| 00:18 | <Hixie> | anyone have an installation of IE that handles flash plugins correctly? |
| 00:18 | <zcorpan_> | i do (i think) |
| 00:19 | <Hixie> | can you go to: http://software.hixie.ch/utilities/js/live-dom-viewer/ |
| 00:19 | <Hixie> | and click "download" |
| 00:20 | <zcorpan_> | sec |
| 00:20 | <Hixie> | and see if it says "pass" in the rendered view? |
| 00:20 | <Hixie> | (i think that's your flash test file, in fact) |
| 00:21 | <zcorpan_> | heh |
| 00:21 | <zcorpan_> | it says pass |
| 00:21 | <Hixie> | ok hold on |
| 00:23 | <zcorpan_> | not in ie6 though, it seems... |
| 00:23 | <Hixie> | download again and see what the log says (bottom box) |
| 00:24 | <zcorpan_> | rendering mode: CSS1Compat |
| 00:24 | <zcorpan_> | document has no title |
| 00:24 | <zcorpan_> | log: [object] |
| 00:24 | <zcorpan_> | log: [object] |
| 00:24 | <zcorpan_> | log: undefined |
| 00:24 | <zcorpan_> | but there's more... |
| 00:25 | <Hixie> | interesting |
| 00:25 | <zcorpan_> | there's an infobar warning me about a blocked download |
| 00:25 | <Hixie> | so basically IE doesn't let you access the Document of a full-plugin pyage |
| 00:25 | <Hixie> | page |
| 00:26 | <zcorpan_> | if i click the infobar and click "download", the page reloads and the log is: |
| 00:26 | <zcorpan_> | rendering mode: CSS1Compat |
| 00:26 | <zcorpan_> | document has no title |
| 00:27 | <zcorpan_> | plus there's a second box below that says "This script puts a function w(s) into the global scope of the test page, " |
| 00:27 | <Hixie> | yeah i just added that box just now |
| 00:27 | <Hixie> | ignore that :-) |
| 00:27 | <zcorpan_> | ok |
| 00:28 | <Hixie> | wait if you reload and click "download" again it doesn't show the flash file at all? |
| 00:28 | <zcorpan_> | now it does |
| 00:29 | <zcorpan_> | rendering mode: CSS1Compat |
| 00:29 | <zcorpan_> | document has no title |
| 00:29 | <zcorpan_> | log: [object] |
| 00:29 | <zcorpan_> | log: [object] |
| 00:29 | <Hixie> | then an exception is raised? |
| 00:29 | <Hixie> | (should show little yellow warning icon in the status bar if so) |
| 00:29 | <zcorpan_> | yeah |
| 00:29 | <zcorpan_> | "Permission denied" or something |
| 00:30 | <Hixie> | k |
| 00:30 | <zcorpan_> | ie's error console isn't very helpful :( |
| 00:31 | <Hixie> | heh |
| 01:22 | <zcorpan_> | as a fulltime job, how many testcases can one create in one day, in avarage? |
| 01:23 | <zcorpan_> | given you know the spec you're writing test cases against and you know how to write test cases? |
| 01:25 | zcorpan_ | is trying to estimate how many test cases he will be able to write this summer |
| 01:25 | <Lachy> | zcorpan_, given roughly 5 minutes per test case, you could get close to 100 per day |
| 01:25 | <Lachy> | that 5 min includes writing and testing in several browsers, and fixing any mistakes you find in your test |
| 01:26 | <Hixie> | zcorpan_: depends |
| 01:26 | <Hixie> | zcorpan_: some test cases take upwards of 2 hours to make |
| 01:26 | <Hixie> | zcorpan_: and then you can write a script that spits out 100 with just 30 minutes of work |
| 01:26 | <zcorpan_> | Hixie: yeah, indeed |
| 01:27 | <Hixie> | zcorpan_: i'd estimate 1 hour per test per person on average over several months |
| 01:27 | <Hixie> | since you'll also have to comment on the spec, learn the spec, etc |
| 01:27 | <Hixie> | so 40 tests a week |
| 01:27 | <Hixie> | these being high-quality tests |
| 01:27 | <Lachy> | wow, I seriously over estimated! |
| 01:27 | <Hixie> | if you're working fulltime and are doing simple tests, you can get more done |
| 01:27 | <Hixie> | it varies |
| 01:28 | <zcorpan_> | 6 weeks, that's 240 test cases |
| 01:28 | <Hixie> | if you're doing testing exclusively you'll get more done than that, probably |
| 01:28 | <Hixie> | depends on the tests |
| 01:29 | <zcorpan_> | i'll probably be working on the implemented parts of html5 that don't already have tests |
| 01:29 | <zcorpan_> | as a start |
| 01:29 | <Hixie> | like History and Location? |
| 01:29 | <zcorpan_> | yeah |
| 01:30 | <Hixie> | for those 90% of the work will be reverse engineering browsers and trying to work out what parts of the spec are wrong |
| 01:30 | <Hixie> | so 1 hour per test is quite reasonable for those sections :-) |
| 01:30 | <zcorpan_> | probably also things that are new but have experimental implementations, like video |
| 01:34 | <zcorpan_> | if we want to have 20,000 tests by 2010, then we'd need at least 3 people working full-time writing test cases |
| 01:35 | <zcorpan_> | given i didn't miscalculate |
| 01:36 | <karlUshi> | 5 days * 8 hours a day = 40 test cases a week |
| 01:37 | <karlUshi> | for the corner cases that Ian was mentionning |
| 01:37 | <karlUshi> | another thing to take into consideration is the format to report the results |
| 01:38 | <karlUshi> | if we have a common format, it will be a lot easier to produce an interop report |
| 01:39 | <Hixie> | i love how every time i commit something and the script twitters the commit message, WHATWG gets new twitter friends |
| 01:39 | <Hixie> | that's hilarious |
| 01:40 | <zcorpan_> | automatically or because people add whatwg as their twitter friend? |
| 01:41 | <Hixie> | i assume people see WHATWG on the public timeline and go "oo, i should add that" |
| 01:41 | <zcorpan_> | ah |
| 01:49 | <zcorpan_> | calculating again from the other end, if one person would work full-time writing test cases, it would take 11.6 years to get 20,000 tests |
| 01:50 | <Hixie> | yeah but if one person worked full time on tests for 11 years, by the end of it they'd be going way faster than one test every hour |
| 01:50 | <Hixie> | :_) |
| 01:51 | <zcorpan_> | true |
| 01:52 | <Hixie> | ok i am not looking forward to the next section i have to do |
| 01:52 | Hixie | shudders |
| 01:52 | <Hixie> | ("scripting") |
| 01:52 | <zcorpan_> | why not? |
| 01:52 | <Hixie> | mostly i'm not sure what it should say |
| 01:54 | <karlUshi> | Hixie: define only one element ;) no scripting :p |
| 01:55 | karlUshi | wonders if Hixie has a template when he's writing sections. As in "having a set of questions, I should go through to write" |
| 02:04 | <Hixie> | i wonder what happens if you invoke a method associated with a document that isn't the active document in a browsing context |
| 03:30 | <Hixie> | ok i'm basically down to the scripting security model section |
| 05:20 | <paulproteus> | Hixie, That is hilarious re: twitter friends. (-: |
| 05:28 | othermaciej | looks for an antacid and an aspirin before reading cwilso's mail |
| 06:09 | <annevk> | is Simon around? |
| 06:13 | <Lachy> | annevk, Simon is asleep |
| 06:15 | <annevk> | He reported a bug on Opera that <foo><dd></foo> should have closed the <dd> while in fact the </foo> is ignored |
| 06:15 | <annevk> | However, this is exactly what html5lib is doing too |
| 06:16 | <Lachy> | is that what the spec says to do? |
| 06:16 | <annevk> | and Firefox |
| 06:16 | <annevk> | maybe I misread the bug report |
| 06:17 | <annevk> | oh, he uses http://simon.html5.org/test/html/parsing/tree-construction/the-main/how-to/in-body/003.htm as a testcase |
| 06:18 | <annevk> | never mind, that's not what Firefox does |
| 06:19 | <annevk> | it's what html5lib does |
| 06:19 | <Lachy> | IE does the same as FF |
| 06:20 | <annevk> | IE can't be relied upon with unknown elements |
| 06:20 | <Lachy> | oh, right |
| 06:20 | <tantek> | that sounds like an overgeneralization |
| 06:21 | <annevk> | Well, unless you involve namespaces you can't |
| 06:23 | <othermaciej> | can't be relied upon? |
| 06:23 | <othermaciej> | does it do something weird with unknown elements? |
| 06:23 | <annevk> | </foo> becomes an element /foo |
| 06:23 | <othermaciej> | ! |
| 06:24 | <annevk> | all unknown tags basically become void elements |
| 06:24 | <annevk> | same for some "known" elements if they are outside their usual context (td, etc.) |
| 06:24 | <othermaciej> | that is pretty weird |
| 06:25 | <othermaciej> | I see now why they are scared to change anything in parsing |
| 06:27 | <annevk> | whoa, I'm quoted in the news |
| 06:27 | <annevk> | well, some tech site |
| 06:29 | <Lachy> | annevk, pointer to the article? |
| 06:30 | <annevk> | I can't find the original |
| 06:30 | <annevk> | http://risal.wordpress.com/2007/04/19/is-the-web-ready-for-html-5/ |
| 06:30 | <annevk> | has the quotes though |
| 06:30 | <annevk> | I did actually say that, but I didn't think I'd get the attribution |
| 06:31 | <annevk> | “HTML5 is about preserving the information people have accumulated over the years,” Opera spokesperson Anne van Kesteren told internetnews.com. “By remaining backward and forwards compatible, we hope to ensure that people will be able to interpret HTML for decades if not centuries to come.” |
| 06:37 | <annevk> | http://www.internetnews.com/dev-news/article.php/3672011 is the original |
| 06:38 | <annevk> | found it by using http://www.google.com/search?q=%22HTML5+is+about+preserving+the+information+people+have+accumulated+over+the+years%22 |
| 06:39 | <Lachy> | heh, IE actually allows .createElement("/foo") to work |
| 06:39 | <Lachy> | it looks like it's the same article |
| 06:40 | <annevk> | I suppose people are just making copies |
| 06:42 | <Lachy> | do you know where you actually said what they quoted? Did they just copy it from some post to whatwg or public-html? |
| 06:43 | <annevk> | yes, no |
| 06:51 | annevk | wonders how it helps if cwilso became chair given his argument about competitors |
| 06:51 | <annevk> | then again, he said that point wasn't as important as his other |
| 06:52 | <Lachy> | the one about Hixie working for Google? |
| 06:52 | <annevk> | yes |
| 06:53 | <Lachy> | I was surprised about that too, especially since Google isn't competing with IE, they're competing with MS in other areas |
| 07:05 | <annevk> | Google has a Firefox dev team |
| 07:06 | <Lachy> | so? |
| 07:06 | <Lachy> | They also build stuff for IE too, like the google toolbar for instance |
| 07:07 | annevk | would count both as competing |
| 07:16 | <krijnh> | Morning people |
| 07:24 | <annevk> | good afternoon |
| 07:24 | <annevk> | apparently JavaScript isn't as interoperable as claimed: http://my.opera.com/hallvors/blog/2007/04/11/new-adventures-in-date-parsing |
| 07:25 | <annevk> | I suppose it's time for another X5 remark |
| 07:25 | <othermaciej> | ECMAScript5? |
| 07:27 | <annevk> | Yeah, I wonder if they are solving these type of issues too |
| 07:28 | <othermaciej> | they plan to write the spec via a reference implementation in SML |
| 07:28 | <othermaciej> | I think spec by reference implementation is a really bad idea |
| 07:30 | <annevk> | the "Web browsers" section of HTML5 is complicated |
| 07:30 | <annevk> | though browsers are like that, I guess :) |
| 07:31 | <Hixie> | ECMAScript is one of the better-written specs |
| 07:32 | <othermaciej> | it suffers mainly from lack of maintenance, but yes, it is written better than HTML4 certainly |
| 07:34 | <annevk> | "The rules for chosing a browsing context given a browsing context name" don't take into account what happens when there's no parent browsing context |
| 07:34 | <annevk> | for instance |
| 07:42 | <annevk> | So it seems that <foo><dd></foo> should indeed close <dd> |
| 07:42 | <annevk> | as in <foo><dd></foo>x "x" should be a sibling of foo |
| 07:48 | <annevk> | Actually, no... |
| 07:49 | <annevk> | when you hit </foo> <dd> is the bottommost node of the spec |
| 07:49 | <annevk> | s/spec/stack/ |
| 07:49 | <annevk> | <dd> is in the special category and therefore </foo> has to be ignored |
| 08:22 | <othermaciej> | I'm really amused by Murray asking for Dan to explain all the technical details to him on public-html |
| 08:22 | <othermaciej> | even with his 30 years of experience with markup languages |
| 08:22 | <annevk> | apparently he misses all the +1 messages to dbaron his e-mail |
| 08:23 | <annevk> | I was wondering whether I should reply that people who agree with dbaron don't do +1 |
| 08:24 | <othermaciej> | I am amazed at how many times you can make arguments against versioning in the doctype or the namespace URI and people keep bringing it up |
| 08:24 | <othermaciej> | not even refuting the arguments, just acting like they didn't happen |
| 08:25 | <annevk> | yeah, arguments -> wiki |
| 08:26 | <annevk> | everyone just ignored John Boyer :) |
| 08:27 | <othermaciej> | well, he didn't say anything relevant to the group |
| 08:27 | <othermaciej> | I wanted to point out that the Forms task force hadn't started yet, but thought better of it |
| 08:29 | <annevk> | I like what marcos said about XForms Transitional |
| 08:29 | <annevk> | "Seems illogical to me to have transitional technology to something that will be incompatible with the web (XForms)." |
| 08:31 | <Hixie> | the xforms wg thinks the forms task force has started, fwiw |
| 08:31 | <Hixie> | see their minutes |
| 08:32 | <othermaciej> | who do they think is in the task force? |
| 08:32 | <othermaciej> | just them? |
| 08:32 | <Hixie> | no idea |
| 08:32 | <Hixie> | didn't investigate |
| 09:04 | <annevk> | bugmode= ... |
| 09:04 | <annevk> | what if all our testcases don't include that attribute? |
| 09:04 | <annevk> | IE will forever fail? |
| 09:05 | <Lachy> | indeed |
| 09:05 | <annevk> | wfm |
| 09:05 | <annevk> | i don't think we should worry much about it |
| 09:05 | <othermaciej> | it would be good to report its non-conformance accfurately |
| 09:06 | <Lachy> | annevk, as I just wrote in my last email, we should worry about it because all authors would be unconditionally required to use it |
| 09:06 | <annevk> | until MS stops seeing the spec as a guide there's not much point in arguing anyway |
| 09:07 | <annevk> | they seem to think fundamentally different about standards |
| 09:07 | <annevk> | unless cwilso responded to some of Hixie's points about quirks mode in other browsers, specs as guide, etc. already and denied that |
| 09:07 | <Lachy> | yes, and given their monopoly position, that is a problem |
| 09:08 | <annevk> | not much we can do about it on the mailing list it seems |
| 09:08 | <Lachy> | not sure what else we can do that would be even remotely successful |
| 09:08 | annevk | was wondering whether a F2F or telcon would be good for this |
| 09:09 | <Lachy> | it should definitely be discussed in the telcon if we have one |
| 09:10 | <Lachy> | I'll even stay awake for the telcon if it's going to be discussed |
| 09:10 | <annevk> | oh right, a telcon was proposed |
| 09:11 | <annevk> | I don't think I'll make it |
| 09:22 | othermaciej | wonders who Terje Bless is |
| 09:54 | <virtuelv> | othermaciej: think he used to work for w3c, but I can't remember as what |
| 09:55 | <hsivonen> | othermaciej: Terje Bless is from the W3C Validator team |
| 09:55 | <othermaciej> | hsivonen: I'm surprised at his views on implementation conformance |
| 09:56 | <hsivonen> | othermaciej: do you mean on IRC yesterday? |
| 09:56 | <othermaciej> | no, in email today |
| 09:56 | <othermaciej> | his apparent view that an implementation which handles content in a nonconformant way can still be conformant |
| 09:57 | <hsivonen> | so it seems |
| 09:58 | met_ | wonders if case location=url1; location=url2 is specified on html5, cannot found it |
| 09:58 | <hsivonen> | I feel pretty strongly that if MS is going to have a switch no matter what, it should be done with an attribute that the spec makes conforming even if we all hold our noses while doing so |
| 10:00 | <othermaciej> | the spec shouldn't make triggering nonconforming behavior based on that attribute conforming, however |
| 10:00 | <othermaciej> | if there is one |
| 10:01 | <met_> | now is behaviour different among browsers, this is why i care |
| 10:02 | <met_> | location=url1; alert(1); location=url2; alert(2); location=url3 etc. |
| 10:02 | <othermaciej> | met_: I specified it in the Window spec I think; I believe the last location within a single batch of script execution wins |
| 10:02 | <met_> | shouldn't be specified recommended behaviour? |
| 10:03 | <othermaciej> | <script>location = l1; location = l2;</script> --> you end up at l2 |
| 10:03 | met_ | is looking on window spec |
| 10:03 | <othermaciej> | <script>location = l1;</script><script>location = l2;</script> |
| 10:03 | <othermaciej> | that ends up at l1 |
| 10:05 | <hsivonen> | othermaciej: if a conformance checker doesn't pass an IE bug mode attribute as conforming, MS will use a switch that is cloaked from conformance checkers |
| 10:06 | <hsivonen> | othermaciej: the net result would be no change to the verdict given by a conformance checker, but there'd be more pain for intermediate tool vendors who build on XML tooling |
| 10:08 | <hsivonen> | othermaciej: I would be OK, though, with making the conformance checker only treat the latest IE bug mode opt-in as conforming |
| 10:08 | <othermaciej> | hsivonen: I'm not opposed to such an attribute, I am just saying it should not be conformant for implementations to actually introduce bugs based on the value or absence of a bugmode attribute |
| 10:09 | <hsivonen> | othermaciej: oh, you were talking about UA conformance. I was talking about document conformance. |
| 10:11 | <othermaciej> | hsivonen: I understood what you meant; sorry for not being sufficiently clear myself |
| 13:18 | <virtuelv> | what does it take to convince the editor that the WA1.0 spec needs to be split into multiple documents |
| 13:19 | <krijnh> | zcorpan_: it probably should, yeah |
| 13:19 | <krijnh> | virtuelv: http://www.whatwg.org/specs/web-apps/current-work/multipage/ ? |
| 13:19 | <virtuelv> | $ lynx -nolist -dump http://whatwg.org/specs/web-apps/current-work/ | wc -w |
| 13:19 | <virtuelv> | 150202 |
| 13:21 | <virtuelv> | krijnh: I'd like something in between |
| 13:21 | <virtuelv> | 3. on one page, then 3.1. on another probably isn't the best thing either |
| 13:21 | <virtuelv> | but thanks anyway, it's much better to work with those slightly smaller docs |
| 13:22 | <Lachy> | virtuelv, yeah, splitting on just the higher level headings, rather than the sub headings would be sensible |
| 13:22 | <krijnh> | Philip` can probably fix that |
| 13:23 | <Lachy> | I don't like the split version though, it makes searching through it annoying |
| 13:23 | <krijnh> | Afaik he made the current multipage stuff |
| 13:23 | <Lachy> | I wish whatwg.org/html5 redirected to the full page version instead |
| 13:43 | <Philip`> | Lachy: I think splitting on just the higher levels (1., 2., 3., etc) wouldn't be much help, since section 3 is about half of the whole document so it'd still be huge |
| 13:44 | <Philip`> | (so you might as well use the single-page version, if you have a computer/browser that doesn't fall to pieces when trying to render it) |
| 13:45 | <Philip`> | ((I don't know why people actually have the problems they've reported - for me (in Opera and Firefox) it just takes ten seconds to load and then it's perfectly fine...)) |
| 13:45 | <Lachy> | doesn't everyone have a dual core, hyperthreaded Pentium 4 with 2GB of RAM that can handle such documents with ease? ;-) |
| 13:46 | <Philip`> | Mine's only single core and the hyperthreading is disabled ;-) |
| 13:46 | <Lachy> | actually, I got the one without hypertheading, but it's dual core |
| 13:47 | <Lachy> | would it be much effort to special case section 3 then, so that it gets split on the sub headings, but the smaller sections were split on the higher level? |
| 13:47 | <Lachy> | or would that just be illogical |
| 13:48 | <Philip`> | virtuelv: The nearly-empty pages for e.g. "3." should be considered a bug - it's designed to work that way, but the design should be changed to merge nearly-empty pages with more contentful pages (but I don't have time to work on that now/soon) |
| 13:49 | <Philip`> | Lachy: Section 3 is actually special-cased already, to split on some third-level headings (so that the big areas like video and canvas are separated) |
| 13:50 | <Lachy> | oh, ok |
| 13:50 | <Philip`> | If I remember correctly, I was trying to keep each of the pages under about 100KB |
| 13:52 | <Philip`> | It should be easy to do the splitting in a different way, so the individual sections are much larger - I guess it'd just be a bit confusing to have three official versions of the same document, and I don't know which variations would be the best to keep |
| 13:55 | <Philip`> | (http://html5.googlecode.com/svn/trunk/spec-splitter/spec-splitter.py in the place where it has 'h3' is how it decides what to split on, and that could just be changed to something different) |
| 13:58 | <Lachy> | I think I finally figured out Chris' insane logic for wanting for wanting version numbers in the DOCTYPE and how that relates to the bug mode switch! |
| 13:58 | <Lachy> | :-) |
| 13:59 | <Lachy> | I think the logic is this:... |
| 14:00 | <Lachy> | 1. <!DOCTYPE html PUBLIC "html5"> triggers HTML5 mode, which will eventually become a frozen bug state |
| 14:00 | <Lachy> | 2. After it becomes a frozen bug state, IE introduces another form of explicit opt-in. e.g. <html ie-version="9"> |
| 14:01 | <Lachy> | 3. IE continues updating the bug mode switch with each new release for several years, each time improving their compliance |
| 14:02 | <Lachy> | 4. HTML6 is released a while after that with a new DOCTYPE. <!DOCTYPE html PUBLIC "html6"> |
| 14:02 | <Lachy> | 5. That provides a brand new opt-in, and by that time IE will be compliant enough to not need a specific bug mode opt-in |
| 14:07 | <Philip`> | I'm not sure about step 5 - I've thought it is "5. That provides a brand new opt-in, which can replace the proprietary ie-version="13", but they still don't quite comply with HTML6, so they'll add a proprietary ie-version="14" back in for the next release" |
| 14:07 | <virtuelv> | Lachy: that is going to be a nightmare for every other browser vendor |
| 14:07 | jdandrea | is catching up on public-html emails ... ! |
| 14:07 | <Lachy> | Philip`, yes, there's always the possibility that MS will keep repeating the cycle |
| 14:08 | <Lachy> | virtuelv, I know. As I said, it's *insane* logic |
| 14:08 | <Philip`> | and then "6. After enough iterations (hopefully only one, but maybe that's optimistic and they want to plan ahead), the rate of new spec versions will be equal to the rate at which new IE versions break sites, so they will converge and ie-version will never be necessary again" |
| 14:08 | Lachy | is also catching up. I have about 242 to read |
| 14:08 | <virtuelv> | Lachy: if then Opera, Mozilla and Safari added their own <html [vendor]-version="n" and multiple incompatible attributes were selected, which one should trump |
| 14:09 | <Lachy> | virtuelv, I have no idea. I hope no other browser but IE introduces a bug mode switch |
| 14:09 | <jdandrea> | Lachy: Ow! OK, I thought I was in worse shape but I'm current as of Thursday 23:00 -0400 (just got past Chris' official response and Hixie's response to that ...) |
| 14:09 | <Lachy> | I wish IE wouldn't either |
| 14:10 | <Dashiva> | IE8 should automatically reform pages to fix its own bugs, sort of like opera's browserjs :) |
| 14:10 | <Lachy> | some of them I already read, but since I was away for the WAF WG meeting, using a crappy email client, I was unable to process them all well |
| 14:12 | <Philip`> | (Assuming this is inevitable, the job of the HTML WG should perhaps be to make step 6 happen as quickly as possible, by making the first IE-with-HTML5 as compliant as possible, presumably by following IE7's behaviour when sane and by providing enough test cases for the IE developers to know what they're doing wrong before release) |
| 14:12 | <Lachy> | right, so what IE is really after is another Transitional DOCTYPE with which they can abuse for their opt-ins |
| 14:12 | <jdandrea> | I was just speaking with a former editor for ANSI C++ about the whole versioning issue (briefly). His $0.02: Detecting multiple versions of any language/markup/etc. are a nightmare. Old versions stay around much longer than anyone imagines. (Then, with no prodding from me, he cited Visual C++ v6 as an example.) :) |
| 14:14 | <Lachy> | jdandrea, yes. Once IE introduces a frozen bug state, it will be up to them to maintain that state forever. |
| 14:14 | <jdandrea> | Ugh. |
| 14:14 | <jdandrea> | He then mentioned that Python does some interesting things WRT versions. I responded: "I don't see anything like a version switch." He replied: "I'm thinking about 'from _future_ import ...'" ! |
| 14:15 | <Lachy> | I don't know what _future_ is in python |
| 14:15 | <jdandrea> | Future versions that aren't known or haven't been released, IIRC. |
| 14:15 | <Philip`> | jdandrea: I don't think one can complain (in the sense of blaming MS) about VC++6 being non-standard-compliant, since it was released before the standard, nor about VC++2003 since that did a good job of being standard-compliant; but it's fair to complain that it took them five years between those releases |
| 14:16 | <jdandrea> | Philip ` - agreed, and I think that was his ultimate point, the length of time it took to make good. |
| 14:17 | <Philip`> | Lachy: Python 2.5 adds the keyword "with", but it's disabled by default (since it would break compatibility with existing code that uses variables/etc named "with") |
| 14:17 | <Philip`> | In Python 2.6, it will be enabled by default, so those programs have to be fixed before then |
| 14:17 | <Philip`> | In the meantime, Python 2.5 programs can write "from __future__ import with" to get the 2.6-like behaviour |
| 14:18 | <jdandrea> | There you go. |
| 14:18 | <Lachy> | unfortunately, that sort of method couldn't work with HTML though |
| 14:19 | <jdandrea> | Aye. I figured it was worth getting other POVs to see if there's any more precedent we can bring to the table, but I gather this horse is beaten to a pulp. |
| 14:19 | <zcorpan_> | quirks mode will be around forever... |
| 14:20 | zcorpan_ | intends to figure out quirks mode and spec it, unless someone else does it |
| 14:20 | <Lachy> | zcorpan_, good luck getting IE to conform to your spec :-) |
| 14:20 | <zcorpan_> | yeah |
| 14:21 | <jdandrea> | LOL |
| 14:21 | <jdandrea> | Bizarro-HTML: http://lists.w3.org/Archives/Public/public-html/2007Apr/1244.html ... sigh ... |
| 14:21 | <Lachy> | it's clear that IE won't make any changes at all to quirks mode, except maybe adding new elements (I assume the new HTML5 elements will be supported in quirks mode) |
| 14:21 | <zcorpan_> | they'd only do that if they become a minority vendor |
| 14:21 | <zcorpan_> | Lachy: it's not clear to me they will add new features to quirks mode |
| 14:21 | <Philip`> | With Python programs, you'd fix them (preferably as soon as Python 2.5 is released and gives you warnings that your variable called "with" is going to break in the next release) or stick with an old version of Python - that doesn't work at all when those 'programs' are HTML pages that you can't change and are owned by somebody who isn't going to change them ever :-( |
| 14:22 | <zcorpan_> | but nevertheless <canvas> is already used in quirks mode |
| 14:22 | <zcorpan_> | for one |
| 14:22 | <Philip`> | Maybe they would decide the amount of <canvas> content is far less than the amount of quirks-mode content that would break if they started handling <canvas> |
| 14:23 | <Lachy> | zcorpan_, if they didn't allow them in quirks mode, it would be a way to force all HTML5 pages to use standards mode, but AFAIK, other browsers are going to make them work in quirks |
| 14:23 | <Lachy> | <canvas> already works in quirks, for instance |
| 14:23 | <Philip`> | (I suppose that's where you need real statistics of how many pages use <canvas> accidentally without wanting the HTML5 interpretation of it) |
| 14:23 | <zcorpan_> | Lachy: indeed |
| 14:23 | <zcorpan_> | Philip`: google for "canvas demo" |
| 14:23 | <zcorpan_> | and check the first few |
| 14:26 | <Philip`> | Yep, those are the real <canvas> content that would be fixed if IE handled it in quirks mode - the problem (maybe less so for canvas than for other elements) is old sites that used <canvas> when its meaning was undefined and they never noticed because every browser ignored it and just displayed the content, and those would be broken if IE started to handle it |
| 14:27 | <Philip`> | Presumably Microsoft needs to be convinced that the second group is nonexistent or negligible compared to the first group, otherwise they wouldn't want to risk handling <canvas> in case it breaks old sites |
| 14:28 | <virtuelv> | heh, Bizarro-HTML it truly is |
| 14:34 | jdandrea | is finally caught up (catches breath). |
| 14:35 | <krijnh> | I'm still 143 mails behind :/ |
| 14:36 | <zcorpan_> | ah! html60 has a name |
| 14:36 | <krijnh> | Dmitry Turin ? |
| 14:36 | <zcorpan_> | yeah |
| 14:37 | <met_> | any way how subscribe for receiving html-wg mails and not be html-wg member? |
| 14:37 | <zcorpan_> | met_: don't think so |
| 14:37 | <krijnh> | Perhaps only via RSS |
| 14:38 | <met_> | krijnh, rss? which url? |
| 14:38 | <krijnh> | http://lists.w3.org/Archives/Public/public-html/ has a feed |
| 14:38 | <krijnh> | And use RssFwd to mail it to you :) |
| 14:38 | <Philip`> | That seems to not have the actual content of the messages, so you've still got to go to the archive to read them |
| 14:38 | met_ | will try, thanks |
| 14:39 | <krijnh> | Ah |
| 14:39 | <krijnh> | Then it doesn't make sense to use it |
| 14:39 | <met_> | no problem with the archive i want to see whic message i read and which not 8-) |
| 14:41 | <Philip`> | The real problem with versioning is that it assumes there will be another version of HTML after HTML5 - but that cannot possibly happen, because Dmitry has already taken the name HTML6 and it would be silly to jump straight from HTML5 to HTML7 and so development of HTML will have to stop |
| 14:45 | <Lachy> | lol |
| 14:57 | <Philip`> | [Correction from some time ago: in Python 2.5 you say "from __future__ import with_statement" (not "with") to get the "with" keyword - http://docs.python.org/whatsnew/pep-343.html] |
| 14:57 | <jdandrea> | Philip `: Thx! |
| 14:59 | <Philip`> | [http://www.python.org/dev/peps/pep-0236/ for more details on __future__ in case anyone's overly interested] |
| 14:59 | <zcorpan_> | Hixie: perhaps the <dialog> example in the spec should have end tags... with the current parsing spec the last DD is left open |
| 15:01 | <zcorpan_> | or the parsing spec is wrong? all browsers except opera close the DD with "<foo><dd></foo>X" |
| 15:02 | <Lachy> | zcorpan_, anne and I discussed that issue in here earlier today. check the logs to see what was said about it |
| 15:08 | <zcorpan_> | Lachy: ok, read it. conclusion was that the bug report is invalid given the current spec? |
| 15:09 | <Lachy> | I think so, but I think it might be a bug in the spec |
| 15:09 | <zcorpan_> | yeah |
| 15:10 | zcorpan_ | mails the list |
| 15:11 | <Lachy> | Nooo!!! I've got enough mail to read already ;-) |
| 15:13 | <zcorpan_> | heh |
| 15:13 | <Lachy> | I got through about 100 today, only 202 left on public-html |
| 15:14 | <Lachy> | and 252 on whatwg |
| 15:15 | <jdandrea> | It's a bear. I have to skim them just to get oriented, then go back and really read each one carefully (esp. the long ones). It's not casual reading by any means. |
| 16:06 | <Dashiva> | zcorpan_: Did you forget a message in your latest mail? |
| 16:06 | <zcorpan_> | Dashiva: ? |
| 16:07 | <Dashiva> | The mail I got for "Parsing: should <foo><dd></foo> close the DD?" didn't have any content after headers. Wondering if it's my mail server acting up, or if it was empty |
| 16:08 | <zcorpan_> | http://lists.whatwg.org/pipermail/whatwg-whatwg.org/2007-April/010968.html |
| 16:09 | <Dashiva> | ok, thanks |
| 17:45 | <zcorpan_> | http://xhtml.se/2007/04/20/im-not-arguing-im-just-making-fun-of-them/ |
| 17:53 | <krijnh> | Nice huh? Public logs :) |
| 17:53 | <zcorpan_> | yeah |
| 17:56 | <Philip`> | Public logs with links that show up when people check Referers, too :-) |
| 17:57 | <krijnh> | Indeed |
| 17:57 | <krijnh> | Now if only they'd link back |
| 17:57 | <krijnh> | That'd be good for my pagerank :p |
| 18:00 | <krijnh> | Probably fair to remove rel="nofollow" on my links then |
| 18:02 | <Philip`> | Hmm, SYSTRAN provides interesting translations for parts of the original article - '"X" wild duck "HTML" sitting in a tree', '[...]HTML5 seems [...] to create the the entire wife solving situation [...]', etc |
| 18:03 | <hasather> | Philip`: no, no, that's correct. It actually said that in Swedish ;) |
| 18:05 | <Philip`> | Swedish seems to have been slightly neglected by automatic translation services, compared to things like French and German :-( |
| 18:05 | <krijnh> | Anybody here using Opera 9.2 ? |
| 18:05 | <hasather> | krijnh: yeah |
| 18:06 | <Philip`> | krijnh: Me too |
| 18:06 | <krijnh> | Why isn't "double click to flag/unflag as important" working anymore.. |
| 18:06 | <krijnh> | It did in 9.1 |
| 18:06 | <Philip`> | It works if I double-click on the line, but not if I double click on the textual part of it |
| 18:06 | <krijnh> | It even works in IE7 :( |
| 18:07 | <Philip`> | (because that just selects the text) |
| 18:07 | <krijnh> | Ah |
| 18:07 | <hasather> | krijnh: also, I get a JS error on the log pages |
| 18:07 | <hasather> | might be related |
| 18:07 | <krijnh> | A JS error? |
| 18:07 | <krijnh> | Or :target ? |
| 18:08 | <gsnedders> | hasather: what is the entire wife solving solution, then? :P |
| 18:08 | <hasather> | krijnh: "message: Statement on line 2634: Could not convert undefined or null to object" |
| 18:08 | <krijnh> | Interesting |
| 18:08 | <krijnh> | Line 2634 does not seem to exist though |
| 18:09 | <hasather> | gsnedders: no idea |
| 18:10 | <hasather> | krijnh: ehum, sorry about that. It was in my of my UserJS files :D |
| 18:11 | <krijnh> | If I needed 2634 lines to make this work I'd suck even more than I do already ;) |
| 18:11 | <hasather> | hehe |
| 18:12 | <krijnh> | Ow well, nobody's using that feature anyway |
| 18:14 | <krijnh> | http://www.google.co.uk/search?hl=en&q=stream+read+error+oneview&btnG=Google+Search&meta= :P |
| 18:15 | <hasather> | krijnh: wow, congrats man, top result for that |
| 18:15 | <krijnh> | Tee hee |
| 18:15 | <krijnh> | One hooray for social software |
| 18:16 | <krijnh> | Err, make that user generated content |
| 18:16 | <zcorpan_> | html5 spec down, two more to go |
| 19:17 | <jdandrea> | Reality check Q (read: "I could be reading this incorrectly.") In section 3.2.3.6, is the "(e.g. no space characters)" at odds with major steps 4 and 6? |
| 19:18 | <jdandrea> | Put another way, is "1, 2, 3, 4" a valid list of integers or must it be "1,2,3,4"? |
| 19:19 | <Philip`> | As I read it, that's the difference between author conformance and UA conformance |
| 19:20 | <Philip`> | Authors must write that list as "1,2,3,4", but UAs must parse the invalid string "1, 2, 3, 4" in the same way |
| 19:21 | <jdandrea> | Ahh, I see - author it the former way, but forgive the spaces (or a leading comma) when parsing. |
| 19:25 | <Philip`> | Excepting any mistakes I've made, ",,,01,2.5--,3 , 4!" would be parsed the same way too |
| 19:27 | <jdandrea> | nice |
| 19:27 | jdandrea | is learning a good deal by fine-toothed-combing through this |
| 19:29 | <Philip`> | This is the kind of place where it'd be nice to have an explicit guide for authors about what they must / must not do - otherwise you have to put on a spec-reading hat and read very carefully, which most people won't do |
| 19:30 | <kingryan> | Philip`: we'll have one eventually. |
| 19:31 | <jdandrea> | True. Hmm ... perhaps I'll be able to contribute to that end. |
| 19:31 | <Philip`> | That will be good - I guess the main problem is just finding someone to write it |
| 19:31 | <jdandrea> | :) |
| 19:32 | <Philip`> | jdandrea: That would be good :-) |
| 19:32 | <jdandrea> | I want to work through it a few more times. Heh. Though I will say this spec is more, how shall I say, literate (?) than others I've attempted to read! |
| 19:32 | <jdandrea> | So - an author guide that describes the must and must nots. |
| 19:33 | <jdandrea> | "So You Want To Write Valid X/HTML 5" |
| 19:35 | <Philip`> | I think it would be useful to have something written like a spec or a reference manual (rather than a tutorial), perhaps based on the same structure as the full spec, but only including the bits that are relevant for conforming documents |
| 19:35 | <zcorpan_> | people asked for a view of the spec that only had the information that applied to authors, i think |
| 19:35 | <Philip`> | Could we add a new HTML element for it? <relevance who="authors"> |
| 19:35 | <zcorpan_> | it would be nice but it's hard to do programmatically as it looks now |
| 19:35 | <zcorpan_> | a class would be sufficient |
| 19:35 | <Philip`> | That's no fun :-( |
| 19:35 | <krijnh> | How about that <switch> element? :) |
| 19:36 | <kingryan> | Philip`: such a reference would probably be most useful if it focused on just one of the serializations (just html5, or just xhtml5) |
| 19:36 | <zcorpan_> | kingryan: why? |
| 19:36 | <zcorpan_> | i mean, the difference would be that the #writing section could be skipped in case you write xhtml5 |
| 19:37 | <zcorpan_> | the rest applies to both |
| 19:37 | <kingryan> | zcorpan_: http://lists.whatwg.org/pipermail/whatwg-whatwg.org/2007-April/010957.html |
| 19:38 | <kingryan> | zcorpan_: because it's hard to track down the specific things that authors need |
| 19:38 | <zcorpan_> | ok |
| 19:38 | <kingryan> | b/c part of the contstraints are in the language and part are in the serialization |
| 19:38 | <kingryan> | if you collapse the 2 for a particular serialization, it'd help authors alot |
| 19:39 | <Philip`> | Hmm, are there any preferences on whether to use/promote <!DOCTYPE HTML> or <!DOCTYPE html> or <!doctype html> etc? Whenever I'm writing a page, I always waste time worrying about which one I should use... |
| 19:39 | <zcorpan_> | Philip`: don't worry about it |
| 19:39 | <zcorpan_> | it's case insensitive, just like tags |
| 19:39 | <jdandrea> | I tend to just go with the latter. YMMV. |
| 19:40 | zcorpan_ | doesn't have a caps lock key :) |
| 19:40 | <Philip`> | zcorpan_: But I can't make myself not worry about it :-) |
| 19:40 | <zcorpan_> | Philip`: then <!doctype html> ;) |
| 19:40 | <jdandrea> | lol |
| 19:40 | <jdandrea> | <!Doctype Html> |
| 19:41 | <Philip`> | I don't worry about tags/quotes because I always do <lowercase thing="quoted"> and anything else looks kind of wrong, but I haven't had time to settle on a spelling for the doctype |
| 19:41 | <zcorpan_> | <!dOcTyPe HtMl> |
| 19:42 | <Philip`> | (I worry about single vs double quotes every time I write JavaScript too) |
| 19:42 | <Philip`> | (Too many choices :-( ) |
| 19:42 | <zcorpan_> | in existing documents, if i am to convert the doctype to the html5 doctype, i usually preserve the previous case |
| 19:42 | <zcorpan_> | Philip`: you must like python |
| 19:43 | <krijnh> | Philip`: Me too :| |
| 19:44 | <krijnh> | But perhaps that's because it does matter in php |
| 19:44 | <zcorpan_> | well. it's like /> vs >. don't waste your time thinking about it :) |
| 19:44 | <krijnh> | With /> it saves a space and a slash |
| 19:45 | <krijnh> | In php it saves some CPU |
| 19:45 | <zcorpan_> | yeah, but if you have a template that has />, would you spend time converting it to >? |
| 19:45 | <Dashiva> | Philip`: Always use ' unless the string contains ' and no ". Simple. |
| 19:45 | <krijnh> | zcorpan_: I probably would, yes |
| 19:45 | <zcorpan_> | i used to as well. now i wouldn't |
| 19:46 | <zcorpan_> | you have to learn to ignore things that are irrelevant :) |
| 19:46 | <krijnh> | Especially if it's a template for my cms |
| 19:46 | <krijnh> | Which breaks on xhtml ;p |
| 19:46 | <jdandrea> | "Old habits ..." |
| 19:46 | <zcorpan_> | krijnh: then you have a bug in your cms... :P |
| 19:47 | <krijnh> | Yeah, irrelevant ones |
| 19:47 | <Dashiva> | Old habits die hard, quirks mode lives forever |
| 19:47 | <jdandrea> | :) |
| 19:47 | <Philip`> | /> vs > is easy - the spec already has a "should" conformance criteria for the latter case |
| 19:47 | <zcorpan_> | Philip`: only for atheists :) |
| 19:48 | <Dashiva> | I prefer 'godless heathen' myself |
| 20:26 | <zcorpan_> | mmm... tinned tuna |
| 20:30 | zcorpan_ | has to get used to eating 6 times a day again |
| 20:40 | <krijnh> | zcorpan_: Working that hard? |
| 20:41 | <zcorpan_> | at the gym, yes |
| 20:41 | <krijnh> | Ah, right |
| 20:41 | <Hixie> | is spartanicus here? |
| 20:48 | <zcorpan_> | wow, firefox does the same wrt <iframe id/name> as it does with <map id/name> |
| 20:55 | <zcorpan_> | @_@ ... opera's behaviour with <a target> freaked me out there |
| 20:57 | <zcorpan_> | if i had <object name=foo data=data:text/plain,x></object> then <a href=foo.txt target=foo>test</a> opened in the object (no surprise). then i removed the data attribute and suddenly the link opened in another object in another tab, that was opened independently from the one where the link was found! |
| 20:58 | <zcorpan_> | if i closed that tab, the link would open in a new window |
| 20:58 | <zcorpan_> | (or well, tab) |
| 21:01 | <zcorpan_> | nevertheless <a target> only works in <object> in opera, it seems, so a more pragmatic way to interop would be to say that <a target> doesn't work with <object> at all (like in ie, firefox) |
| 21:23 | <zcorpan_> | hm. <map> is an inline element in html4 |
| 21:24 | <zcorpan_> | block in html5 |
| 21:37 | zcorpan_ | uploads test cases without pass conditions... i.e. demos |
| 21:40 | <zcorpan> | ...or not, can't connect with the server :| |
| 21:45 | <zcorpan> | conclusion of my tests anyway is that <iframe name> and <map name> should be conforming (name perhaps required), but UAs must support both id and name for both. <object> shouldn't work with <a target> at all |
| 21:45 | <zcorpan> | i'll send this to the list as soon as i can upload my tests |
| 21:52 | met_ | examined source of http://www.microsoft.com/silverlight/asp/default.aspx it's html but inside it is <?xml ?> declaration, awful |
| 21:56 | <krijnh> | I like how they presented IE7 with a nice stylesheet with hacks |
| 21:57 | <krijnh> | IE6 that is |
| 22:05 | <Philip`> | I like how the "cross-browser, cross-platform plug-in" doesn't support either the browser or the platform that I'm currently using |
| 22:14 | <Philip`> | (Actually, it looks like it wouldn't support either even if I rebooted into Windows (2000), though I'm unsure whether that's support in the "we will provide official support" sense or the "we will not refuse to install" sense (though my experience is that the latter is most often the case for Microsoft software...)) |
| 22:42 | <zcorpan> | test |
| 22:58 | <ajnewbold> | it worked |