| 00:08 | gsnedders | realizes the last time he played on a PS2 was in Mandelieu |
| 00:09 | <Dashiva> | That sounds more like a place |
| 02:37 | Hixie | adds a new feature to the live dom viewer |
| 03:28 | <Hixie> | ok i guess next up is html5/selectors interaction |
| 05:46 | <Hixie> | zcorpan: IE supports 'transparent'?! |
| 05:48 | <zcorpan> | Hixie: yep |
| 05:48 | <Hixie> | wow |
| 05:48 | <Hixie> | i put your mail into my pile, i'll get to it probably later this month or next month, but i haven't forgotten it |
| 05:48 | <Hixie> | i'm about to start writing the section that deals with selectors |
| 05:48 | <Hixie> | i wonder where to put it |
| 05:49 | <zcorpan> | put it in a new section before rendering? |
| 05:50 | <Hixie> | that's what i was thinking, but it seems kinda weird to have such a small section be at such a high level |
| 05:50 | <Hixie> | i'm thinking a subsection at the end of 4 The elements of HTML |
| 05:51 | <zcorpan> | might be good, that'll probably get more attention by authors, too |
| 06:30 | <zcorpan> | Hixie: "hr[align=left] { margin-left: 0; margin-right: auto; }" - why margin-left: 0? |
| 06:36 | <Hixie> | zcorpan: consistency with the center case |
| 06:36 | <Hixie> | doesn't really matter, 0 is the default anyway |
| 06:39 | <zcorpan> | it would override user styles |
| 06:40 | <Hixie> | oh, good point, that's another reason |
| 06:41 | <zcorpan> | why would you want to override user styles? |
| 06:41 | <Hixie> | author rules override user rules |
| 06:41 | <Hixie> | that's how css works |
| 06:41 | <Hixie> | if the author says align=left, then you align left |
| 06:42 | <Hixie> | if we didn't say margin-left: 0, you might end up centering |
| 06:43 | <zcorpan> | ok |
| 06:47 | <zcorpan> | Hixie: the spec doesn't say what to do when parsing a legacy color returns an error (for e.g. <hr color>) |
| 06:47 | <Hixie> | bummer. can you drop me a mail or file a bug? |
| 06:47 | <zcorpan> | sure |
| 06:47 | <Hixie> | thanks |
| 06:59 | <hsivonen> | Sigh. alt thread rides again. |
| 07:00 | <zcorpan> | Hixie: there's still "body { color: black; background: white; }" |
| 07:23 | <Hixie> | (oops, i forgot to checkin the last set of edits and my next checkin will be a mixed bag of stuff) |
| 07:24 | <Hixie> | zcorpan: that's right, no? |
| 07:24 | <Hixie> | hsivonen: my bad, sorry |
| 07:24 | <Hixie> | zcorpan: btw use the "save" link for future live dom viewer links |
| 07:26 | <zcorpan> | Hixie: no. consider this author rule: html { color:green; background:yellow } |
| 07:26 | <Hixie> | hm, i guess i'll have to define them as inital values then |
| 07:26 | <Hixie> | if you remind me in ten minutes i'll do it then |
| 07:27 | <zcorpan> | if you remind me to remind you in ten minutes then i'll remind you :) |
| 07:27 | <Hixie> | :-) |
| 07:30 | <Hixie> | ok |
| 07:30 | <Hixie> | what was it again? :-) |
| 07:30 | <Hixie> | oh yes. colors. |
| 07:31 | <Hixie> | Ok, done that. |
| 07:40 | <zcorpan> | Hixie: maybe an element with a tabindex attribute should also match :active? |
| 07:40 | <Hixie> | it can't really be activated? |
| 07:40 | <Hixie> | i wasn't sure what to do |
| 07:41 | <zcorpan> | you usually put elements in the tab order because they can be activated (with js) |
| 07:42 | <Hixie> | i guess |
| 07:47 | <zcorpan> | Hixie: typo: |
| 07:47 | <zcorpan> | + <dt><dfn> title="selector-read-write"><code>:read-write</code></dfn></dt> |
| 07:47 | <Hixie> | thx |
| 07:48 | <zcorpan> | hsivonen: v.nu doesn't catch the above mistake :) |
| 07:49 | <hsivonen> | what's the typo? |
| 07:49 | <hsivonen> | ooh |
| 07:50 | <hsivonen> | blame Hixie and SGML :-) |
| 07:50 | <zcorpan> | might be useful to look at pages that have > in data to see how often it is a mistake |
| 07:51 | <zcorpan> | (most people think it has to be escaped as >) |
| 07:52 | <Hixie> | html5 has > unescaped in a huge number of cases |
| 07:57 | <hsivonen> | ap: is it within your power to make the Finder team normalize to NFC when they *export* text to clipboard? |
| 07:58 | <ap> | hsivonen: the most effective way to proceed would be for you to file a bug at http://bugreport.apple.com, and for me to track comments on it |
| 07:58 | <hsivonen> | ap: ok |
| 07:59 | <hsivonen> | ap: upon login attempt: "An error has occurred. Please report the error to Apple Inc. by emailing the error detail to devbugs⊙ac" |
| 07:59 | <hsivonen> | (no error detail) |
| 08:00 | hsivonen | mails devbugs |
| 08:00 | <ap> | hsivonen: weird. that happened for me a few years ago, and devbugs⊙ac did help |
| 08:03 | hsivonen | tries to remember to file a bug against erroneous glyph assignment in Apple Symbols, too |
| 08:59 | <zcorpan> | annevk5: xhtml compat is nice! don't kill it! :) |
| 09:01 | <hsivonen> | what's annevk5 killing? |
| 09:02 | <Hixie> | nothing to see there |
| 09:02 | <Hixie> | pay no attention to anne |
| 09:02 | <Hixie> | :-P |
| 09:03 | <zcorpan> | hsivonen: table[frames=void] > tr > td |
| 09:03 | <hsivonen> | zcorpan: ah. |
| 09:05 | <annevk5> | *innocent* |
| 09:11 | <Lachy> | Hixie, re http://lists.w3.org/Archives/Public/www-style/2009Feb/0328.html - where in the spec does it say that input type=hidden can't be disabled? |
| 09:11 | <Hixie> | #selector-enabled |
| 09:11 | <Hixie> | oh wait |
| 09:11 | <Hixie> | misread your question |
| 09:12 | <Hixie> | hm, i guess disabled does apply |
| 09:12 | <Hixie> | i'll fix :disabled |
| 09:13 | <Lachy> | ok, so the intention is still that neither :enabled or :disabled match hidden inputs? |
| 09:13 | <Hixie> | yeah |
| 09:13 | <Lachy> | OK, I will go reopen our bug on the issue. |
| 09:16 | <annevk5> | cc me in case I am not |
| 09:17 | <annevk5> | it seems we also need to change our implementation of :indeterminate |
| 09:17 | <Hixie> | oh? |
| 09:17 | <annevk5> | the new rules make that Selector far less expensive I think |
| 09:18 | <annevk5> | http://www.whatwg.org/specs/web-forms/current-work/#relation ":indeterminate Matches radio form control elements when none of the radio buttons in their group are checked." |
| 09:18 | <annevk5> | I'm happy with not adding that back in for what it's worth |
| 09:20 | <Hixie> | ah |
| 09:20 | <Hixie> | yeah i just ignored wf2 when writing this new text |
| 09:20 | <Hixie> | didn't even look at the old text :-) |
| 09:20 | <annevk5> | :read-write and :read-only also changed |
| 09:21 | <Hixie> | yeah that was in response to feedback |
| 09:21 | <Hixie> | see the list |
| 09:21 | <annevk5> | is it really useful to have them apply to all elements and have them be influenced by disabled and such? |
| 09:21 | <annevk5> | k |
| 09:21 | <Hixie> | if you want it changed, change css3ui |
| 09:21 | <Hixie> | i'm just following the spec |
| 09:21 | <Hixie> | i agree the spec is dumb |
| 09:23 | <annevk5> | Hixie, grmbl |
| 09:23 | <Hixie> | indeed |
| 09:23 | <annevk5> | Hixie, :active also applies to disabled <input> controls at this point and only to <button> controls that are not disabled |
| 09:24 | <Hixie> | k, hold on |
| 09:25 | <annevk5> | just so that it's considered, :enabled should still apply to <link href> with .disabled set to true, right? |
| 09:25 | <hsivonen> | Is there a URI for a collection of Philip`'s ill-formedness accomplishments? |
| 09:26 | <Hixie> | annevk5: yes. |
| 09:26 | <Hixie> | annevk5: i was considering making :link/:enabled/etc only apply to <link> when the rel="" is a hyperlink and not a resource link, though |
| 09:26 | <Hixie> | (not gonna do it) |
| 09:26 | <Hixie> | (it'd be too expensive) |
| 09:26 | <annevk5> | (agreed) |
| 09:26 | <Hixie> | (would be cool though) |
| 09:27 | <annevk5> | (UAs scraping a wiki to figure out whether or not :link applies? :p ) |
| 09:27 | <Lachy> | Hixie, why would it be cool? What problem would it solve? |
| 09:27 | <Hixie> | cool things rarely solve actual problems :-P |
| 09:27 | <Lachy> | ok |
| 09:27 | <Lachy> | then I'm glad you're not doing it |
| 09:28 | <Hixie> | it would be cool because it would make :link match <link rel=next> and other such things, which are links, but not <link rel=stylesheet>, which isn't a link the user would care about |
| 09:28 | <Lachy> | (it's arguable whether a user would care about <link rel=next> anyway) |
| 09:29 | <Lachy> | (given that few browsers have any UI for it) |
| 09:29 | <Hixie> | HEH |
| 09:30 | <Hixie> | er |
| 09:30 | <Hixie> | heh |
| 09:30 | <Lachy> | actually, I think Opera's mysterious Fast Forward button does |
| 09:30 | jgraham | is still recovering from the loudness of the heh |
| 09:31 | <Hixie> | d'oh |
| 09:31 | <Philip`> | hsivonen: I wouldn't call them "accomplishments" - all I did was stick some U+FFFFs into query strings :-p |
| 09:31 | <Hixie> | i have caps-lock rebound to scroll-lock on this keyboard |
| 09:31 | <Hixie> | and my scroll-lock is right next to my f12 |
| 09:31 | <Hixie> | which i hit a lot when on irc |
| 09:31 | <Hixie> | so i keep hitting capslock by mistake |
| 09:31 | <Hixie> | not sure where else to put capslock though |
| 09:32 | hsivonen | has considered getting rid of capslock entirely |
| 09:32 | <annevk5> | it seems making the background-color initial value to be white rather than transparent would break an awful lot of pages |
| 09:32 | <annevk5> | that making* |
| 09:32 | <Lachy> | Hixie, what purpose does scroll lock have? Or do you just use it because it's a useless key that does nothing |
| 09:32 | <Philip`> | hsivonen: There isn't any collection that I'm aware of; the closest thing is http://diveintomark.org/archives/2008/03/09/no-fury-like-dracon-scorned#comment-11442 which lists a few examples |
| 09:32 | <jgraham> | I alays try to disable capslock, except I am lazy at the moment because the Gnome UI seems to have lost the option |
| 09:33 | <annevk5> | Hixie, I think you want something else there, e.g. saying that you expect the canvas to be white and html { color:black } |
| 09:33 | <Hixie> | Lachy: none, hence why i use it for capslock |
| 09:33 | <hsivonen> | as far as key remapping goes, I still haven't found a way to introduce an Insert key onto my keyboard for testing Orca and NVDA |
| 09:33 | <Hixie> | annevk5: valid, fixing... |
| 09:33 | <jgraham> | Although it does have a bizzare number of capslock-related options |
| 09:33 | <annevk5> | Hixie, though CSS might already say that the canvas is white somewhere |
| 09:33 | annevk5 | forgot |
| 09:33 | <Philip`> | Hixie: You could remap capslock onto the 'hold down shift' key sequence, and then you wouldn't have to spare an extra key for it |
| 09:33 | <zcorpan> | Hixie: is it useful to have :link and :visited match anything but <a href>? |
| 09:34 | <zcorpan> | <link> and <area> are display:none anyway |
| 09:34 | <hsivonen> | Philip`: sticky shift has slightly different semantics than capslock, though |
| 09:34 | <Hixie> | Philip`: yeah, fair point |
| 09:34 | Hixie | tries to remember how to unmap keys so that his useless capslock key turns into a useless scrolllock key again |
| 09:35 | <Hixie> | oh actually it's not scroll lock |
| 09:35 | <Hixie> | it's printscr/sysreq |
| 09:36 | <zcorpan> | Hixie: i usually remove the key altogether |
| 09:36 | <Hixie> | wow, i crashed my keyboard. |
| 09:36 | <Hixie> | oh i remember why i had capslock bound to this key now |
| 09:37 | <Hixie> | it's because i'd rebound that key elsewhere, and then turned it into a macro for command-tab |
| 09:37 | <Hixie> | so now if i miss my f12 key instead of getting caps lock, i'll end up in a different app |
| 09:37 | <Hixie> | oh well |
| 09:38 | hsivonen | has mapped two key for tab, one for each hand |
| 09:38 | <Lachy> | zcorpan, Opera matches <link href=""> and <area href=""> with :link and :visited. It may be useful with selectors API, even if its not so useful with CSS |
| 09:38 | <zcorpan> | Lachy: oh we do? can we click <link>s too? |
| 09:39 | <annevk5> | "not when using proper tools" |
| 09:39 | <annevk5> | "writing proper tools is non-trivial" |
| 09:39 | <annevk5> | -> problem |
| 09:39 | <Lachy> | zcorpan, I don't understand the question |
| 09:39 | <hsivonen> | Philip`: OK. I ended up referring to http://diveintomark.org/archives/2008/03/09/no-fury-like-dracon-scorned#comment-11442 via www-archive indirection |
| 09:40 | <zcorpan> | Lachy: if you make a <link> visible and have content, does clicking it make opera follow the link? |
| 09:40 | jgraham | successfully locates the GUI option to disable capslock |
| 09:40 | <Lachy> | I don't know |
| 09:41 | <zcorpan> | hsivonen: interesting, which is the other key you have mapped to tab? |
| 09:41 | <Lachy> | zcorpan, no, that doesn't work |
| 09:42 | <hsivonen> | zcorpan: I've mapped what used to be caps lock to tab, and I have mapped what used to be right-hand ctrl to tab |
| 09:42 | <zcorpan> | hsivonen: ok |
| 09:43 | <Hixie> | i don't have a right-hand control anymore |
| 09:43 | <Hixie> | my right hand modifiers are alt, command, and shift |
| 09:43 | <Lachy> | why is there even a need to disable caps lock? It's useful on rare occasions and it's not that hard to avoid pressingit |
| 09:43 | <Hixie> | (and left hand are shift, control, and command) |
| 09:44 | <Hixie> | Lachy: no need, i just needed the key for another purpose (one-key command-tab so i don't have to stretch my hand to hit that combination) |
| 09:44 | <jgraham> | Lachy: It is the single most pointless key on the keyborad |
| 09:44 | <Hixie> | (i was getting hand strain from it) |
| 09:44 | <Hixie> | jgraham: it's high up there, but it's at least third, maybe fourth |
| 09:44 | <jgraham> | For me it causes problems far more than it solves them |
| 09:44 | <jgraham> | So it has negative value |
| 09:44 | <jgraham> | Whereas others just have no value |
| 09:45 | <Lachy> | Hixie, you said before that you had it mapped to scroll lock, and now you're saying it's Cmd+Tab? |
| 09:45 | <Hixie> | Lachy: my caps lock key is mapped to print screen (i thought it was scroll lock, but i was mistaken), and the print screen function is mapped to a macro that hits command-tab |
| 09:45 | jgraham | wonders what Hixie has F12 doing |
| 09:46 | <Hixie> | f12 inserts a line in my irc window so i know how far i've read |
| 09:46 | <Hixie> | (and in emacs, it brings up the shell) |
| 09:46 | <jgraham> | Interesting |
| 09:49 | <Hixie> | (i also rebound the "international" key to command-tilde) |
| 09:51 | <Lachy> | what's the "international" key? |
| 09:53 | <Hixie> | it's labelled "| \ Insert", dunno what it is |
| 09:53 | <Hixie> | (i have another key with "| \" on the other side) |
| 09:54 | <Lachy> | I've never seen a key labelled like that. What keyboard layout do you have? |
| 09:54 | <Lachy> | obviously it's not a standard US layout |
| 09:54 | <zcorpan> | Hixie: the margin collapse thing is still wrong because the margins are 0 even when there's a border (and so the margins cannot collapse with each other) |
| 09:54 | <zcorpan> | Hixie: on body, th, td |
| 09:54 | <Hixie> | lachy: standard US qwerty/dvorak dual label kineses advantage |
| 09:54 | <Hixie> | kinesis |
| 09:56 | <annevk5> | Hixie, the initial value of the color property is also not expected to be black; e.g. <span style=color:green><span>this is green</span></span> |
| 09:57 | <hsivonen> | Lachy: the "international key" also appears on typical European PC keyboard hardware |
| 09:57 | <hsivonen> | i.e. there's a key there |
| 09:57 | <Hixie> | annevk5: color is inherited |
| 09:57 | <hsivonen> | but the U.S. layout makes it a duplicate of the backslash key |
| 09:58 | <Lachy> | Hixie, this one? http://www.fentek-ind.com/kinesis_dvorak.gif |
| 09:58 | <Lachy> | although I don't see any key labelled "| \ Insert" |
| 09:58 | <hsivonen> | Lachy: the international key is the one to the left of left arrow on that chart |
| 09:58 | <Hixie> | Lachy: yeah |
| 09:59 | <Hixie> | Lachy: next to ~` and <- and -> |
| 09:59 | <zcorpan> | Hixie: i think annevk5 is saying that per css, color's initial value is 'inherit', but per html5 it's instead 'black' |
| 09:59 | <Hixie> | zcorpan: anne is wrong then. no property has initial value 'inherit' |
| 09:59 | <Lachy> | oh, that's just the backslash key. Why are you calling it the "international" key? |
| 09:59 | <Hixie> | the backslash key is the one above the right shift |
| 10:00 | <Hixie> | below the question mark / forward slash key |
| 10:00 | <Hixie> | (and the question mark / forward slash key is the backslash key on the qwerty layout) |
| 10:01 | <zcorpan> | Hixie: ok, but still, the initial value of background-color is transparent |
| 10:01 | <Hixie> | wow i just had the hardest time typing qwerty, despite actually now being on a qwerty keyboard |
| 10:01 | <Hixie> | zcorpan: already fixed |
| 10:01 | <zcorpan> | ok |
| 10:02 | <Lachy> | ok, so the international key which is labelled with "| \" on the left, doesn't actually type | or \? |
| 10:02 | <Hixie> | it does on the US standard layout, but on european layouts, e.g., it would do something else |
| 10:02 | <Hixie> | depending on the layout |
| 10:02 | <Hixie> | it's basically an extra key for layouts that ned another key |
| 10:02 | <Hixie> | many US keyboards don't physically have this key |
| 10:03 | <Lachy> | ok, here's a standard slightly less confusing european layout http://a248.e.akamai.net/7/248/51/3049331182291562/www.info.apple.com/images/kbase/304933/304933_08b.gif |
| 10:04 | <annevk5> | ah you're right |
| 10:04 | <annevk5> | maybe we should just change the initial value to 'black' |
| 10:04 | <annevk5> | in css3-color or so |
| 10:05 | <hsivonen> | Hixie: how do I find out if a script has done "crazy modifications" without always doing thread synchronization in the AAA and without keeping crazy state in the tree? |
| 10:05 | <jgraham> | Lachy: The Apple idea of what constitutes a non_US english keyboard layout is different to anyone else's idea |
| 10:07 | <annevk5> | Hixie, a/canvas's/canvas'/ ? |
| 10:08 | <Hixie> | hsivonen: pass an assertion along to the main thread, and rewind if the assertion fails (it rarely will) |
| 10:08 | <Lachy> | jgraham, aside from the Cmd key, isn't it close enough? |
| 10:09 | <hsivonen> | Hixie: in that case, I'd have to keep around enough state to be able to rewind |
| 10:09 | <hsivonen> | Hixie: keeping around that state as seldom as </script> occurs is OK, keeping it for many token boundaries is less so |
| 10:10 | <Hixie> | hsivonen: that's pretty much how i feel about your alternative suggestion :-) |
| 10:10 | <Hixie> | hsivonen: you want everyone to keep this state instead of just you, to be able to do something that doesn't seem particularly useful in the first place :-) |
| 10:11 | <hsivonen> | Hixie: I agree that it probably doesn't make sense to spec it without there being a demo implementation showing benefit |
| 10:12 | <jgraham> | Lachy: It moves a bunch of the punctuation around compared to the standard en-GB keyboard |
| 10:12 | <Hixie> | it's not like you have to keep much state to rewind anyway |
| 10:12 | <Hixie> | just the position in the input stream, and what data was inserted into the stream after the assertion (which can be done by annotating insertions with an age) |
| 10:13 | <Hixie> | just include that information with the assertion |
| 10:13 | <Hixie> | and have the main thread include it when it complains |
| 10:35 | <annevk5> | Hixie, HTMLInputElement still does not have width and height DOM attributes |
| 10:40 | <Hixie> | fixed |
| 10:54 | <annevk5> | so it's not legal to write <table><tr> in HTML or it is and I'm misunderstanding the writing section? |
| 11:00 | <Lachy> | annevk5, it is legal because the tbody element's start tag is optional |
| 11:41 | <annevk5> | funny |
| 11:41 | <annevk5> | "the tools will save us" on google gives "XML.com: XML on the Web Has Failed" as top 3 result |
| 11:45 | <hsivonen> | hah |
| 12:14 | <yecril71> | The Scroll Lock key toggles the Scroll Lock mode. |
| 12:14 | <yecril71> | A fictional language cannot be invented; art is not an invention. |
| 12:15 | <yecril71> | And a fictional language cannot be used, except by fictional characters; once it is used, it stops being fictional. |
| 12:17 | <yecril71> | I hope non-interactive documents will be readable and have a similar organization in various browsers. |
| 12:17 | <yecril71> | (organisation = visual layout) |
| 12:18 | <yecril71> | Otherwise, HTML cannot be used for publishing and authors will have to stay with PDF. |
| 12:19 | <yecril71> | (I admit it is far from the Web Applications trend) |
| 12:21 | hendry | wonders if there is already a tool to split up a large HTML file into seperate HTML files (with next/prev). Want to chunk an HTML book. |
| 12:22 | <yecril71> | Split on H2 or what? |
| 12:23 | <hendry> | yecril71: split on 1000 lines or something I was thinking |
| 12:23 | <jgraham> | hendry: Philip` just wrote a custom script for splitting the HTML5 spec |
| 12:23 | <yecril71> | 1000 lines no good, the user will be lost. |
| 12:24 | <hendry> | jgraham: ah cool, could you point me to the source? or Philip` ? |
| 12:24 | <hendry> | yecril71: these are project Gutenberg ebooks. HTML formatting is not great. I think I am probably better off with text. |
| 12:25 | <hendry> | yecril71: and using `split` and wrap it in some HTML. |
| 12:27 | <jgraham> | hendry: I guess I can do something like |
| 12:27 | <yecril71> | If they are in HTML, I would count top-level elements, not lines |
| 12:27 | <jgraham> | Philip`: ^ |
| 12:27 | <hsivonen> | hendry: I'm curious: what kind of reading environment needs splitting? A device with 64 MB of RAM handles unsplit Gutenberg ebooks unsplit |
| 12:28 | <yecril71> | I think, whatever the hardware is, the software should be smart enough to handle that somehow. |
| 12:28 | <hendry> | hsivonen: this HTC device running Pocket IE seems to have issues with 200K files. |
| 12:28 | <hsivonen> | hendry: you need a better browser :-) |
| 12:30 | <hendry> | hsivonen: i don't have too much choice as this widget runtime uses that browser control. :-) so you read ebooks eh? How do you 'bookmark' the page you're on? |
| 12:30 | <yecril71> | The bookmarks should already be there. |
| 12:31 | <hendry> | yecril71: explain? |
| 12:31 | <hsivonen> | hendry: I try to avoid quitting the browser and failing that, remembering which chapter I was in |
| 12:31 | <yecril71> | Headers have identifiers, these identifiers constitute bookmarks you can use. |
| 12:31 | <yecril71> | You can discover them by looking at the table of contents. |
| 12:32 | <yecril71> | Failing that, remember a key phrase and perform a text search. |
| 12:32 | <Philip`> | hendry: http://code.google.com/p/html5/source/browse/trunk/spec-splitter/spec-splitter.py |
| 12:33 | <killerboy> | hello |
| 12:33 | <Philip`> | hendry: but that's very much designed for HTML5, not for general-purpose use |
| 12:33 | <Philip`> | hendry: (and it splits on <h2>s and a custom list of elements IDs, not on page size) |
| 12:33 | <yecril71> | The assignment is trivial once you have the DOM. |
| 12:33 | <jgraham> | A good ebook reader would allow you to easilly resume from an arbitary point |
| 12:33 | <hendry> | Philip`: thanks and understood |
| 12:34 | <killerboy> | i'd like to know who can read localStorage. Is access domain based, and if such, what about ISP which give you names like: http://www.ispdomain.com/~mysite/ |
| 12:35 | <Philip`> | I keep my place in ebooks by printing them out and using a scrap of paper as a bookmark |
| 12:35 | <annevk5> | killerboy, http://www.whatwg.org/specs/web-apps/current-work/multipage/structured-client-side-storage.html#cross-directory-attacks |
| 12:37 | <killerboy> | hmm, isn't path based access possible? |
| 12:37 | <annevk5> | no |
| 12:37 | <killerboy> | and why not ;-) |
| 12:37 | <annevk5> | because you can circumvent it as the spec says |
| 12:37 | <killerboy> | how is it possible to circumvent it? |
| 12:37 | <killerboy> | some examples? |
| 12:38 | <annevk5> | e.g. inserting an <iframe> with a different path on the same domain and then running a script in that context |
| 12:38 | <killerboy> | ok |
| 12:38 | <killerboy> | but same problem with domains isn't it? |
| 12:39 | <killerboy> | why you cannot treat paths as "long domains" and use same solution as for domains |
| 12:39 | <annevk5> | no, <iframe>s have a cross-origin restriction |
| 12:40 | <annevk5> | so it's a different problem |
| 12:40 | <annevk5> | (or no problem, depending on how you look at it) |
| 12:40 | <killerboy> | so you can add this restriction and make it path based, but add some tag to iframe, to turn it off etc. |
| 12:41 | <killerboy> | or something similar |
| 12:41 | <annevk5> | I'd suggest you first get a basic understanding of the Web security model before trying to change it |
| 12:41 | <killerboy> | ok |
| 12:41 | <annevk5> | well, more than basic would probably be good :) |
| 12:41 | <killerboy> | some links? |
| 12:42 | <annevk5> | whatwg.org/html5 defines a bunch of it |
| 12:42 | <killerboy> | ok |
| 12:42 | Philip` | doesn't remember having ever seen a decent description of what the Web security model actually is |
| 12:42 | <killerboy> | i'll come back after reading |
| 12:44 | <jgraham> | Philip`: I thought it was defined to be "duct tape and string" |
| 12:44 | <annevk5> | killerboy, cool! |
| 12:44 | <Philip`> | Mostly I've just seen that it's complex and fragile and poorly designed, and nobody is willing to make it worse in any way, and it's very hard to make it better in some ways without making it worse in other ways, so nothing much ever changes |
| 12:45 | <annevk5> | Philip`, e.g. http://dev.w3.org/2006/waf/access-control/ is a pretty major change |
| 12:45 | <hsivonen> | was Same Origin in Netscape 2.0 or was it added later? |
| 12:46 | <Philip`> | annevk5: Since the first draft was almost four years ago and it's still not a feature I can use today, I'll interpret that as evidence that changing the security model is very hard :-) |
| 12:48 | <annevk5> | indeed :) |
| 13:07 | <annevk5> | hsivonen, per https://developer.mozilla.org/En/Same_origin_policy_for_JavaScript it was 2.0 |
| 13:07 | <hsivonen> | annevk5: thanks |
| 13:23 | gsnedders | waves |
| 13:26 | <takkaria> | heh |
| 13:26 | <takkaria> | in my Logic lectures at the moment, we're doing axiomatic proofs |
| 13:27 | <takkaria> | I wanted to shout "Fear the axiomatic proof!" |
| 13:27 | <killerboy> | is it ##logic channel ;-) ? |
| 13:27 | <annevk5> | takkaria, lol, the moment you said axiomatic proofs I looked up http://ln.hixie.ch/?start=1041735552&count=1 |
| 13:28 | <takkaria> | :) |
| 13:29 | <Philip`> | How can a non-axiomatic proof be a proof at all? |
| 13:30 | <Philip`> | It's kind of hard to do anything unless you're starting with some axioms |
| 13:30 | <takkaria> | if you have a system that uses natual deduction instead of axioms you can |
| 13:30 | <annevk5> | haha: http://www.google.com/search?q=axiomatic+proof |
| 13:30 | <killerboy> | takkaria, :-) |
| 13:32 | <killerboy> | annevk5, i've read security related parts of html5 |
| 13:32 | <gsnedders> | takkaria: What are you doing at uni? |
| 13:32 | <killerboy> | all it says is that it's domain based |
| 13:33 | <Philip`> | takkaria: Hmm, I appear to know nothing about that :-( |
| 13:33 | <takkaria> | gsnedders: philosophy |
| 13:34 | <annevk5> | killerboy, it's origin based |
| 13:34 | <killerboy> | annevk5, which is domain based |
| 13:34 | <gsnedders> | takkaria: That's what I thought. I guess logic has an effect on that. :) |
| 13:34 | <killerboy> | +scheme etc... |
| 13:35 | <killerboy> | +port, +certificates, and other additional data |
| 13:35 | <takkaria> | gsnedders: formal logic is pretty much inseperable from western philosophy as of the early twentieth century |
| 13:35 | <killerboy> | takkaria, if you're interested in logic, come to ##logic, you're welcomed |
| 13:36 | <annevk5> | now combine that with the fact that it has been like that since Netscape 2.0 and try to comprehend that making it path based at this point just because localStorage does not work on shared hosting providers that use directories is not going to happen |
| 13:36 | <takkaria> | gsnedders: the most prominent western philosophers from 1880-1930 or so were also logicians and mathematitians |
| 13:36 | <takkaria> | killerboy: hm, I may do |
| 13:36 | <killerboy> | takkaria, which year of philo? |
| 13:37 | <takkaria> | killerboy: second |
| 13:38 | <killerboy> | annevk5, i understand that. you might remember this line: <killerboy> hmm, isn't path based access possible? <annevk5> no |
| 13:38 | <killerboy> | annevk5, maybe we used "possible" in different meanings |
| 13:39 | <killerboy> | takkaria, ok |
| 13:40 | <annevk5> | killerboy, ok |
| 13:58 | <Philip`> | hsivonen: Does it count as a bug whenever nu.validator.htmlparser.tools.XML2XML produces output that is not well-formed XML? |
| 13:58 | <Philip`> | e.g. when the input is <?xml version="1.1"?><x xmlns:x=""/> |
| 14:28 | <killerboy> | was canvas influenced by postscript? |
| 14:28 | <annevk5> | ask Apple :) |
| 14:29 | <annevk5> | but I think that might be the case |
| 14:30 | <Philip`> | killerboy: I'd guess PDF is part of the chain that led to the canvas API |
| 14:30 | <killerboy> | it's because function names are similar :-) |
| 14:31 | <Philip`> | killerboy: like maybe PS -> PDF -> Quartz -> canvas, or something, though I know very little about anything but the last of those stages, so maybe I'm totally wrong |
| 14:31 | <killerboy> | ok |
| 14:52 | <hsivonen> | Philip`: yes, it counts as a bug. I was relying on the XML parser enforcing input XML 1.0-compliance. I guess one can't trust any third-party code. |
| 14:53 | <annevk5> | to wit, that is XML 1.0 compliant |
| 14:54 | <hsivonen> | umm. XML 1.0 plus Namespaces in XML 1.0 |
| 14:54 | <annevk5> | that's not what I meant, but I'm wrong |
| 14:54 | <annevk5> | even in 5th edition xmlns:x="" is incorrect |
| 14:55 | <Philip`> | hsivonen: Does the parser ever claim to be an XML 1.0 parser? |
| 14:56 | <hsivonen> | Philip`: in my defense, my serializer meets its spec. but my choice of SAX event generator doesn't. So you win. |
| 14:56 | <hsivonen> | good question |
| 14:56 | <Philip`> | hsivonen: But your serialiser's spec seems to be "if you give it input that won't result in ill-formed output, then it will serialise to well-formed output" :-p |
| 14:58 | <gsnedders> | Yay! Now building index of RFCs from XML index in less than two seconds. |
| 14:58 | <hsivonen> | Philip`: the spec is that text content, attribute values and PI data are considered untrusted but namespace mappings, element names, attribute names and PI targets are considered trusted |
| 14:58 | <gsnedders> | (from a local copy of the XML index — doing it with urllib2 and getting the index takes around five minutes :P) |
| 14:58 | <hsivonen> | Philip`: also, namespace mappings are not required |
| 14:58 | <hsivonen> | Philip`: so in this case, a namespace mapping is fed to the serializer and the mapping is trusted |
| 15:00 | <hsivonen> | Philip`: it seems that SAXParserFactory is silent on XML version |
| 15:02 | <hsivonen> | let this be evidence that XML 1.1 is Trouble |
| 15:06 | <Dashiva> | Who is the picture of in the latest lastweek? |
| 15:06 | <hsivonen> | Philip`: it turns out that SAX makes detecting the situation pretty annoying, too. |
| 15:07 | <hsivonen> | Dashiva: Hixie |
| 15:11 | <killerboy> | thanks for all, bye |
| 15:13 | <gsnedders> | Ah, without glasses. |
| 15:17 | <hsivonen> | what does the following sentence mean? "That the world has been doing evil while processing XML as a browser format (MSIE and Mozilla being potentially the biggest actors) is not debatable anymore and justifies fully the lack of widespreadness." |
| 15:18 | <annevk5> | I have no idea; I was also wondering how we obscure prossing models in browsers |
| 15:18 | Philip` | didn't understand it when he first read |
| 15:18 | <gsnedders> | Where is that from? |
| 15:19 | <hsivonen> | gsnedders: www-tag |
| 15:19 | <Philip`> | (nor any subsequent times I read it) |
| 15:20 | <annevk5> | s/prossing/processing/ (maybe this is how) |
| 15:22 | <jcranmer> | lemme try to understand it |
| 15:23 | <jcranmer> | a. The world has been doing evil |
| 15:23 | <jcranmer> | a.1 This evil is processing XML as a browse format |
| 15:23 | <jcranmer> | a.2 MSIE and Mozilla are potentially the biggest actors |
| 15:23 | <jcranmer> | b. This fact is not debatable anymore |
| 15:24 | <jcranmer> | c. This fact also justifies why it has failed to become widespread |
| 15:24 | <jcranmer> | although the antecedent of the last it eludes me |
| 15:24 | <hsivonen> | jcranmer: so is "the world" or "MSIE and Mozilla" doing evil? |
| 15:24 | <annevk5> | I'm guessing he meant that XML did not became widespread because browsers processed it as HTML and not as XML and the fault lies with Mozilla/MSIE |
| 15:25 | <jcranmer> | I'm guessing "the world" == browser vendors, primarily "MSIE and Mozilla" |
| 15:25 | <hsivonen> | but Mozilla does process XML as XML |
| 15:25 | <annevk5> | I'm taking several leaps here of course |
| 15:26 | <annevk5> | hsivonen, I'm guessing he wanted XHTML as text/html to be processed as XML too |
| 15:26 | <annevk5> | (if anyone really wants to know they should ask the person who wrote it, to be clear) |
| 15:28 | <hsivonen> | annevk5: I think the blame for text/html handling lies largely on the W3C itself and O'Reilly |
| 15:29 | <annevk5> | yeah, the old HTML WG decided that |
| 15:29 | <annevk5> | I'm glad they did |
| 15:31 | <MikeSmith> | hsivonen: why O'Reilly? |
| 15:33 | <hsivonen> | MikeSmith: when Netscape/Mozilla made the decision, O'Reilly xml.com was already serving ill-formed XHTML as text/html |
| 15:33 | <MikeSmith> | ah |
| 15:34 | <Dashiva> | Which side of the draconian debate was O'Reilly on? |
| 16:16 | annevk5 | figures out how to read the Web IDL ECMAScript binding part on "Operations" |
| 16:16 | annevk5 | is glad he's not writing that specification |
| 16:31 | <MikeSmith> | has anybody used Yahoo BrowserPlus much? |
| 16:41 | Philip` | refrains from mentioning how even if XOM guaranteed well-formed output, it could result in DOS attacks when, say, your web site stores the visitors' IP addresses in an SQL database, and then appends the IP values as text nodes when generating a document to display to another user |
| 16:41 | <Philip`> | ...e.g. in a wiki-type situation |
| 16:42 | <Philip`> | ...and then the user tricks your system by sending an X-Forwarded-For header with invalid bytes in it, which your code treats as the user's IP address and stores in the database |
| 16:42 | <Philip`> | ...and when you try to construct a XOM tree for subsequent users, you get a fatal exception because of those invalid bytes |
| 16:44 | <Philip`> | (At least it's better than using string concatenation, where you'll probably get an XSS attack rather than a DOS attack) |
| 16:45 | <annevk5> | haha |
| 16:45 | annevk5 | blogged it |
| 16:49 | <hsivonen> | what was the meeting where W3C decided to stop working on HTML? in 2000 or thereabouts |
| 16:49 | <Philip`> | annevk5: It's only a bug, it's not that exciting really :-p |
| 16:50 | <annevk5> | last XML {day} was also just a bug |
| 16:51 | <Philip`> | annevk5: That one wasn't that exciting really either :-) |
| 16:51 | <annevk5> | indeed, my blog is not that exciting :) |
| 16:51 | <annevk5> | just a blog after all |
| 16:54 | <hsivonen> | has MS ever acknowledged the disbanding of the IE team after IE6? |
| 16:54 | <hsivonen> | or is the official story that they were working on IE7 all along? |
| 16:55 | <Philip`> | hsivonen: Is there a sensible reason why you output 
 for \r in comments, instead of just dropping the character or outputting a space or something? |
| 16:55 | <Philip`> | hsivonen: (Anything is going to cause data loss, but 
 seems a weirder thing to pick than the alternatives) |
| 16:55 | <hsivonen> | Philip`: the reason is code reuse, and yes, you've found a bug |
| 16:56 | <Philip`> | hsivonen: (particularly since "--" is written as "- -", not "-&#xwhatever;") |
| 16:56 | <Philip`> | hsivonen: Ah, okay |
| 16:56 | <Philip`> | hsivonen: (It affects PIs too) |
| 16:56 | <hsivonen> | Philip`: thanks |
| 16:58 | <MikeSmith> | hsivonen: it was 1998 |
| 16:58 | <MikeSmith> | http://www.w3.org/MarkUp/future/ |
| 16:59 | <hsivonen> | MikeSmith: thanks |
| 16:59 | <MikeSmith> | workshop somewhat ironically (in hindsight) titled "Shaping the Future of HTML" |
| 17:00 | <MikeSmith> | lots more history stuff at http://esw.w3.org/topic/HTML/history |
| 17:00 | <hsivonen> | MikeSmith: thanks |
| 17:00 | <hsivonen> | (I'm writing the history part for my HTML5 lecture) |
| 17:02 | <Philip`> | MikeSmith: That Future thing doesn't sound much like XHTML |
| 17:03 | <MikeSmith> | hsivonen: please make any updates to that page if you find anything to add to it |
| 17:03 | <hsivonen> | MikeSmith: OK |
| 17:04 | <MikeSmith> | Philip`: maybe the right Future has not arrived yet |
| 17:05 | <MikeSmith> | or maybe that workshop actually caused an alternate universe to spring into existence -- one where everything it working just as envisioned by the participants |
| 17:06 | <Philip`> | (I suppose it sounds somewhat like XHTML2, though) |
| 17:07 | <Philip`> | Clearly the problem was that Mozilla and Google didn't participate in that workshop |
| 17:08 | <Philip`> | I can't imagine why they thought it was unimportant they could simply ignore it |
| 17:08 | <Philip`> | s/was/was so/ |
| 17:17 | <Philip`> | http://www.w3.org/MarkUp/future/papers/roconnor.html - <?IS10744:arch name="HTML 5.0" public-id="-//W3C//NOTATION HTML 5.0 ARCHITECTURE//EN" dtd-public-id="-//W3C//DTD HTML 5.0//EN" doc-elem-form="HTML"> - yeah, I think I prefer <!doctype html> |
| 17:19 | <MikeSmith> | gsnedders: maybe you can do a comparative analysis of RDF, XSD, and Architectural Forms |
| 17:20 | <gsnedders> | MikeSmith: In terms of how horrible they are? :) |
| 17:22 | <Philip`> | http://web.archive.org/web/20010606223122/www.sisu.se/~william/Papers/Image6.gif - wow |
| 17:23 | <Dashiva> | That's so... 90s |
| 17:25 | <Philip`> | I don't remember the 90s being that bad |
| 17:26 | <Lachy> | LOL, "The transition to supporting HTML architectures can be made easier by allowing user agent support to be optional" (from that HTML 5.0 paper Philip` linked to) |
| 17:27 | <Lachy> | I just wonder how intentional lack of support can make any transition easier |
| 17:27 | <Dashiva> | "Most authors will validate their documents against the HTML DTD since it provides enough structure." |
| 17:28 | <Philip`> | http://datapeak.net/images/Windows-95-Start-Button.png - even Win95 had nice 3Dish buttons and full-colour background images |
| 17:28 | <Philip`> | There's no excuse for using dark cyan as your primary UI colour |
| 17:28 | <Dashiva> | That's more like win3.11 days |
| 17:29 | <Dashiva> | Early 90s then :) |
| 17:29 | <Philip`> | But that screenshot was from 1998 |
| 17:29 | <Dashiva> | Legacy software |
| 17:29 | <Philip`> | and the developers can't have kept it alive from the early 90s up until 1998 without tearing their eyeballs out |
| 17:30 | <Dashiva> | Maybe they're color blind? |
| 17:31 | <Philip`> | Does colour blindness give you the ability to read horrendously dithered logos? |
| 17:31 | <MikeSmith> | I'm looking forward to the day where people look back on the history of HTML5 and make smartass comments about what idiots we all were. |
| 17:31 | <MikeSmith> | wait, people are doing that already |
| 17:31 | <Dashiva> | We're ahead of the curve |
| 17:31 | <MikeSmith> | so we are way ahead of the curve |
| 17:31 | <MikeSmith> | damn, we even think the same things at the same time |
| 17:32 | <Lachy> | you idiots! |
| 17:32 | <MikeSmith> | HTML5 borg collective |
| 17:32 | <gsnedders> | Another thing I'd like in html5lib: some way to indent a whole tree nicely |
| 17:32 | <Dashiva> | The vast #whatwg hivemind |
| 17:33 | <MikeSmith> | gsnedders: postprocess it through xmllint --format --html |
| 17:33 | <MikeSmith> | if's it html |
| 17:34 | <gsnedders> | MikeSmith: How does that cope with optional tags? Does it just parse it and serialize it? If so, libxml2's serializer has issues |
| 17:34 | <Philip`> | Lachy: That wasn't a smartass comment, that was just hurtful and abusive :'-( |
| 17:34 | <MikeSmith> | gsnedders: dunno |
| 17:35 | <Lachy> | Philip`, :-) |
| 17:35 | <Philip`> | Does it indent <pre> and <textarea> contents too? |
| 17:35 | <Dashiva> | Philip`: How do you think XHTML sites feel when you push invalid content on them? |
| 17:35 | <MikeSmith> | Philip`: probably |
| 17:37 | <mpt> | Dashiva, jaundiced |
| 17:37 | <Philip`> | Dashiva: They are relieved to be free from the burden of publishing large quantities of information, and happy to sit back and relax in the warm yellow glow |
| 17:38 | <gsnedders> | Philip`: But I use WebKit based browsers! :'( |
| 17:42 | <MikeSmith> | "I repeat B.S again." is an interesting statement |
| 17:45 | <MikeSmith> | maybe we should have a rule that anybody working on standards related to browser technologies should have to work at a browser project for 3 months or something |
| 17:45 | <MikeSmith> | doing QA and responding to bug reports |
| 17:45 | <MikeSmith> | like military service |
| 17:47 | <mpt> | Do user interface bug reports count? |
| 17:49 | <MikeSmith> | mpt: they count for half |
| 17:52 | <Philip`> | Is 3 months long enough for people to stop saying "RESOLVED INVALID: your page is ugly tag soup and doesn't validate, you should fix it before complaining about bugs" in response to every report? |
| 17:53 | <MikeSmith> | Philip`: that's what we have the electric shocker for |
| 17:53 | <MikeSmith> | all part of the plan |
| 17:54 | <mpt> | That was part of the original plan for Mozuki |
| 17:54 | <mpt> | If pages didn't validate, style sheets would be ignored |
| 17:55 | <mpt> | I don't think anyone ever used that browser, though |
| 17:55 | <Dashiva> | What about @style? |
| 18:00 | <sicking> | hsivonen, ping |
| 18:05 | <Philip`> | mpt: http://web.archive.org/web/20000817010453/http://www.infidels.org/~meta/mozuki/ ? |
| 18:06 | Philip` | fails to find any more information on it |
| 18:07 | <mpt> | That's the one! |
| 18:21 | <Philip`> | hsivonen: Is there a sensible reason why processingInstruction adds a space at the end if 'data' ends in '?'? |
| 18:21 | <Philip`> | (i.e. processingInstruction("foo", "?") becomes <?foo ? ?> instead of <?foo ??>) |
| 18:24 | <annevk5> | hober, dude, there's no 1998 XHTML namespace |
| 18:33 | <Philip`> | annevk5: http://www.google.com/search?q=%22www.w3.org%2F1998%2Fxhtml - some people disagree |
| 18:34 | <annevk5> | that's awesome |
| 18:44 | mpt | is momentarily confused as to why annevk5 isn't annevk6n |
| 18:44 | <annevk5> | http://twitter.com/WoollyMittens/statuses/1199866236 ? |
| 18:44 | <annevk5> | mpt, what's the n? |
| 18:45 | <mpt> | annevk5, the last letter, as in i18n and l10n |
| 18:46 | <annevk5> | :) |
| 18:50 | <hober> | annevk5: yeah, I was just repeating what larry said in his email. s/8/9/g. |
| 19:24 | <roc> | hsivonen: I don't know what you mean by "official" |
| 19:25 | <roc> | but various Microsoft employees have acknowledged it in public statements |
| 19:25 | <roc> | I'm not aware of anyone trying to pretend they were working on IE7 all along |
| 19:37 | <millam> | Anybody here working on video and media elements (related to accessibility in particular) that I can yak with? |
| 21:23 | <Hixie> | holy crap microsoft just licensed the IE logo as creative commons non-commercial by-attribution share-alike |
| 21:24 | <Philip`> | It's just a shame it's 17x17, not 16x16 :-) |
| 21:24 | <virtuelv> | Hixie: sauce? |
| 21:24 | <Hixie> | turns out i don't need it anymore either |
| 21:24 | <Hixie> | :-) |
| 21:26 | <Philip`> | virtuelv: http://lists.w3.org/Archives/Public/public-html/2009Feb/0240.html |
| 21:28 | <Philip`> | In response to a message which isn't on public-html yet: Microsoft's test feedback cycle appears to be an awful lot slower than the usual approach where I comment on IRC to the person who wrote the test and they fix it |
| 21:30 | <Lachy> | it's too bad the share-alike licence makes it unusable in the spec too |
| 21:31 | <Lachy> | well, maybe it doesn't. It may just mean you can't use a more restrictive licence than share-alike |
| 21:31 | <Philip`> | That's better than the other icons, which don't come with any licensing information whatsoever |
| 21:37 | <heycam> | i wonder if showing it as less than full opacity is considered a derivative |
| 21:38 | <heycam> | *at less |
| 21:40 | <Lachy> | aargh, I wish that namespace issue in the HTMLWG would just be closed already. There's little point even discussing the issue further, since there is absolutely no technical justification for us using an alternative namespace. |
| 21:41 | <Lachy> | but it's nice to see Rob blatantly ignoring technical arguments again |
| 21:45 | <yecril71> | Windows Help used green for hyperlinks, and so did the Macintosh port. |
| 21:45 | <yecril71> | QuickView. |
| 21:45 | <Lachy> | yecril71, it would be nice to have some context for your comments |
| 21:46 | <yecril71> | > Every color browser I have ever heard of back to the dawn of the Web uses blue for links. |
| 21:46 | <Lachy> | link? |
| 21:47 | <Lachy> | or at least a subject line |
| 21:47 | <yecril71> | Re: [whatwg] [html5] Rendering of interactive content |
| 21:47 | <yecril71> | Admittedly, neither was a Web browser, of course. |
| 21:50 | <virtuelv> | windows help, while a hypertext system isn't HTML either |
| 21:50 | <virtuelv> | (isn't/wasn't, I don't know much about Microsoft after windows 2000, or thereabouts) |
| 21:51 | <yecril71> | It is blue now :-) |
| 21:51 | <Lachy> | virtuelv, the old windows help with green links wasn't. But they've since replaced it with HTML based help, now with blue links. |
| 21:52 | yecril71 | envies virtuelv |
| 21:54 | <yecril71> | And AppleGuide had it blue too |
| 21:54 | heycam | remembers writing horrible RTF and compiling to .hlp files many moons ago... |
| 21:55 | Philip` | remembers not writing Windows help files, because he could never figure out how to do it or what tools he needed |
| 21:55 | yecril71 | concurs to heycam |
| 21:56 | <virtuelv> | speaking of hypertext |
| 21:56 | <virtuelv> | I'm a bit curious why no one ever made a hypertext-based text adventure |
| 21:57 | <virtuelv> | zork-as-html |
| 21:57 | <svl> | virtuelv: I've seen several back in the late 90s |
| 21:58 | <svl> | http://www.bradthegame.com/ |
| 21:58 | <svl> | (some nsfw text a couple of clicks in) |
| 21:59 | <virtuelv> | neat idea |
| 22:00 | <virtuelv> | I still enjoy typing blindly as well, though |
| 22:03 | <svl> | Also http://www.versificator.co.uk/hamlet/ and http://www.geocities.com/Vienna/2662/highmess.html |
| 22:05 | <svl> | (and undoubtedly hundreds of others, but those are examples of the main three ways of controlling a game; plus of course most classics like zork and h2g2 have been ported to javascript and java and flash and ... implementations) |
| 22:27 | <sicking> | Hixie, ping |
| 22:47 | <Hixie> | sicking: pong |
| 22:50 | <sicking> | Hixie, so, this whole feedback from DOM into parser thing |
| 22:50 | <Hixie> | yeah |
| 22:50 | <sicking> | Hixie, what is the purpose of it? Why do we need to do different things depending on if there are children or not? |
| 22:51 | <Hixie> | it's residual style handling |
| 22:51 | <sicking> | Hixie, i tried to make some tests in the HTML5 livedom thing, but i couldn't figure out when having children made a difference |
| 22:51 | <sicking> | Hixie, do you have an example? |
| 22:51 | <Hixie> | let's see if i can construct one |
| 22:59 | <Hixie> | sicking: http://software.hixie.ch/utilities/js/live-dom-viewer/saved/5 |
| 23:00 | <Hixie> | i guess we could remove the check |
| 23:00 | <Hixie> | and just always clone |
| 23:00 | sicking | is still looking at example |
| 23:01 | <sicking> | Hixie, i need to look at that in henris HTML5 dom-live-viewer right? |
| 23:01 | <Hixie> | wait, hsivonen was asking for a flag for whether it had parser-inserted children |
| 23:01 | <Hixie> | sicking: or webkit |
| 23:02 | <Hixie> | sicking: the AAA in HTML5 is the same as webkit's residual style handling (so long as you only have one block) |
| 23:02 | <Hixie> | hsivonen: when would that algorithm ever _not_ have parser-inserted children? |
| 23:02 | <sicking> | Hixie, "the AAA"? |
| 23:02 | <Hixie> | Adoption Agency Algorithm. residual style in html5. |
| 23:02 | <sicking> | ah |
| 23:05 | <Hixie> | (from the spec: "Because of the way this algorithm causes elements to change parents, it has been dubbed the "adoption agency algorithm" (in contrast with other possibly algorithms for dealing with misnested content, which included the "incest algorithm", the "secret affair algorithm", and the "Heisenberg algorithm").") |
| 23:05 | <Hixie> | heisenberg's = mozilla, secret affair = opera, incest = ie |
| 23:05 | <sicking> | yeah, i know what the algorithm does |
| 23:05 | <sicking> | what still confuses me is when it makes a difference which of the two forks it uses |
| 23:06 | <sicking> | i.e. the has-children and don't-have-children forks |
| 23:06 | <Hixie> | i'm actually unsure as to when you can ever not have children without scripting |
| 23:06 | <Hixie> | and why we care about that case |
| 23:06 | <Hixie> | i'm thinking maybe we can remove the if statement altogether |
| 23:06 | <sicking> | that would be nice :) |
| 23:07 | <Hixie> | i hate this algorithm, i only barely understood it when i was writing it and implementing it |
| 23:07 | <sicking> | since i won't have to challange you to a fight of doom over the feedback |
| 23:07 | <Hixie> | hah |
| 23:08 | <Hixie> | i'm gonna remove the if statement and see if anyone complains. |
| 23:08 | <sicking> | haha |
| 23:08 | <sicking> | sold! |
| 23:08 | <sicking> | i'm still not sure how to detect a difference between the two forks |
| 23:09 | <sicking> | but i don't care as long as this thing goes :) |
| 23:09 | <Hixie> | oh, wait... |
| 23:09 | Hixie | studies further |
| 23:10 | <sicking> | Hixie, in your example, if i remove the script, the DOM still looks the same |
| 23:11 | <Hixie> | yeah i was wrong |
| 23:11 | <Hixie> | here: |
| 23:11 | <Hixie> | it's the difference between: |
| 23:11 | <Hixie> | <!DOCTYPE html><i><em><div></i>xxxx |
| 23:11 | <Hixie> | <!DOCTYPE html><i><em>xxxx<div></i>xxxx |
| 23:11 | <Hixie> | in the former case, html5 just moves the original <em> down |
| 23:11 | <Hixie> | in the latter case, it clones it |
| 23:12 | <Hixie> | (the <div> doesn't count as being a child of the <em> because it's removed earlier in the algorithm) |
| 23:12 | <sicking> | so we're trying to avoid creating the extra <em>? |
| 23:12 | <Hixie> | yeah |
| 23:13 | sicking | start to warm up the fight-of-doom arena |
| 23:13 | <Hixie> | it becomes a big deal when you have a lot of missing end tags (e.g. a lot of <font> with no </font>) iirc |
| 23:13 | <sicking> | right |
| 23:13 | <Hixie> | the problem with keeping track of whether the parser inserted any nodes into each node in the stack is that that's actually not enough |
| 23:13 | <Hixie> | you have to keep track of how many children |
| 23:13 | <sicking> | why is there always an 'i' outside the <div> though? |
| 23:14 | <Hixie> | (0 or 1 or 2) |
| 23:14 | <Hixie> | 2+ rather |
| 23:14 | <Hixie> | the element being closed is always cloned because not doing so would make the algorithm orders of magnitude more complex still |
| 23:15 | <Hixie> | iirc |
| 23:15 | <sicking> | ok |
| 23:15 | <sicking> | that's probably why i failed to figure this out last time |
| 23:16 | <Hixie> | yeah this is definitely one of the most obscure parts of the spec |
| 23:16 | <Hixie> | easily on par with xbl2 in terms of headaches |
| 23:16 | <sicking> | actually, that doesn't seem right either |
| 23:16 | <sicking> | <!DOCTYPE html><i><em><em><em><div></i>xxxx |
| 23:16 | <sicking> | that clones all but the inner most <em> |
| 23:16 | <sicking> | so if you have lots of opened <font>s, you'll still clone all but the last it looks like |
| 23:16 | <sicking> | unless henri has a bug |
| 23:19 | <sicking> | Hixie, and safari seems to clone them all |
| 23:19 | <sicking> | though maybe we're trying to do better than safari? |
| 23:22 | <Hixie> | hm, maybe it's not an optimisation |
| 23:22 | <Hixie> | yeah maybe we should simplify this by not removing the furthest block node from its parent and then just not checking for children |
| 23:22 | <Hixie> | and always cloning |
| 23:23 | <sicking> | Hixie, my first question is what we're trying to accomplish with the check :) |
| 23:24 | <sicking> | Hixie, optimizting away empty element i suspect is actually worth it |
| 23:24 | <Hixie> | yeah, but as you point out, we're not doing it enough |
| 23:24 | <Hixie> | if we want to do it we should do a proper job |
| 23:24 | <Hixie> | which is a whole other problem |
| 23:24 | <sicking> | yup |
| 23:24 | <sicking> | Hixie, i do know that we (=gecko) suffer, or at least used to suffer, on pages with lots of opened-but-not-closed <font>s |
| 23:25 | <Hixie> | yeah |
| 23:25 | <sicking> | but we might never have considered it common enough to fix |
| 23:25 | <Hixie> | ok i've changed it to not do the check |
| 23:26 | <Hixie> | but i think we might want to make a more general fix at one point |
| 23:26 | <Hixie> | to get rid of the empty nodes |
| 23:26 | <sicking> | agreed |
| 23:27 | <sicking> | at which point i'll book the DOOM conference room so we can discuss how :) |
| 23:28 | <sicking> | Hixie, try "<!DOCTYPE html><i><font><font><font><font><div>w</div><div>f</div>" in gecko now |
| 23:28 | <sicking> | Hixie, we create loads of elements |
| 23:28 | <sicking> | Hixie, way more than HTML5 does |
| 23:30 | <Hixie> | yeah, html5 doesn't move all the nodes to inside the blocks |
| 23:30 | <Hixie> | that isn't affected by the change just now afaict |
| 23:31 | <sicking> | funny thing is, if you just close one <font> after the last div, we dramatically change what we do |
| 23:31 | <sicking> | which is not what i would have guessed |
| 23:31 | <sicking> | right, html5 deals with this stuff dramatically different |
| 23:32 | <Hixie> | the biggest problem with gecko is that what it does differs based on packet boundaries |
| 23:33 | <sicking> | right |
| 23:33 | <sicking> | the lookahead is bad |
| 23:33 | <Dashiva> | "With namespaces, users pay while spec authors get off scott free for their inability to do the hard work of writing specs and forging agreement." |
| 23:33 | <sicking> | i'm not at all suggesting we do what gecko does |
| 23:34 | <sicking> | just saying that dealing with lots of unclosed <font>s might not be critically important to do |
| 23:34 | <sicking> | the packed boundry thing could be fixed in gecko, and still keep the lookahead |
| 23:35 | <Philip`> | Dashiva: But with distributed extensibility, *everyone* is a spec author |
| 23:35 | <sicking> | but we'd still produce much more bloated DOMs than the current spec |
| 23:36 | <sicking> | The only real advantage in what we do is that we can implement a real SAX API even for malformed documents |
| 23:37 | <Hixie> | sicking: yeah, the spec might be good enough at this point |
| 23:38 | <Hixie> | sicking: well, you still can't do a real SAX API for <table>s, right? |
| 23:38 | <Hixie> | or for <html> and <body> nodes |
| 23:39 | <Hixie> | ok bbiab |
| 23:41 | <sicking> | Hixie, ah, yeah, you're right |
| 23:45 | <Lachy> | Hixie, re the :enabled and :disabled issue, are you now going to make them match hidden inputs based on the changes to Selectors? |
