| 00:07 | <aboodman> | sicking: what's the question? |
| 00:08 | <sicking> | sorry, something came up in another channel, bbiab |
| 00:09 | <sicking> | so the issue is how to specify JS versions |
| 00:10 | <sicking> | something that is important in order for 'yeild' to work for example |
| 00:10 | <sicking> | (ff only for now, hopefully will be implemented cross browser) |
| 00:11 | <olliej> | sicking: you could try { eval("function(){var yield = 1;}") } catch(e) { /* no yield */} or something |
| 00:11 | <olliej> | surely? |
| 00:11 | <olliej> | i''m assuming "yield" is considered a keyword |
| 00:11 | <sicking> | olliej, no, i mean that we as an implementation has to put the JS parser in a special mode when parsing |
| 00:11 | <sicking> | olliej, it's not unless the parser is put in such a mode |
| 00:12 | <olliej> | ah |
| 00:12 | <sicking> | olliej, since introducing keywords breaks stuff |
| 00:12 | <olliej> | true |
| 00:12 | <othermaciej> | I think there should be an in-script way of specifying you want new keywords |
| 00:12 | <zcorpan_> | sicking: can't importScript assume the latest and greatest? |
| 00:12 | <othermaciej> | an external flag is not robust |
| 00:12 | <zcorpan_> | sicking: surely there's no legacy to worry about |
| 00:13 | <Philip`> | You just need to grep the web for all identifiers used in all JS scripts, and then choose keywords which don't conflict with any identifier usage |
| 00:13 | <sicking> | zcorpan_, that means that a page loaded in safari will work, but not in FF |
| 00:13 | <sicking> | zcorpan_, since the 'yeild' in FF will be parsed as a keyword, but in safari/opera/chrome not |
| 00:13 | <aboodman> | sicking: does it even make sense to load a js 1.5 script into a, say, js 1.4 environment? |
| 00:13 | <olliej> | sicking: you forgot ie |
| 00:13 | <olliej> | sicking: ;) |
| 00:13 | <zcorpan_> | sicking: you don't expect authors who write worker scripts to test in firefox? |
| 00:13 | <sicking> | othermaciej, external meaning the server side content-type for example |
| 00:14 | <othermaciej> | sicking: yeah, that seems bogus to me |
| 00:14 | <sicking> | zcorpan_, nope |
| 00:14 | <othermaciej> | sicking: it means if I download the script, it won't work any more |
| 00:14 | <othermaciej> | since there's nothing to provide the Content-Type |
| 00:14 | <othermaciej> | so now I have to remember to set a flag wherever I embed it |
| 00:14 | <sicking> | aboodman, it seems to work in HTML today. You specify version per <script> element |
| 00:14 | <othermaciej> | versioning should be in the script itself, like the proposed "use strict" |
| 00:15 | <aboodman> | sicking: ok, i never use that feature, so i'm unfamiliar. sorry. |
| 00:15 | <othermaciej> | Safari completely ignores the version on the <script> element |
| 00:15 | <othermaciej> | it's basically a Firefox-only feature |
| 00:15 | <aboodman> | sicking: maybe there could be an optional extra param in the ff version |
| 00:15 | <sicking> | othermaciej, how would you deal with yield? 'use yield'? |
| 00:15 | <Philip`> | sicking: Most people specify versions like 1.1 or 1.3, and all browsers treat that as exactly the same as 1.5, so it doesn't seem like a particularly successful feature |
| 00:15 | <sicking> | aboodman, problem is that the function takes a list of uris |
| 00:16 | <othermaciej> | sicking: I think whatever language version introduces it should have a "use es4" or "use harmony" or whatever to request new keywords |
| 00:16 | <sicking> | aboodman, i suppose we could check if the last one looks like a version number, but that seems like a hack |
| 00:16 | <sicking> | othermaciej, makes sense to me. Have you guys talked about this? |
| 00:16 | <aboodman> | or you could support something like: |
| 00:16 | <aboodman> | importScripts([[foo.js, 1.5], [bar.js, 1.3]]) |
| 00:17 | <othermaciej> | sicking: I have talked some about how the old-world ES4 approach to versioning was lame, but I am not sure what anyone else thinks |
| 00:17 | <othermaciej> | I'll bring it up on es-discuss |
| 00:17 | <aboodman> | horrific i know |
| 00:17 | <zcorpan_> | why are you concerned about js keywords but not new dom features? |
| 00:17 | <sicking> | Philip`, the differences between 1.1 and 1.3 doesn't generally matter. 1.5 to 1.8 is different |
| 00:17 | <zcorpan_> | i mean why should authors have to opt in to get new js keywords but not to get new dom features |
| 00:18 | <sicking> | zcorpan_, because introducing new DOM features generally doesn't break existing content |
| 00:18 | <Philip`> | zcorpan_: Because new JS keywords are likely to make perfectly valid not-uncommon scripts written in the old language version fail to compile in the new version, whereas you never get the same problem with new DOM features |
| 00:18 | <sicking> | zcorpan_, introducing new JS keywords does |
| 00:19 | <sicking> | i guess for now we can add something like 'mozilla.importScriptVersion' and let people set that before calling importScripts |
| 00:22 | <Philip`> | If people are setting the script version, presumably they are using features of the new language version, so their scripts won't work in old language versions, so their scripts won't work in any other browsers anyway, so the API doesn't need to be compatible with other browsers |
| 00:22 | <Philip`> | so it could just be a function like mozImportScriptsWithVersion(version, uris...) instead of calling importScripts |
| 00:23 | <Hixie> | sicking: use the same technique as is used for javascript: URLs, or onfoo="" event handlers, or scripts sent over FTP |
| 00:24 | <Hixie> | sicking: which is to say, fix JS so that it is forwards and backwards compatible like HTML and CSS and you don't need a specific out-of-band indicator |
| 00:24 | <erlehmann> | js 5, yay \o/ |
| 00:55 | <sicking> | Hixie, true. I don't think we have a way to use other version in onfoo attributes right now |
| 00:55 | <sicking> | Philip`, exactly, that's why i suggested the mozilla.importScriptVersion syntax. Your syntax would also work |
| 00:59 | Hixie | has been telling the JS group this for some time now |
| 00:59 | <Hixie> | but they keep pushing back |
| 00:59 | <Hixie> | which is sad, since lack of versioning being the way to go has been proven over and over again to be the only sane way to design web-resistent languages |
| 01:00 | <Hixie> | versions don't work when you have unsynchronised widely-buggy implementations and widely varied content providers that work together without collaborating |
| 01:02 | <sicking> | Hixie, at the same time, HTML5 has <DOCTYPE HTML> |
| 01:02 | <sicking> | Hixie, i mostly agree though |
| 01:02 | <Hixie> | <!DOCTYPE HTML> is not versioning. it's explicitly no versioning, in fact. |
| 01:03 | <Hixie> | i would have removed it altogether (we did in XHTML5) but then we wouldn't be triggering standards mode |
| 01:03 | <sicking> | Hixie, exactly, so the web needs at least two versions |
| 01:03 | <Hixie> | quirks mode was a huge mistake |
| 01:03 | <Hixie> | we should never have done it |
| 01:04 | <Hixie> | we only "need" four versions (xml, quirks, limited quirks, no quirks) because of historical inertia |
| 01:05 | <Hixie> | we should make as sure as possible that we never add another version again |
| 01:05 | <sicking> | agreed, though there is a 'as possible' there |
| 01:05 | <Hixie> | it's possible with JS |
| 01:06 | <Hixie> | e.g. using special syntax for new keywords, or whatever |
| 01:06 | <Hixie> | or instrumenting browsers to find out what keywords are safe to add, the same way we scanned the web to determine safety for new html5 element names |
| 01:07 | <Hixie> | or importing new symbols using a page-wide "opt-in" syntax |
| 01:13 | <BenMillard> | webben, I disagree with that markup for tag clouds: http://krijnhoetmer.nl/irc-logs/whatwg/20081007#l-372 |
| 01:14 | <BenMillard> | webben, it makes the content desperately verbose and sighted users don't have a convenient way of getting the actual frequencies |
| 01:22 | <BenMillard> | webben, I'd put the values in the title attribute and it seems "ampz" had the same idea in the final comment to that 24 ways article: http://24ways.org/2006/marking-up-a-tag-cloud#c001139 |
| 01:23 | <BenMillard> | webben, the 24 ways markup also puts the link at the end of the item, so it's painful to use linearly |
| 01:26 | <Dashiva> | Geez, what is it with people thinking that just because something is in some order, it has to be an ordered list |
| 01:26 | <Hixie> | the spec is clear that that is not what <ol> means. :-) |
| 01:27 | <Dashiva> | http://www.whatwg.org/specs/web-apps/current-work/multipage/ |
| 01:27 | <Dashiva> | Where's my spec? |
| 01:27 | <Hixie> | it's there |
| 01:27 | <Dashiva> | Oh, there it is. I must've hit the page just before the index.html was generated |
| 01:27 | <Hixie> | looks like |
| 01:33 | <BenMillard> | Dashiva, "I put my list items in order, therefore it's an ordered list." |
| 01:35 | <Dashiva> | Sounds like a job for intelligent sort |
| 01:36 | <Philip`> | Dashiva: I'd blame Hixie for that - the actual generation (currently performed by me) happens on a different server and it just uploads a .tar.something with the output, so it's Hixie's fault if that doesn't get uncompressed and linked into place atomically :-) |
| 01:36 | <Dashiva> | Hmm |
| 01:37 | <Dashiva> | You could uncompress on the side, but is it possible to redirect a symlink atomically? |
| 01:38 | <Philip`> | How could it possibly be impossible? |
| 01:38 | <Philip`> | Have directories called "001" and "002", symlink multipage to 001, uncompress into 002, replace the symlink, delete directory 001 |
| 01:38 | <Philip`> | then repeat next time with 002 and 003 |
| 01:39 | <Dashiva> | But replacing the symlink might mean delete + create unatomically |
| 01:39 | <Philip`> | The symlink can't really be doing anything other than pointing at either one of the directories, so it's got to be pretty much atomic |
| 01:40 | <Philip`> | "Replace" is just a renaming operation, and I'd assume all sane filesystems do atomic renaming |
| 01:40 | <Dashiva> | That's what you think |
| 01:40 | <Philip`> | Indeed it is :-) |
| 01:42 | <Philip`> | But anyway, delete+create would still be, uh, molecular even if it's not atomic, and you'd never manage to make an HTTP request in that tiny time period, so it's much better than deleting everything and then letting bunzip2 crunch some megabytes for a few seconds |
| 01:43 | <Dashiva> | But if we leave it at that, Hixie won't have an excuse for not making it work |
| 01:47 | <Hixie> | I'm getting there. Only <textarea>, <output>, common attributes, validity, and form submission formats to go. |
| 01:49 | <Dashiva> | Poor type=email-list, never to see the light of day |
| 01:49 | <Hixie> | oh well once i've done all this i have 500+ e-mails to reply to with new feature requests |
| 01:49 | <Hixie> | i'm just talking about the wf2->html5 integration |
| 01:50 | Hixie | decides to drop <textarea pattern=""> |
| 01:56 | <blooberry> | hixie: why so? |
| 02:14 | <BenMillard> | Hixie, now 1,002 entries in my 2008 collection as the "To Do" entries often have several interesting uses of markup in them |
| 02:14 | <BenMillard> | Hixie, maybe we can fit a few sessions to run through it? We can do both planned and impromptu meetings, depending how it goes. |
| 02:14 | <Hixie> | blooberry: what's the use case? |
| 02:14 | <Hixie> | BenMillard: sure |
| 02:15 | <Hixie> | BenMillard: not too long, but i can do an hour or so |
| 02:15 | <BenMillard> | Hixie, my feeling exactly :) |
| 02:15 | <BenMillard> | I don't want to spend all day looking at collected web pages when I'm in the south of France |
| 03:09 | <blooberry> | hixie: Sorry, was away. Bad word filtering in blog or forum system? |
| 06:12 | <Hixie> | gah |
| 06:12 | <Hixie> | i missed "textarea" when fixing the parser last time |
| 08:11 | hsivonen_ | sees the financial crisis comes to the damowmow portal |
| 08:23 | <Hixie> | man, why do people watch that page all the time :-) |
| 08:24 | <olliej> | Hixie: ? |
| 08:24 | <olliej> | Hixie: damowmow has a portal? whoah |
| 08:24 | <Hixie> | http://☺.damowmow.com/ |
| 08:34 | <hsivonen_> | krijnh: your linkifying regexp doesn't work with IRIs |
| 08:39 | <jruderman> | or http://www.damowmow.com/portal/ if you want an address you can type |
| 08:40 | <Hixie> | click on the <h1> to go to the idn url |
| 08:40 | <Hixie> | but seriously? you people use that page? |
| 08:40 | <Hixie> | what's with that! :-P |
| 08:41 | <hsivonen_> | Hixie: it has useful links and interesting charts |
| 08:45 | <Hixie> | fair enough :-) |
| 08:45 | <Hixie> | i encourage you to make your own though, it'll be more useful :-) |
| 09:00 | <hsivonen_> | Hixie: I meant the electoral vote charts--not your portfolio. |
| 09:11 | <annevk2> | krijnh, you fail on http://☺.damowmow.com/ |
| 09:12 | <annevk2> | oops, hsivonen_ said that already |
| 09:13 | <Philip`> | krijnh: Thank you for not linkifying http://☺.damowmow.com/, since we don't want to encourage the use of untypeable domain names |
| 09:13 | <annevk2> | Hixie, new Option() takes an optional fourth argument as well |
| 09:13 | <annevk2> | Hixie, it sets the selectedness |
| 09:15 | <hsivonen_> | can async scripts or defer scripts do document.write? |
| 09:18 | <hsivonen_> | so the first level of document.write is atomic if it doesn't write "<script>" |
| 09:18 | <hsivonen_> | if it does write "<script>", is it supposed to return early and leave the rest of its data in the inputstream? |
| 09:35 | <hsivonen_> | Hixie: can "script that will execute as soon as possible" execute when parsing document.written data? |
| 09:38 | <hsivonen_> | "If the tree construction stage is being called reentrantly, say from a call to document.write():" Can it be called re-entrantly except from document.write? |
| 10:45 | <hendry> | has anyone studied how mail should be marked up on the Web? http://natalian.org/archives/2008/10/07/marking-up-mail/ |
| 14:09 | <Philip`> | "IE8 also supports an event named storagecommit that is not part of HTML5 which tells you when the data has been written to an XML backend format IE8 uses." - hmm, XML? I wonder if you can store values that make that XML ill-formed... |
| 14:28 | <jcranmer> | see if storing '; DROP DATABASE; &-) <!-- fails... |
| 14:29 | annevk2 | was not planning on booting back into Windows anytime soon, but would be interested in results of such tests :) |
| 14:31 | Philip` | wishes he had a VM with IE8, that worked in Linux and didn't cause kernel panics, since that'd make it much easier to test |
| 15:31 | <erlehmann> | how can i style an element that it doesn't disturb the flow of other block elements ? |
| 15:36 | <annevk2> | that sounds confused :) |
| 15:36 | <erlehmann> | okay, i want a layout like |
| 15:36 | <erlehmann> | xxxxxxxx yy |
| 15:36 | <erlehmann> | zzzzzzzz yy |
| 15:36 | <erlehmann> | zzzzzzzz yy |
| 15:36 | <erlehmann> | zzzzzzzz yy |
| 15:37 | <erlehmann> | and in the source, elements are in order <x> <y> <z> |
| 15:37 | <erlehmann> | y has fixed width and floats left |
| 15:37 | <erlehmann> | err, right |
| 15:37 | <annevk2> | use position:absolute for y instead? |
| 15:37 | <erlehmann> | x and z are dynamic |
| 15:39 | <erlehmann> | annevk2: how could that help if i have several <y> class elements ? |
| 15:40 | <annevk2> | you didn't say that |
| 15:41 | <erlehmann> | hihi, but i have |
| 15:41 | <erlehmann> | like <x> <y> <y> <y> <y> <z> |
| 15:41 | <annevk2> | well, group them? |
| 15:41 | <erlehmann> | and i dont want to make a semantic mess of it ? |
| 15:41 | <erlehmann> | group ? |
| 15:41 | <erlehmann> | like, enclosing it in a div ? |
| 15:42 | <erlehmann> | that would mako the layout way too inflexible |
| 15:42 | <zcorpan> | how do you define CharacterData.data? |
| 15:43 | <zcorpan> | "The data attribute must, on getting, return the data of the node, and on setting, must change the node's data to the new value. " |
| 15:44 | <annevk2> | maybe you should have a "data-concept" just like Hixie has a concept for value |
| 15:44 | <annevk2> | erlehmann, then I don't know, good luck! :) |
| 15:46 | <zcorpan> | annevk2: and how do i define that? |
| 15:55 | <annevk2> | zcorpan, each "1" object has a "2". member "3" of "1" modifies "2" |
| 15:55 | <annevk2> | or something in that direction |
| 15:56 | <annevk2> | I notice Hixie hasn't actually defined "2" yet :) |
| 15:56 | <annevk2> | (but it's referenced all over the place) |
| 16:49 | <zcorpan> | when is a node read-only as defined in dom3core? |
| 16:49 | <zcorpan> | http://www.w3.org/TR/DOM-Level-3-Core/glossary.html#dt-readonly-node |
| 16:53 | <annevk2> | "As for Entity nodes, EntityReference nodes and all their descendants are readonly." |
| 16:53 | <annevk2> | couldn't find anything else |
| 16:54 | <annevk2> | (though potentially there are other specs that allow you to obtain a readonly DOM node/fragment or some such, but I can't think of any) |
| 16:57 | <zcorpan> | ok |
| 16:58 | <annevk2> | you didn't fix textContent btw, it has setting and getting reversed |
| 16:59 | <zcorpan> | fixed |
| 17:25 | <krijnh> | hsivonen_, annevk2, Philip`: fixed |
| 17:26 | <annevk2> | ta |
| 17:26 | <krijnh> | (only that case btw :) |
| 17:27 | <hsivonen_> | krijnh: thanks |
| 17:41 | <Philip`> | krijnh: I didn't want you to fix it :-( |
| 17:41 | <gsnedders> | Fix what? |
| 17:41 | <Philip`> | krijnh: What he just fixed |
| 17:41 | <Philip`> | Uh |
| 17:42 | <Philip`> | s/krijnh/gsnedders/ |
| 17:42 | <gsnedders> | Which was? |
| 17:42 | <gsnedders> | linking http://☺.damowmow.com/? |
| 17:44 | <Philip`> | gsnedders: I think so |
| 17:47 | <zcorpan> | Philip`: it's easy to type ☺ |
| 17:48 | <gsnedders> | zcorpan: You were saying anolis cut off <http://simon.html5.org/specs/web-dom-core>;? |
| 17:48 | <gsnedders> | zcorpan: You find what caused that? You got the source file? |
| 17:48 | <zcorpan> | gsnedders: yeah, it did, twice, at different places |
| 17:49 | <zcorpan> | http://simon.html5.org/specs/web-dom-core.src |
| 17:50 | <gsnedders> | zcorpan: But it works now? You know why? |
| 17:50 | <zcorpan> | once after the <!-- TypeInfo dropped --> comment and once before the <!-- Notation dropped --> comment, iirc |
| 17:50 | <annevk2> | zcorpan, shouldn't isEqualNode consider many other attributes as well? |
| 17:51 | <gsnedders> | zcorpan: Odd. |
| 17:51 | <zcorpan> | annevk2: dunno |
| 17:52 | <annevk2> | might be an idea to define the various properties of a node and an element, etc. as definitions first and then "implement" the members of the interfaces using those terms |
| 17:53 | <zcorpan> | annevk2: can you give an example of what you mean? |
| 17:56 | <annevk2> | e.g., define an element as having a namespace, name, prefix, list of attributes, list of children |
| 17:57 | <annevk2> | then for Node.namespaceURI you say it returns null unless it is an element or attr object in which case it returns namespace-concept |
| 17:57 | <zcorpan> | ok |
| 17:58 | <annevk2> | but maybe that's not needed, not sure |
| 18:00 | <zcorpan> | gotta go now anyway |
| 18:35 | <BenMillard> | Hendry, the markup for mail archives is something I've thought about but not looked into closely. |
| 18:38 | <BenMillard> | the way Google turns them into somewhat natural web pages seems general approach, although the specific markup they use could be improved a lot. Such as using heading elements around heading-like text: http://groups.google.com/group/Google-Maps-API/browse_thread/thread/c81a00da42bd64d9 |
| 18:39 | <BenMillard> | s/seems general approach/seems like the best general approach/ |
| 18:55 | <ehird> | Is profile="" still removed from <head>? |
| 18:59 | <smedero> | ehird: it is not present in the current draft, no: http://www.whatwg.org/specs/web-apps/current-work/multipage/semantics.html#the-head-element |
| 19:00 | <ehird> | smedero: :| |
| 19:07 | <smedero> | Microformat folks recommend using <a> or <link> + @rel=profile with HTML 5: http://microformats.org/wiki/profile-uris#.3Chead_profile.3E |
| 19:07 | <BenMillard> | Hendry, here's a fascinating UI for mailng lists: http://zesty.ca/zest/out/ |
| 19:10 | <BenMillard> | Hendry, putting the most important phrase from a message in the left panel turns skimming a thread into something really usable and convenient. |
| 19:10 | BenMillard | bookmarks it for future reference. |
| 19:13 | <hallvors> | In the HTML5 draft we're planning to re-use "drop" events for pasting data. Has anyone investigated or tried to standardise what should happen if you paste formatted content that contains images into an online rich text editor? |
| 19:13 | <zcorpan> | BenMillard: here's one for you: http://krijnhoetmer.nl/irc-logs/whatwg/20081008#l-18 |
| 19:14 | <BenMillard> | zcorpan, I don't understand that :P |
| 19:14 | <BenMillard> | was it plain text? |
| 19:15 | <zcorpan> | BenMillard: a checkbox with the label "I accept the terms*" where "terms" is a link and "*" indicates required |
| 19:15 | <BenMillard> | zcorpan, ah |
| 19:16 | <BenMillard> | zcorpan, maybe something like this: <input type="checkbox" name id> <label for>I accept the <a href>terms</a> (required)</label> |
| 19:17 | <BenMillard> | (with sensible values for the attributes) |
| 19:17 | <zcorpan> | BenMillard: <label><a href> is not allowed in html5 and clicking the link would check the checkbox |
| 19:18 | <BenMillard> | zcorpan, that isn't how I expect it to work as an author...hmm |
| 19:18 | <zcorpan> | <a href> doesn't stop the event from bubbling up |
| 19:19 | <BenMillard> | zcorpan, could you prevent that from script (or capture it and make it a no-op?) |
| 19:19 | <zcorpan> | yeah you could |
| 19:20 | <BenMillard> | ok, well as an author and a user I expect clicking a link to only click the link :) |
| 19:20 | <BenMillard> | if that's not what happens natively, scripting that to be the behaviour seems OK |
| 19:21 | <zcorpan> | wonder if it's possible to make the default action for <label> have knowledge about whether a descendant link was clicked |
| 19:22 | <BenMillard> | zcorpan, there might be a more general rule there for making nested interactivity work |
| 19:22 | <BenMillard> | I imagine there are sites which rely on what's currently implemented in this area, though? |
| 19:23 | <zcorpan> | changing the default action of <label> to do nothing if a descendant interactive element was clicked seems like a safe change |
| 19:24 | <zcorpan> | whilch probably improves the usability of pages that do the above |
| 19:26 | <BenMillard> | zcorpan, yeah that sounds good, even when implicit label association is being used |
| 19:27 | <BenMillard> | zcorpan, <label><input type=checkbox ...> I accept the <a href>terms</a> (Required)</label> |
| 19:28 | <BenMillard> | clicking <a hreF> shouldn't click the <label>. Clicking <input> shouldn't click <label> since that would only click the <input> again. |
| 19:28 | <BenMillard> | cool :) |
| 19:29 | <zcorpan> | hmm that's a good point |
| 19:31 | BenMillard | feels good when zcorpan agrees with him. |
| 19:31 | <BenMillard> | 19:30 and I haven't eaten lunch...I need to offset my sleep cycle before the TPAC |
| 19:32 | <zcorpan> | actually this is how it works in opera |
| 19:38 | <zcorpan> | and webkit |
| 19:39 | <zcorpan> | and ie |
| 19:40 | <zcorpan> | well then i don't see a reason for <label> to not accept interactive descendants in html5 |
| 19:47 | <zcorpan> | and actually when reading the spec it seems it allows <label><a href> |
| 19:50 | <zcorpan> | i found a bug in v.nu |
| 19:51 | <BenMillard> | zcorpan, because the content model permits "phrasing content" and that includes <a href>? http://www.whatwg.org/specs/web-apps/current-work/multipage/forms.html#the-label-element |
| 19:51 | <BenMillard> | (and <a href> isn't a "labelable form-associated element" afaict) |
| 19:52 | <BenMillard> | it talks about "the platform's label behaviour" being matched, though |
| 19:52 | <zcorpan> | BenMillard: yeah. i thought there was an implicit rule that interactive content can't contain interactive descendants, but that doesn't seem to be the case |
| 19:53 | <BenMillard> | zcorpan, I thought I read that in HTML5 as well, in a different section... |
| 19:55 | <zcorpan> | the content model for <a> is Transparent, but there must be no interactive content descendant. |
| 19:55 | <BenMillard> | zcorpan, the definition for <a> says it mustn't have interactive descendents |
| 19:55 | <BenMillard> | ah, we spotted that at the same time :) |
| 19:55 | <Philip`> | localStorage["foo"] = "\ud800" |
| 19:56 | <Philip`> | IE stores that in an XML file saying <root><item name="foo" value=" /></root> (sic) |
| 19:58 | <Philip`> | If you have multiple items, the XML serialiser just stops after the first \ud800 and outputs the ' /></root>' immediately |
| 19:58 | <BenMillard> | zcorpan, HTML4 also allows <label><a href> since <a> is part of %special; which is part of %inline; here: http://www.w3.org/TR/html4/interact/forms.html#edef-LABEL |
| 19:59 | <Philip`> | (Same for \ufffe etc) |
| 19:59 | <BenMillard> | and an exception isn't made for it in the prose, that I've seen |
| 20:00 | <zcorpan> | html4 also allows <a href><label> |
| 20:01 | <Philip`> | (annevk2: ^ ) |
| 20:01 | <BenMillard> | zcorpan, I wonder if there's a use-case for that... |
| 20:02 | <BenMillard> | I can't think of one off-hand and my 2008 collection hasn't found <a href><label> |
| 20:07 | <BenMillard> | zcorpan, using part or all of a link's text to label a form control seems very confusing to use |
| 20:07 | <BenMillard> | zcorpan, things like <a href>Google <label for=q>Search</label></a>: <input name=q id=q> could be written <label for=q><a href>Google</a> Search:</label> <input name=q id=q> or <label><a href>Google</a> Search: <input name=q></label> and seem more usable for it |
| 20:09 | <zcorpan> | BenMillard: i agree |
| 20:11 | <zcorpan> | hmm html5 doesn't allow <label for=a><input id=a></label> |
| 20:11 | <Philip`> | annevk2: I appear to be unable to comment on your blog |
| 20:11 | <Philip`> | annevk2: It keeps saying "What did you do? You tried to pass the comment system with an invalid comment! Note that Opera has a submission bug; it converts entities on submit...", even when I use Firefox |
| 20:12 | <BenMillard> | Philip`, you have to write <p> elements and everything last time I used it. |
| 20:13 | <Philip`> | BenMillard: I did, and the preview thing was happy to accept it |
| 20:13 | <BenMillard> | Philip`, in that case I'm no help to you. :) |
| 20:14 | <Philip`> | BenMillard: I'll just blame anne, then :-) |
| 20:16 | <Philip`> | Clearly XML is hard |
| 20:16 | <zcorpan> | Philip`: you need to break your habit to include invalid byte streams |
| 20:16 | <Philip`> | zcorpan: I wasn't including invalid byte streams at all! |
| 20:17 | <Philip`> | zcorpan: at least, not in the attempted blog comment itself |
| 20:31 | <Philip`> | Argh |
| 20:31 | <Philip`> | IE8's XSS 'protection' breaks the Live DOM Viewer |
| 20:32 | <Philip`> | If you try a permalink to anything containing <script>, then it complains about XSS and replaces the page's own <script> tags with <sc#ipt> |
| 20:33 | <Philip`> | Admittedly it's true that the Live DOM Viewer is one giant XSS vulnerability, but that's the whole point of it |
| 20:35 | <csarven> | http://unicodesnowmanforyou.com/ |
| 20:36 | <Philip`> | IE8 seems to calculate remainingSpace based on the sum of all key and value string lengths, measured in UTF-16 code units |
| 20:48 | <Philip`> | which is pretty silly since it takes 61+len(key)+len(value) to store each item, so you can store maybe three hundred thousand items with zero-length value and minimal-length key, taking up about 20MB of file space and maybe 2MB of post-compression disk space, which is much more than the notional 5e5-byte quota |
| 20:48 | <gsnedders> | Today's challenge: Describe me in one to three sentences. |
| 20:50 | <Philip`> | gsnedders: Are expletives allowed? |
| 20:50 | <gsnedders> | Philip`: yes |
| 20:51 | <hallvors> | hm, so there is actually a use case for the "do not use XSS protection on this page" HTTP header? ;-) |
| 20:51 | <hallvors> | Ian has to add one for the live DOM viewer.. |
| 20:53 | <Philip`> | hallvors: Ooh, good idea |
| 20:53 | <gsnedders> | Philip`: I wonder what you're planning now :) |
| 20:54 | <Philip`> | Hixie: Please add X-XSS-Protection: 0 to http://software.hixie.ch/utilities/js/live-dom-viewer/ so that it works correctly in broken browsers like IE8 |
| 21:18 | <Hixie> | the lastweekinhtml5 blog's quality went up a little bit in the last post |
| 21:18 | <Hixie> | if it goes up in quality much more, it'll be good enough for the blog |
| 21:25 | <Hixie> | you know with all the attention on this channel, you'd have thought we'd have been quoted on bash.org by now |
| 21:32 | <smedero> | http://bash.org/?913 |
| 21:32 | <hsivonen_> | I guess those who'd quote use http://quotes.burntelectrons.org/search?query=tag%3A%23whatwg instead |
| 21:48 | Philip` | would be surprised if anything said on this channel had sufficiently widespread appeal and quality to be suitable for bash.org |
| 21:49 | Hixie | falls down a hole and gets trapped under quotes.burntelectrons.org |
| 21:49 | <Hixie> | Philip`: have you _read_ bash.org quotes?! |
| 21:49 | <Philip`> | Hixie: Yes :-) |
| 21:49 | <Philip`> | (Far too many :-/ ) |
| 21:51 | <Hixie> | and you think we'll have an appeal and quality problem? :-) |
| 22:05 | <gsnedders> | Hixie: Yes. |
| 22:05 | <gsnedders> | We're a horrible bunch. |
| 22:06 | <gsnedders> | I mean, <http://lastweekinhtml5.blogspot.com/>; proves it. |
| 22:55 | <Dashiva> | Is being listed on that lastweek blog a badge of honor yet? |
| 23:07 | <virtuelv> | wildly off topic, but this is brilliant: http://www.webmonkey.com/blog/YouTube_Commenters_Hear_Their_Own_Gibberish |
| 23:07 | <virtuelv> | every site should have that |
| 23:23 | <Hixie> | Dashiva: neither really |
| 23:41 | <fakeolliej> | gsnedders: lame-o -- someone already linked to the penny arcade comic i was thinking of linking to lastweekinhtml |
| 23:42 | <olliej> | oh well, ours not to question the mind of a troll |
| 23:42 | <olliej> | or a trolls troll as the case may be |
| 23:44 | <Hixie> | it's kind of weird that the guy is doing it anonymously, it seems like a perfectly legitimate thing to actually do |
| 23:45 | <olliej> | Hixie: yes, it would be, however his behaviour indicates that he's a troll, and trolls don't want to actually be answerable to anyone |
| 23:46 | <Dashiva> | Maybe he's a performance artist |
| 23:46 | <Hixie> | olliej: if he's trying to troll, he's not doing it right :-) |
| 23:46 | <olliej> | Hixie: hehe |
| 23:46 | <olliej> | Dashiva: heheheh :D |
| 23:46 | <Hixie> | olliej: trolls aren't supposed to be useful and fun :-) |
| 23:47 | <olliej> | Hixie: if he felt he were legitimately critiquing the whatwg (reasonable) he would not have felt the need to be anonymous |
| 23:49 | <othermaciej> | I think I know know who the lastweek troll is |
| 23:49 | <olliej> | oh? |
| 23:49 | <olliej> | i was suspecting one of the people i passed the url to |
| 23:50 | <othermaciej> | just a guess |
| 23:50 | <olliej> | (who is also an avid penny arcade fan :D ) |
| 23:55 | <othermaciej> | no, I mean I think I know who is the author of the last week in html5 blog |
| 23:55 | <Dashiva> | Is it Fake Steve Jobs? |
| 23:55 | <othermaciej> | not the comment that Mr Last Week mistakenly attributed to me |
| 23:55 | Dashiva | wonders if there will be a Fake Ian Hickson blog one day |
| 23:55 | <olliej> | hehehe |