ln.hixie.ch/?start=1180683614&order=1 |
| 06:00 | <annevk> | a nice reminder of the old days |
| 06:00 | <annevk> | report a serious implementation issue, no progress for years |
| 06:04 | <annevk> | "DOMException.code is deprecated per DOM4 spec." it isn't really... |
| 06:04 | <annevk> | I guess a little bit |
| 06:04 | <annevk> | oh well |
| 06:10 | <Von_Davidicus> | Question: Do you think XHTML 2.0 would have gained more traction if IE hadn't been so slow in adopting XHTML in general? |
| 06:13 | <jwalden> | more, I think yes |
| 06:13 | <jwalden> | much more? less likely |
| 06:15 | <Von_Davidicus> | I've read that it would have helped if XHTML2.0 had a different name. I mean, WML got somewhere. |
| 06:17 | <Hixie> | Von_Davidicus: WML didn't get much of anywhere except on devices that didn't already support HTML (so where there was no backcompat issue) |
| 06:17 | <Hixie> | Von_Davidicus: and when compat with HTML became an issue, it died |
| 06:18 | <Hixie> | XHTML2's main problem was that it was trying to replace something with a massive installed base, without offering a commensurate advantage |
| 06:18 | <Hixie> | if XHTML2 had been amazing, it could have overcome its backcompat issue |
| 06:19 | <Hixie> | but it barely did anything more than HTML4 |
| 06:19 | <Hixie> | so there really was no incentive to use it at all |
| 06:19 | <Von_Davidicus> | Oh, and Hixie... you mentioned in your blog you'd never found a use for the style="" attribute. I DID have one, though circumstances changed and it was removed. |
| 06:20 | <Hixie> | what was it? |
| 06:21 | <Von_Davidicus> | I had a page demonstrating all named colours--specifically, a table with the name, the hex, rgb, and hsl codes, AND a cell at the end of the row with the colour in question as a background. Since those were the only elements in the entire WEBSITE that used those styles, I decided it was better to simply use the style="" attribute than to go through creating 100+ rules, one for each of those table cells. |
| 06:23 | <Von_Davidicus> | Later on, though, I did a webpage on that same website describing CSS values--along with a list of CSS colour names. Now that I had two elements on the website using each styling, I decided to move the contents of the style="" attributes over to a stylesheet. |
| 06:26 | <MikeSmith> | the many problems with XHTML2 start with the fact that from its TOC you can't even tell what new features it might be adding |
| 06:26 | <MikeSmith> | instead you see the word "Module" about 75 times |
| 06:26 | <Von_Davidicus> | *Nods.* |
| 06:27 | <annevk> | MikeSmith: you should tell Jeff about that |
| 06:27 | <annevk> | MikeSmith: he seems to think we should modularize again |
| 06:27 | <MikeSmith> | then when you explore those modules you start to see that it's all the same stuff that was in HTML4, except now it has Modules |
| 06:28 | <Von_Davidicus> | What was the whole idea behind Modularizing? |
| 06:29 | <MikeSmith> | then once you've wasted a bunch of time doing that exploration you get to realizing what Hixie said about "no commensurate advantage" |
| 06:29 | <MikeSmith> | Von_Davidicus: ask the people who were in the XHTML WG |
| 06:29 | <Hixie> | Von_Davidicus: sounds like what you had was a graphic, not something that HTML would be really best used for |
| 06:30 | <Hixie> | Von_Davidicus: consider, how would you read it to someone on the phone? |
| 06:31 | <MikeSmith> | annevk: I've tried at least to make it clear to the team that modularization is not a priority to the actual implementors the spec is intended for |
| 06:32 | <MikeSmith> | and that it even causes problems for end users |
| 06:33 | <Von_Davidicus> | Good question, Hixie, but it was more a spreadsheet than anything, and the last cell in that table was simply a sample of the colour. May I send you a link to the page in question? |
| 06:34 | <MikeSmith> | witness the case of Microdata, where because of the way it's split out, you have the HTML5 spec specifying conformance restraints but needing to do it while pretending that Microdata doesn't exist, and the separate W3C Microdata spec needing to override that with the actual constraints |
| 06:34 | <MikeSmith> | but many people read just the constraints in the HTML5 spec but don't see the ones in the Microdata spec |
| 06:35 | <MikeSmith> | there have been several validator bugs filed about this already |
| 06:35 | <Hixie> | Von_Davidicus: i'm familiar with the kind of page you mean, it's ok |
| 06:35 | <Hixie> | Von_Davidicus: i think something like svg might be more appropriate for those though |
| 06:38 | <Von_Davidicus> | Not really. I wanted as many browsers as possible to be able to read the website, and older versions of Internet Exploder tend to get their knickers in a knot when SVG is involved. |
| 06:40 | <Hixie> | Von_Davidicus: ah, well, if the use case is "support IE", then I agree there's plenty of reasons to use style="" |
| 06:40 | <Hixie> | and <table> for layout, for that matter |
| 06:40 | <Hixie> | especially older IEs |
| 06:40 | <Hixie> | :-) |
| 06:40 | <Hixie> | i was asking more on a theoretical level |
| 06:41 | <Von_Davidicus> | Well, in this case, I use <table> for a spreadsheet-style webpage. |
| 06:51 | <zcorpan> | argh... why do spec editors feel the urge to "deprecate" stuff instead of jumping directly to the desired outcome and remove altogether directly |
| 06:55 | <smaug____> | what is wrong with deprecation? |
| 06:55 | <smaug____> | it is a strong hint that the stuff will go away at point |
| 06:55 | <zcorpan> | it doesn't work on the web |
| 06:55 | <zcorpan> | authors don't care if things are deprecated |
| 06:56 | <zcorpan> | they use what works |
| 06:57 | <Hixie> | yeah if you can't remove it straight away, you'll probably never remove it |
| 06:59 | <zcorpan> | right |
| 06:59 | <Von_Davidicus> | Who uses <font> these days? |
| 06:59 | <nesta_> | good morning! :) |
| 07:00 | <Ms2ger> | jwalden, the status of WF2 is that volkmar works on B2G instead |
| 07:00 | <jwalden> | plausible |
| 07:00 | <jwalden> | I had thought there was one or so other person looking at it, tho |
| 07:01 | <Von_Davidicus> | Question: Should one still test websites in the Lynx browser? |
| 07:01 | <Ms2ger> | And K9O isn't going to help |
| 07:02 | <jwalden> | Von_Davidicus: depends on your target audience |
| 07:02 | jwalden | has occasionally found it useful to browse sites with lynx |
| 07:03 | <jwalden> | but anyone who uses lynx to browse is a bit of an edge case, for most sites |
| 07:04 | <zcorpan> | Von_Davidicus: of 8915 front pages from the top 10,000 sites, i see 1850 pages using a <font> tag |
| 07:05 | <zcorpan> | data set http://www.paciellogroup.com/blog/2012/04/html5-accessibility-chops-data-for-the-masses/ $ grep -aPihrc "<font\s" . |
| 07:05 | <zcorpan> | so more than 20% |
| 07:05 | <Von_Davidicus> | Target: Mostly farmers, some of whom might be using browsers old enough that CSS support is dodgy. |
| 07:06 | <zcorpan> | that's more than pages using the html5 doctype in this data set |
| 07:07 | <Von_Davidicus> | I created a webpage for my dad's business, and I tested it in Lynx. |
| 07:20 | <annevk> | zcorpan: you mean the BlobBuilder stuff? |
| 07:20 | <annevk> | zcorpan: I don't get that either, just nuke it from the draft alright |
| 07:20 | <annevk> | if it turns out we absolutely must support BlobBuilder we can always add it back in |
| 07:20 | <Von_Davidicus> | Sorry, what is BlobBuilder? |
| 07:21 | <Ms2ger> | Something that will die a painful death soon enough |
| 07:21 | <smaug____> | hopefully not painful |
| 07:22 | <zcorpan> | annevk: yeah, but i recall seeing the same behavior before (dunno what it was about then) |
| 07:25 | <annevk> | zcorpan: init*Event() perhaps |
| 07:26 | <annevk> | zcorpan: Travis thought we should deprecate it first... |
| 07:26 | <annevk> | so much nonsense |
| 07:36 | <volkmar> | Ms2ger: i might have some minions this summer to do forms stuff |
| 07:45 | <Ms2ger> | http://24.media.tumblr.com/tumblr_m2efjw7RN61rrf1eeo1_1280.jpg |
| 07:47 | Von_Davidicus | chuckles |
| 07:49 | <Von_Davidicus> | Oh, question: Will HTML5 bring back <marquee> and <blink>? * :D Ducks bricks. :D * |
| 07:50 | <Ms2ger> | Both are specified in the HTML spec |
| 07:50 | <Von_Davidicus> | seriously? |
| 07:50 | <Ms2ger> | Yes |
| 07:51 | <Von_Davidicus> | Oh, my goodness, there it is. TLTAAAW... |
| 07:53 | <Von_Davidicus> | Oh, sorry, I should explain that abbreviation. It stands for "The Looney Tunes Are Alive And Well"; it's a phrase some of my friends and I use when we try to make what we think is an utterly ridiculous statement--and discover it's perfectly true. |
| 07:58 | <zcorpan> | https://twitter.com/AskDotCom/status/190521773945667584 LOL |
| 09:13 | <annevk> | zcorpan: seems kind of a weird reply |
| 09:24 | <zcorpan> | "If a type that is not one of DOMString [WebIDL], ArrayBuffer [TypedArrays], or Blob is used as a blobParts member, the user agent MUST throw an InvalidStateError." http://dev.w3.org/2006/webapi/FileAPI/#dfn-Blob |
| 09:24 | <zcorpan> | doesn't webidl throw TypeError already in that case? |
| 09:24 | <zcorpan> | or convert? |
| 09:28 | Ms2ger | sighs |
| 09:36 | <Ms2ger> | zcorpan, converts to DOMString, yes |
| 09:37 | Ms2ger | sends email |
| 09:38 | <zcorpan> | i filed a bug |
| 09:38 | <Ms2ger> | wfm |
| 09:38 | <zcorpan> | https://www.w3.org/Bugs/Public/show_bug.cgi?id=16721 |
| 09:42 | <annevk> | also, should we reconsider that constructor in light of ES6 changes? |
| 09:42 | <annevk> | a signature of (options, data...) prolly makes more sense |
| 09:43 | <zcorpan> | maybe, but not right now :-) |
| 09:45 | <hsivonen> | sigh. I'm looking at script data escaped less than sign state :-( |
| 09:45 | <hsivonen> | I didn't want to look here again |
| 09:46 | <zcorpan> | why are you looking at it? |
| 09:47 | <hsivonen> | zcorpan: crash in View Source |
| 09:47 | <zcorpan> | ah |
| 09:47 | <hsivonen> | you are in a maze of state transitions all alike |
| 09:49 | <zcorpan> | you are likely to be eaten by the W3C grue |
| 09:50 | <hsivonen> | smaug____ is going to have fun reviewing this |
| 09:55 | <annevk> | hsivonen: pointer to patch? |
| 09:55 | annevk | is curious |
| 09:56 | <zcorpan> | so do i understand correctly that new Blob("foo") is equivalent to new Blob(["f", "o", "o"]) per webidl? |
| 09:57 | <Ms2ger> | What makes you think that? |
| 09:57 | <zcorpan> | or is it equivalent to new Blob(["foo"])? |
| 09:57 | <Ms2ger> | It throws a TypeError |
| 09:58 | <zcorpan> | ah |
| 09:58 | <Ms2ger> | http://dev.w3.org/2006/webapi/WebIDL/#es-sequence |
| 10:00 | <zcorpan> | ok |
| 10:00 | <zcorpan> | cheers |
| 10:01 | <zcorpan> | new Blob(window) doesn't throw though, since it supports indexed properties :-) |
| 10:01 | <hsivonen> | annevk: patch not ready yet |
| 10:02 | <Ms2ger> | zcorpan, and it will append "[object Window]", "[object Window]", "[object Window]", ... ? |
| 10:02 | <zcorpan> | yeah |
| 10:02 | <zcorpan> | i'm totally gonna test that case |
| 10:02 | <Ms2ger> | Do it! :) |
| 10:07 | <zcorpan> | seems you can also overwrite the toStrings on the windows |
| 10:09 | <annevk> | zcorpan: if we don't change it now we can't change it in the future |
| 10:10 | <annevk> | and we'll make arv_afk sad |
| 10:10 | <zcorpan> | well, we don't support es6 today and i'm supposed to write tests for Blob today |
| 10:10 | <Ms2ger> | No more innerHTML? :( |
| 10:10 | <zcorpan> | not today |
| 10:11 | <zcorpan> | the impl for that is finished (should match the spec as it is, roughly) |
| 10:12 | <zcorpan> | hint: gecko and webkit don't match the spec |
| 10:12 | <zcorpan> | for innerHTML in xml that is |
| 10:13 | <Ms2ger> | Unsurprising |
| 10:13 | <Ms2ger> | We've got someone rewriting our innerHTML implementation soon, though |
| 10:14 | <hsivonen> | Ms2ger: for XML? |
| 10:15 | <hsivonen> | Ms2ger: I thought what was about to get rewritten is the innerHTML getter in the HTML mode |
| 10:15 | <Ms2ger> | Not clear from the bug |
| 10:15 | zcorpan | was talking about the setter |
| 10:15 | <hsivonen> | Ms2ger: it's a perf bug basically and XML perf doesn't matter |
| 10:16 | <Ms2ger> | :) |
| 10:16 | <Ms2ger> | zcorpan, including the silly thing with the foo-- element you showed earlier? |
| 10:17 | <hsivonen> | Ms2ger: the context is doing badly on some benchmark |
| 10:17 | <Ms2ger> | Yeah |
| 10:17 | Ms2ger | doesn't care about benchmarks much |
| 10:17 | <zcorpan> | Ms2ger: that too |
| 10:19 | <zcorpan> | but in particular, gecko and webkit don't pick up the prefix from the element when there's no ns decl |
| 10:19 | <zcorpan> | and gecko likes throwing weird exceptions and webkit likes returning null, or some such |
| 10:21 | <Ms2ger> | Oh, we do like our weird exceptions |
| 10:22 | <annevk> | zcorpan: you're gonna file bugs on 'endings' not being defined and such? |
| 10:35 | <hsivonen> | the annoying thing about test cases is that they show your patch isn't done yet |
| 10:38 | <smaug____> | hsivonen: whaaaat |
| 10:38 | <smaug____> | what should I review now |
| 10:38 | <annevk> | hsivonen: sounds like writing a standard without sufficient tests |
| 10:38 | smaug____ | scrolls up |
| 10:38 | <annevk> | hsivonen: or a standard with tests, but none written by zcorpan |
| 10:38 | <hsivonen> | smaug____: an upcoming patch for a view source crash |
| 10:39 | <Ms2ger> | smaug____, EVERYTHING! ;) |
| 10:40 | <smaug____> | noooouuu |
| 10:41 | <Ms2ger> | Bah, who reviewed our Blob ctor? |
| 10:41 | <smaug____> | but shadow dom proposal yes |
| 10:41 | smaug____ | doesn't remember if he reviewed that. perhaps not |
| 10:41 | <Ms2ger> | sicking |
| 10:41 | <Ms2ger> | That explains a lot |
| 10:52 | <annevk> | Ms2ger: it doesn't even match the spec? |
| 10:52 | <annevk> | Ms2ger: I have the feeling sicking just guesses what the spec says and then instead of reading it, asks on the mailing list about it |
| 10:52 | <Ms2ger> | Not for new Blob(window), no :) |
| 10:52 | <annevk> | it's really weird |
| 11:09 | <hsivonen> | annevk: in case you are still curious, it turned out to be a 3-line fix: https://bugzilla.mozilla.org/attachment.cgi?id=614726&action=diff |
| 11:18 | <zcorpan> | ok does the endings values in new Blob([], {endings: ... }) include the quotes or not? |
| 11:22 | <Ms2ger> | '"transparent"' or "transparent"? |
| 11:24 | <zcorpan> | yeah |
| 11:25 | <Ms2ger> | The former, per spec, I guess |
| 11:29 | <zcorpan> | sigh, i'm not gonna get to writing tests with a spec like this |
| 11:29 | <Ms2ger> | Does WebIDL actually define to use the defaults for dictionaries? |
| 11:32 | <zcorpan> | when the argument is omitted, you mean? |
| 11:32 | <Ms2ger> | Yes |
| 11:32 | <Ms2ger> | It would be silly to use the default when the argument is present, no? :) |
| 11:33 | <zcorpan> | the property can still be omitted |
| 11:33 | <zcorpan> | new Blob([], {}) |
| 11:34 | <Ms2ger> | Ah, no |
| 11:34 | <Ms2ger> | When the *property* is omitted, I can't read |
| 11:34 | <zcorpan> | http://dev.w3.org/2006/webapi/WebIDL/#dfn-dictionary-member-default-value |
| 11:36 | <zcorpan> | i guess webidl should treat omitted dictionary the same as an empty dictionary |
| 11:36 | <zcorpan> | to not have all specs need to say that |
| 11:41 | <Ms2ger> | "the value the dictionary member is to be considered to have when not present" |
| 11:41 | <Ms2ger> | That could come straight out of a CSS spec |
| 11:42 | <zcorpan> | heh |
| 11:51 | <zcorpan> | "If an invalid key is used or if the dictionary argument is invalid, user agents MUST throw an InvalidStateError." |
| 11:51 | <zcorpan> | aargh |
| 11:53 | <Ms2ger> | Yeah :/ |
| 12:13 | zcorpan | wonders how to test endings:"native" |
| 12:14 | <Ms2ger> | confirm("Are you on Windows?"); |
| 12:14 | <zcorpan> | i guess newlines will always either be one or two chars, so i can test for that |
| 12:40 | <zcorpan> | hmm so what should happen for {endings:"bogus"} |
| 12:40 | <zcorpan> | webidl only says to use the default value when the key is not present |
| 12:40 | Ms2ger | goes for throwing |
| 12:41 | <Ms2ger> | "If S is not one of E�s enumeration values, then throw a TypeError." |
| 12:42 | <zcorpan> | values aren't necessarily enumerated |
| 12:42 | <zcorpan> | e.g. type can be anything |
| 12:43 | <Ms2ger> | endings should be an enum, I guess |
| 12:43 | <Ms2ger> | So, is there a way to get the value of endings out again? |
| 12:44 | <zcorpan> | not directly |
| 12:44 | <zcorpan> | making endings enum might be a good idea |
| 12:58 | <smaug____> | someone should write a good article about MutationObserver |
| 12:59 | smaug____ | is obviously trying to delegate some work. |
| 13:18 | zcorpan | decides to go with var isWin = navigator.platform.substr(0, 3) == 'Win'; to test native endings |
| 13:32 | <zcorpan> | what's an example of a platform array object? |
| 13:42 | <annevk> | smaug____: preferably one that can serve as introduction in the specification |
| 13:43 | <annevk> | hsivonen: code looks daunting |
| 13:43 | <annevk> | zcorpan: oh you found normative requirements around "endings", nice |
| 13:44 | <zcorpan> | does <select> support indexed properties? |
| 13:44 | <zcorpan> | it has getter Element item(unsigned long index); but i can't find that it says "supports indexed properties" |
| 13:45 | <annevk> | might be a bug then per IDL "getter" requirements |
| 13:46 | <zcorpan> | oh, the magic phrase is "determine the value of an indexed property" |
| 13:49 | <zcorpan> | hmm, no, the idl getter and long is enough |
| 13:49 | <zcorpan> | http://dev.w3.org/2006/webapi/WebIDL/#dfn-indexed-property-getter |
| 13:58 | <zcorpan> | heycam|away: you know if anything uses webidl "platform array object"? |
| 14:07 | <annevk> | zcorpan: fwiw, I often use "<code>stringvalue</code>" in specs, is that confusing? |
| 14:08 | <annevk> | though not entirely consistently if it's both a name and a potential string value |
| 14:08 | <annevk> | like say utf-8 |
| 14:08 | <annevk> | that's unquoted |
| 14:08 | <zcorpan> | i think that's fine; file-api had <dfn>"native"</dfn> |
| 14:09 | <annevk> | oh |
| 14:09 | <zcorpan> | and "a DOMString which can take the values "transparent" or "native"" |
| 14:09 | <annevk> | when spec writers don't pay attention to details it gets messy |
| 14:10 | <annevk> | anyway, back tonight |
| 14:24 | <zcorpan> | hmm. should new Blob() take ArrayBuffer or ArrayBufferView? |
| 14:25 | <zcorpan> | the latter, right? |
| 14:27 | <zcorpan> | man this is probably a record on bugs for a single feature for a day |
| 14:30 | <Ms2ger> | Blame Typed Arrays for that :) |
| 16:35 | <charlvn> | what's going on with this page - http://www.whattf.org/mailing-list - it's being returned as text/plain |
| 16:48 | <Hixie> | whattf? |
| 16:48 | <Hixie> | try whatwg :-) |
| 17:02 | <gsnedders> | Oh I bet I know what caused my Gmail account to get suspended again. I logged in from Windows… Or at least that's the only thing different about last night. |
| 17:16 | <charlvn> | Hixie: yeah it's been a very long time since i last looked at that site - not sure who is running / maintaining it |
| 17:17 | <charlvn> | the domain is registered by an Elika Etemad |
| 17:17 | <TabAtkins> | That would be fantasai from the CSSWG. |
| 17:18 | <charlvn> | ah |
| 17:18 | <charlvn> | strange that i recognise fantasai and not her real name :) |
| 17:18 | <Hixie> | gsnedders: send me an e-mail to ianh⊙gc and i'll poke around again |
| 17:19 | <charlvn> | talking about whatwg.org - the one block looks a little bit strange in firefox 11 on ubuntu 11.10 http://i.imgur.com/Odmex.png |
| 17:19 | <charlvn> | in chromium it's perfect http://i.imgur.com/zKaw9.png |
| 17:20 | <TabAtkins> | Different line-wrapping. |
| 17:20 | <charlvn> | yeah |
| 17:20 | <smaug____> | different font |
| 17:20 | <charlvn> | not that it really matters or anything, just something i noticed |
| 17:21 | <kbrosnan> | different fonts have different widths |
| 17:21 | <kbrosnan> | the one ubuntu chooses for the default font causes extra wrapping |
| 17:23 | <kbrosnan> | you can try using something like http://en.wikipedia.org/wiki/Liberation_fonts which has similar metrics to the MS core fonts |
| 17:27 | <tantek> | well this is an funny snapshot: http://www.whattf.org/specs/web-apps/current-work/ |
| 17:29 | <charlvn> | 7 July 2008 - nice |
| 17:29 | <charlvn> | ok i'm off - bbl |
| 17:46 | <annevk> | inbox explosion |
| 17:56 | <annevk> | kennyluck: online? |
| 17:56 | <annevk> | hmm, it seems to be 3AM in Tokyo |
| 17:57 | <kennyluck> | annevk, I am no longer in Tokyo, but what's up? |
| 17:58 | <annevk> | oh okay |
| 17:58 | <annevk> | kennyluck: wondering about a summary of the mailing list discussion since running it all through Google Translate would be cumbersome |
| 17:58 | <annevk> | where are you now? |
| 18:01 | <kennyluck> | annevk, I am in Beijing now. So, the mailing list discussion… I don't think there's any good progress, besides two Taiwanese friends of mine suggest that merging big5-hkscs into big5 is not a good idea. |
| 18:01 | <annevk> | btw, does anyone know if Gecko/WebKit have open bugs on fixing the way they represent characters in URL that cannot be expressed in a certain encoding? |
| 18:01 | <annevk> | kennyluck: k, but no data? |
| 18:01 | <annevk> | kennyluck: I saw something about a lang= attribute too |
| 18:02 | <annevk> | kennyluck: what's their solution when using Chrome? |
| 18:02 | <kennyluck> | I somehow come to think there's not likely to be more big5-hkscs than big5-uao in Taiwan for a very simple reason: big5-hkscs doesn't support Japanese well, and that was a main use case why people want to extend big5 at all. |
| 18:03 | <kennyluck> | annevk, nothing to scientific but I start to collect a list of those that are best decoded as big5-hkscs or big5-uao → http://www.w3.org/html/ig/zh/wiki/Big5#big5-uao |
| 18:03 | <kennyluck> | in terms of data |
| 18:03 | <kennyluck> | foolip is working on something that'll be more systematic. |
| 18:04 | <annevk> | yeah, the question is more I guess how much extended big5 there is in Taiwan to begin with |
| 18:04 | <annevk> | I've seen threads on people having to override Chrome because it does not do hkscs, I've seen no such thing about uao |
| 18:06 | <kennyluck> | annevk, for the @lang thing. A guy suggested that we might want to pick either big5-hkscs or big5-uao based on the @lang attribute, and I was replying and said it wasn't really help much because the in the list of urls foolip collected, there's only about 10% that has @lang specified. |
| 18:06 | <annevk> | kk |
| 18:06 | <kennyluck> | I think this idea is just not worthy. |
| 18:07 | <annevk> | which idea? |
| 18:07 | <kennyluck> | The lang idea. |
| 18:07 | <kennyluck> | annevk, by the way, I noticed that HK people use big5-uao too, specifically for Japanese content. I collected two examples in the link above. |
| 18:08 | <annevk> | good times |
| 18:09 | <annevk> | evolving big5 without coordination was such a stupid idea |
| 18:09 | <TabAtkins> | s/without coordination/ |
| 18:10 | <TabAtkins> | s/evolving// |
| 18:12 | <kennyluck> | It's easier to collect big5-uao content: for example the sequence \x92\xd4 is 実 in big5-uao, a rather common Kanji, and 琔 in big5-hkscs, which is a rather uncommon Chinese character I can't pronouce. So if you Google 琔, you get those big5-uao pages, and that's because Google decodes content as big5-hkscs. |
| 18:13 | <kennyluck> | Well, a bunch of pages that actually use the char 琔, but a little bit that uses 実. |
| 18:14 | <kennyluck> | I haven't found a way to get big5-hkscs specific pages so I am waiting for foolip's work :p |
| 18:19 | <annevk> | I think foolip is looking at .tw pages only |
| 18:20 | <annevk> | pretty awesome he's helping out, he had a nice strategy for getting more content to study |
| 18:28 | <gsnedders> | Hixie: emailed, thx |
| 18:28 | <gsnedders> | Hixie: (and now you know my uni email :P) |
| 18:54 | <annevk> | zewt: you're not answering my question |
| 18:54 | <annevk> | zewt: my question was if there's anything that does not deal with CRLF |
| 18:54 | <annevk> | zewt: obviously the answer is not going to be Notepad, because that's all it deals with |
| 19:30 | <annevk> | in a """ Python block \n still works? |
| 19:31 | <gsnedders> | Yes. |
| 19:51 | <annevk> | ah shit |
| 19:51 | <annevk> | forgot to explain in my email why I limited things to 0x10000 |
| 19:52 | <annevk> | btw, I tested some of the alternatives and none were really faster |
| 19:52 | <annevk> | the problem is in the amount of data (just over a million iterations), not the individual aspects |
| 19:53 | <annevk> | oh sweet, http://lists.w3.org/Archives/Public/www-archive/2012Apr/att-0039/test-x-mac-cyrillic.html run directly |
| 19:53 | <annevk> | XSS galore |
| 19:53 | <annevk> | (I didn't attempt to do anything nasty though) |
| 19:55 | <annevk> | email itself: http://lists.w3.org/Archives/Public/www-archive/2012Apr/0039.html |
| 20:03 | <zewt> | annevk: um, you don't want to be writing CRLF in *nix, heh |
| 20:04 | <zewt> | a great way to irritate users (who, being the weirdos running linux on a desktop, are even more likely to notice and complain about it) |
| 20:05 | <annevk> | dunno if I ever noticed that really |
| 20:05 | <annevk> | maybe my software was too evolved; I mean, dealing with CR, LF, and CRLF is a major hassle |
| 20:07 | <Ms2ger> | Eww, CRLFs |
| 20:09 | <zewt> | anyone know off-hand if {a:1, a:2} is guaranteed in JS to be {a:2} (and never a:1)? |
| 20:10 | <annevk> | zewt: per Web ECMAScript handmade dictionaries are ordered I think |
| 20:11 | <zewt> | i mean, the second duplicate 'a' always clobbers the first |
| 20:13 | <Ms2ger> | Yes, it does |
| 20:13 | <Ms2ger> | See http://es5.github.com/#x11.1.5 |
| 20:13 | <zewt> | (not terribly important, just affects how I answer someone's question) |
| 20:13 | <annevk> | oh sweet, the tests when hosted on lists.w3.org are all screwed because of some default charset setting |
| 20:14 | <Ms2ger> | Chaaaaaaaaaarseeeeeeeeets |
| 20:14 | <zewt> | as a random gruntly aside, MongoDB by design depends on the order of dictionaries, which makes me unhappy |
| 20:14 | <annevk> | I don't understand why languages don't have ordered dictionaries by default |
| 20:15 | <annevk> | I think Python might have something now, but no convenient syntax |
| 20:15 | <zewt> | well, it seems like optimizing for it would compete with optimizing for fast lookups |
| 20:15 | <zewt> | python has ordereddict i think, but without a builtin syntax it's cumbersome |
| 20:17 | <annevk> | heh ordered dict support used html5lib as argument http://www.python.org/dev/peps/pep-0372/ it's in Python 2.7; I have 2.6 |
| 20:50 | <annevk> | welcome https://lkml.org/lkml/2012/4/13/180 o_O |
| 20:50 | <annevk> | via mpt |
| 20:51 | <zewt> | where's that troll stench coming from |
| 21:59 | <Hixie> | abarth: i think a document's sandboxing flags must be set, when the Document is created, to the union of the sandboxing flags on the document's container iframe, and the sandboxing flags on the document's container iframe's document, and the sandboxing flags that come from CSP |
| 21:59 | <Hixie> | abarth: does that sound right? |
| 21:59 | <Hixie> | abarth: i think that automatically makes the flags transitive across multiple levels of nested iframes |
| 22:00 | <abarth> | correct |
| 22:01 | <Hixie> | ok |
| 22:01 | <Hixie> | that seems tractable |
| 22:01 | <abarth> | http://trac.webkit.org/browser/trunk/Source/WebCore/loader/FrameLoader.cpp#L3111 |
| 22:01 | <abarth> | "m_forcedSandboxFlags" comes from CSP |
| 22:02 | <Hixie> | ooh, "forced", good terminology |
| 22:02 | <abarth> | and possibly from the app that's embedding WebKit, but that's not relevent for the spec |
| 22:04 | <abarth> | we compute them when we're initializing the security origin and such for the document: http://trac.webkit.org/browser/trunk/Source/WebCore/dom/Document.cpp#L4806 |
| 22:53 | <Hixie> | abarth: ok, i've updated the spec |
| 22:53 | <abarth> | Hixie: thanks! |
| 22:53 | <Hixie> | abarth: there's a "forced sandboxing flag set" http://www.whatwg.org/specs/web-apps/current-work/#forced-sandboxing-flag-set |
| 22:54 | <Hixie> | abarth: that you should set on the resource as it is loaded, based on its HTTP headers |
| 22:54 | <Hixie> | just say something like "A resource with the foo header with the bar value has a /forced sandboxing flag set/ with the following flags set:" |
| 22:54 | <abarth> | ok |
| 22:54 | <abarth> | let me try to do that now |
| 22:55 | <Hixie> | and then copy the way i phrased it for iframes, here: http://www.whatwg.org/specs/web-apps/current-work/#attr-iframe-sandbox |
| 22:55 | <Hixie> | but note i'm about to add a new flag |
| 22:55 | <Hixie> | allow-popups |
| 22:55 | <abarth> | i defer to HTML to parse the string anyway |
| 22:55 | <Hixie> | maybe i should just give you a shorthand to set all the flags at once |
| 22:55 | <Hixie> | oh |
| 22:55 | <Hixie> | hm |
| 22:55 | <Hixie> | interesting |
| 22:55 | <abarth> | or, i would prefer that |
| 22:55 | <abarth> | i get a string from the header parsing algorithm |
| 22:56 | <abarth> | and i'd like to just have the HTML spec parse it |
| 22:56 | <Hixie> | ah ok |
| 22:56 | <Hixie> | hm |
| 22:56 | <Hixie> | ok, that will take longer. i'll get back to you! |
| 22:56 | <abarth> | directive-name = "sandbox" |
| 22:56 | <abarth> | directive-value = token *( 1*WSP token ) |
| 22:56 | <abarth> | token = <token from RFC 2616> |
| 22:57 | <abarth> | i can change that if necessary |
| 22:57 | <Hixie> | that's just your authoring conformance requirements, doesn't affect me at all |
| 22:57 | <Hixie> | but i'll give you a hook you can use |
| 22:58 | <abarth> | thx |
| 23:00 | <abarth> | <p>When enforcing the <code>sandbox</code> directive, the user agent |
| 23:00 | <abarth> | MUST parse the <code>directive-value</code> as an unordered set of unique |
| 23:00 | <abarth> | space-separated tokens that are ASCII case-insensitive. The user agent |
| 23:00 | <abarth> | MUST include these tokens in the forced sandboxing flag set for the |
| 23:00 | <abarth> | protected document.</p> |
| 23:00 | <abarth> | how does that look? (I'll add links to the relevant definitions.) |
| 23:02 | <Hixie> | i'm doing the splitting for you |
| 23:02 | <Hixie> | (the /forced sandboxing flag set/ takes flags, not tokens) |
| 23:02 | <Hixie> | (the tokens actually remove certain flags) |
| 23:04 | <abarth> | ic |
| 23:05 | <abarth> | (yeah, I don't want to list the valid tokens because I'd like HTML to be able to add them without having to change the CSP spec) |
| 23:05 | <Hixie> | ok here's what you need to say: |
| 23:06 | <Hixie> | <p>[When enforcing the <code>sandbox</code> directive, ] the user agent MUST /parse the sandboxing directive/ using [the <code>directive-value</code>] as the /input/ and [the resource's] /forced sandboxing flag set/ as the output.</p> |
| 23:06 | <Hixie> | the bits in square brackets are whatever you need on your end |
| 23:07 | <abarth> | ok, great |
| 23:07 | <abarth> | thanks |
| 23:07 | <Hixie> | this has to be done as you receive the http headers, before i create teh Document |
| 23:08 | <abarth> | ok. i've been writing these as static requirements, but I can change it to be triggered on receiving the header |
| 23:08 | <abarth> | the <meta> tag is gone now, so the only entry point is when processing headers |
| 23:09 | <abarth> | <p>Upon receiving an HTTP response containing at least one |
| 23:09 | <abarth> | <code>Content-Security-Policy</code> header field, the user agent |
| 23:09 | <abarth> | MUST <a href="#enforce">enforce</a> the policy contained in the |
| 23:09 | <abarth> | <em>first</em> such header field. The user agent MUST ignore |
| 23:09 | <abarth> | subsequent such header fields.</p> |
| 23:09 | <abarth> | so, it looks like "enforce" gets called at the right time |
| 23:11 | <abarth> | yeah, we're using the "resource representation" terminology in this spec |
| 23:11 | <abarth> | so, that should all work out fine |
| 23:12 | <abarth> | it's slightly confusing that we call it the "protected document" even though it's really the HTTP response that will be used to create the document |
| 23:13 | <abarth> | tlr will be sad that I'm going to link to the whatwg version of the spec, but I don't think that can be helped :) |
| 23:15 | <Hixie> | the w3c version will actually contain this stuff too |
| 23:15 | <Hixie> | until someone asks for a revert, anyway |
| 23:15 | <abarth> | I think i'm going to rename "protected document" to "protected resource" to make it clearer that this is happening before the document is created |
| 23:15 | <Hixie> | yeah calling things documents that aren't Documents is confusing |
| 23:15 | <zewt> | docun'ts |
| 23:15 | <Hixie> | also even if you did have the <meta> it wouldn't matter here cos it's too late by the time you see hte <meta> to add sandboxing flags |
| 23:16 | <abarth> | yeah |
| 23:16 | <abarth> | if we add it back, i'll add a note explaining that |
| 23:16 | <Hixie> | ok allow-popups is next on the pile |
| 23:16 | <Hixie> | microsoft will be happy |
| 23:16 | <abarth> | how often does http://dev.w3.org/html5/spec/single-page.html update? |
| 23:17 | <abarth> | (or rather, is there a w3c version that has this stuff now?) |
| 23:17 | <Hixie> | should be |
| 23:17 | <Hixie> | MikeSmith: any idea why that's not updating? |
| 23:18 | <Hixie> | i imagine the date is wrong because they were publishing so the date is faked right now |
| 23:18 | <Hixie> | that's also why it'll say WD and not ED |
| 23:18 | <Hixie> | looks like it should work on my side |
| 23:19 | <Hixie> | dunno |
| 23:19 | <Hixie> | bbl |
| 23:27 | <abarth> | Hixie: ok, pushed. thanks for your help! |
| 23:50 | <SlydidarSlider> | I'm not going to study logic, I don't want to... what? wait, what? :) |
| 23:51 | <SlydidarSlider> | bye |
| 23:51 | SlydidarSlider | waves |
