| 00:45 | <astearns> | Buttons (and other form elements) are non-replaced elements in the current spec. Did they used to be replaced elements? |
| 00:46 | <jamesr> | what on earth is SVG <script>? |
| 00:46 | <jamesr> | just <script> in the SVG namespace? |
| 00:50 | <heycam> | yes |
| 00:51 | <heycam> | but slightly different, of course! |
| 00:51 | <heycam> | :) |
| 00:51 | <heycam> | it uses xlink:href, but we're moving away from xlink for SVG2 |
| 00:54 | <Dashiva> | astearns: Not as far as I know |
| 00:54 | <jamesr> | what's the difference between xlink:href and href? |
| 00:55 | <heycam> | nothing practical |
| 00:56 | <Dashiva> | Except being different |
| 00:56 | <TabAtkins> | It's in the xlink namespace. |
| 00:57 | <heycam> | can't remember for <script>, but at least for <a> you have the SVG DOM kind of DOM accessors |
| 00:57 | <heycam> | like a.href.baseVal |
| 01:16 | <TabAtkins> | Okay, MathML only collides with "<image>". |
| 01:16 | <TabAtkins> | SVG collides with <a>, <font>, <image>, <script>, <style>, and <title>. |
| 01:17 | <TabAtkins> | <a>, <script>, and <style> should just unify; that's pretty easy to argue. |
| 01:18 | <TabAtkins> | <image> is roughly the same. Unification is possible. |
| 01:19 | <TabAtkins> | <font> and <title> very clearly can't unify. |
| 01:19 | <TabAtkins> | Darn. |
| 01:20 | <TabAtkins> | MathML's <image> is completely different. |
| 01:21 | <TabAtkins> | Possibly SVG2 could switch to <img>, leaving <image> for MathML. |
| 01:21 | <heycam> | well |
| 01:21 | <TabAtkins> | And the magical "new Element()" function could just switch based on that. |
| 01:21 | <heycam> | that's not going to help for existing content |
| 01:21 | <heycam> | or is this only for SVG-in-text/html? |
| 01:22 | <TabAtkins> | I dunno. |
| 01:26 | <TabAtkins> | heycam: At the very least, this would help with whatever-in-HTML. But it would be great if we could drop namespaces in general for SVG2 and MathML. |
| 01:27 | <heycam> | "in general" is a big ask. :) would it be XML but all elements in the null namespace? |
| 01:28 | <TabAtkins> | That's one possibility. |
| 01:29 | <TabAtkins> | (I'm shooting for "in general" because I don't want inconsistance between SVG-in-HTML and SVG-in-HTML-in-SVG. |
| 01:29 | <heycam> | I agree that inconsistency would be bad |
| 01:36 | <TabAtkins> | heycam: I know that maintaining backwards-compat is a goal for SVG2 (though I'm unsure how strongly you guys feel about it), but if we dropped that, made some breaking changes for SVG2, and then declared that SVG2 and HTML were BFFs, that may help some of the integration issues. |
| 01:38 | <heycam> | TabAtkins, I think it's worth raising as a possibility, anyway |
| 01:48 | <Hixie> | if you set an expiry date on a resource fetched over http, the browser won't hit the network when using it right? |
| 01:51 | <Hixie> | > I wasn't very clear when I say "the main page should not be cached". |
| 01:51 | <Hixie> | > I was saying, we should still keep the main page cached, |
| 01:51 | <Hixie> | ... |
| 01:58 | <othermaciej> | TabAtkins: sag:image does have some significantly different attributes and behavior compared to html:img |
| 01:58 | <othermaciej> | not sure it is practical |
| 01:59 | <othermaciej> | I could imagine giving alternate non-colliding names for the elements that do collied |
| 01:59 | <TabAtkins> | othermaciej: It's got x, y, and xlink:href, obviously, and the SVG DOM. What else is going on with it? |
| 02:00 | <TabAtkins> | And then the various SVG-specific global attributes. |
| 02:00 | <TabAtkins> | (Which x and y might qualify as?) |
| 02:01 | <othermaciej> | in WebKit, SVG image generates a completely different kind of render object |
| 02:01 | <TabAtkins> | But in any case, the unification was from the point of (a) merging namespaces, and (b) making "new Element('foo')" work. |
| 02:01 | <othermaciej> | its layout behavior is not the same |
| 02:01 | <othermaciej> | new Element('foo') seems like a half-baked idea to me |
| 02:02 | <TabAtkins> | So having <svg><image/></svg> or new Element('image') make an SVG image, while <html><image /></html> makes an HTML img, is probably okay. |
| 02:02 | <othermaciej> | I do not see the advantage over new FooElement, and it is also kind of a lie, because normally it will not give you a direct instance of Element |
| 02:02 | <othermaciej> | I can't think of any other case in JS APIs where new FooBar returns something other than a direct instance of the FooBar class |
| 02:02 | <TabAtkins> | That's a good point. |
| 02:03 | <Hixie> | othermaciej: Image? :-) |
| 02:03 | <TabAtkins> | That's because the HTML element is misspelled. ^_^ |
| 02:03 | <Hixie> | (new Image() returns an HTMLImageElement which is an <img>) |
| 02:03 | <othermaciej> | Image and Option are exceptions in that they are aliases for what was later decided were the true names of those things |
| 02:03 | <Duke_> | :P |
| 02:04 | <othermaciej> | and I suppose Audio follows their pattern |
| 02:04 | <Hixie> | yeah i was gonna say, Audio was added knowingly :-) |
| 02:04 | <othermaciej> | unfortunately I don't think that pattern can be extended indefinitely because defining P, I, B and U in the global namespace would surely be a breaking change |
| 02:05 | <othermaciej> | however, having a constructor secretly be a factory function that produces instances of different things seems like very poor design to me |
| 02:05 | <othermaciej> | so I would still prefer new PElement to new Element('p') |
| 02:05 | <TabAtkins> | $('<p>') |
| 02:06 | <TabAtkins> | That looks remarkably like an emoji. |
| 02:06 | <othermaciej> | and I don't think the slightly more verbose new HTMLParagraphElement is all that bad |
| 02:06 | <othermaciej> | I don't really dig strings as APIs |
| 02:06 | <jamesr> | what about to make a raw element? new ElementElement()? |
| 02:06 | <Hixie> | speaking of emojis |
| 02:06 | <TabAtkins> | Nah, I don't either, except that it lets you set attributes on it as well. |
| 02:06 | <Hixie> | have you guys checked out the new Apple Colour Emoji font? |
| 02:06 | <jamesr> | or do you not do that? |
| 02:06 | <TabAtkins> | jamesr: new Element('') |
| 02:06 | <othermaciej> | but if you wanted something that short you could define a $p() function |
| 02:06 | <othermaciej> | the desire to make a raw element is rare enough that I don't care |
| 02:06 | <othermaciej> | Element intrinsically needs a tag name to exist |
| 02:07 | <othermaciej> | I'm satisfied with document.createElement() covering that use case |
| 02:07 | <Hixie> | U+1F533 WHITE SQUARE BUTTON in Apple's font is a round purple blob. |
| 02:07 | <TabAtkins> | If we can fix the constructors to allow setting attributes and content immediately, I'd be a lot happier. |
| 02:07 | <othermaciej> | almost always you want an HTMLElement anyway, not an Element |
| 02:07 | <TabAtkins> | Hixie: wut |
| 02:07 | <Hixie> | and U+1F537 LARGE BLUE DIAMOND is a gray diamond. |
| 02:07 | <jamesr> | but wouldn't new HTMLElement() make an <html> in that API? |
| 02:08 | <zewt> | since when do unicode characters specify color |
| 02:08 | <Hixie> | since the emoji set! |
| 02:08 | <othermaciej> | currently you make one with new HTMLHtmlElement |
| 02:08 | <othermaciej> | so I would expect new HtmlElement to be the new-style version, assuming we even cared |
| 02:08 | <zewt> | i expect almost no font renderers would even be capable of that, since color and glyphs should be fully orthogonal |
| 02:08 | <othermaciej> | s/currently/if we followed the interface names literally/ |
| 02:08 | <Hixie> | zewt: Mac OS X Lion has a colour truetype font |
| 02:09 | <Hixie> | zewt: it's pretty sweet |
| 02:09 | <othermaciej> | since currently you obviously use createElement if you use anything |
| 02:09 | <othermaciej> | if we did want a tag-dispatching version I'd do Element.create(tagName, attributes…) |
| 02:09 | <othermaciej> | that seems more obviously a factory method, doesn't lie and claim to be a constructor |
| 02:10 | <Hixie> | Element.create() would be nice |
| 02:10 | <othermaciej> | and Element.create('p') is not materially more verbose than new Element('p') |
| 02:10 | <Hixie> | Element.create(tagName /* picks namespace magically */, {attributes}, [children...]) |
| 02:10 | <othermaciej> | and if we cared we could add Element.createNS() |
| 02:11 | <othermaciej> | that's a neat version, being able to include the children |
| 02:11 | <TabAtkins> | We dont' care, and shouldn't do it. >_< |
| 02:11 | <TabAtkins> | Mix in "new Text('foobar')" and I'm sold. |
| 02:12 | <TabAtkins> | Or just make strings auto-wrap themselves into textnodes when inserted into the DOM. |
| 02:12 | <othermaciej> | it could be an escape hatch for cases of HTML/SVG/MathML namespace conflicts |
| 02:12 | <othermaciej> | or give .create() an optional fourth namespace attribute |
| 02:12 | <TabAtkins> | othermaciej: I'd rather resolve the conflicts. ^_^ |
| 02:12 | <Hixie> | i really think we should just make strings work |
| 02:12 | <othermaciej> | I dunno how practical it is |
| 02:12 | <Hixie> | the only mathml conflict, iirc, is <image> |
| 02:12 | <Hixie> | and that's a three-way conflict |
| 02:13 | <othermaciej> | at the very least all these elements would need magical rules about what renderer is created depending on whether they arein SVG or HTML content |
| 02:13 | <TabAtkins> | Hixie: That's right, I just checked. |
| 02:13 | <zewt> | still, unicode specifying colors on glyphs just seems out of scope |
| 02:13 | <Hixie> | (i had to look all this stuff up when doing the parser) |
| 02:13 | <othermaciej> | hmm, I suppose .create() can handle namespaces by providing xmlns in the list of attributes |
| 02:13 | <Hixie> | for the dom api, i'd just make 'image' map to the svg one |
| 02:13 | <Hixie> | 'a' is more troublesome |
| 02:13 | <othermaciej> | it won't actually affect descendants unless you use innerHTML on it |
| 02:13 | <Hixie> | what are the other conflicts? |
| 02:14 | <TabAtkins> | a, font, image, script, style, title |
| 02:14 | <TabAtkins> | a, script, and style should be unified. |
| 02:14 | <TabAtkins> | font and title pretty much can't be. |
| 02:14 | <Hixie> | font=>svg, image=>svg, script=>html, style=>html |
| 02:14 | <Hixie> | (for this api) |
| 02:14 | <Hixie> | a and title are tough |
| 02:14 | <Hixie> | title kinda doesn't matter |
| 02:14 | <TabAtkins> | How often do you actually create <title> elements in HTML via script> |
| 02:15 | <Hixie> | nobody generating dynamic stuff this way is going to use either the html or svg one |
| 02:15 | <othermaciej> | you can pick either default and use xmlns in the attributes dictionary as an escape hatch |
| 02:15 | <Hixie> | a is a bigger problem |
| 02:15 | <TabAtkins> | (Probably SVG should drop <title> in favor of global @title.) |
| 02:15 | <Hixie> | how bad is merging svg <a> and html <a>? |
| 02:15 | <Hixie> | can we make svg just support the html <a> where appropriate? |
| 02:15 | <othermaciej> | Elemet.create('a', {'xmlns': svgNS, 'xlink:href', '….'}, new Text('xxx')) |
| 02:15 | <TabAtkins> | othermaciej: Or continue to ignore NS, but have a fourth paramter that just accepts "html | svg | mathml", with possible future extensions. |
| 02:16 | <Hixie> | or just have the tagname be optionally prefixed by "svg:" to force it into svg |
| 02:16 | <TabAtkins> | othermaciej: Since the NS itself is dumb, and we just need to know what language you're talking about. |
| 02:16 | <Hixie> | and then have sane defaults |
| 02:16 | <othermaciej> | since NS is rarely needed in the first place it would be nice not to be in the business of defining a translation table |
| 02:16 | <othermaciej> | just pick sane defaults and have a way to override |
| 02:16 | <Hixie> | yeah maybe just default to html and have svg: and mathml: prefixes |
| 02:16 | <TabAtkins> | I don't see why the override mechanism should continue to be difficult. |
| 02:16 | <Hixie> | or svg: and m: |
| 02:17 | <othermaciej> | I like the idea of Element.create() |
| 02:17 | <TabAtkins> | But predefined prefixes work too. |
| 02:17 | <othermaciej> | I should post to public-webapps |
| 02:17 | <Hixie> | Element.create('svg:a', {href: 'example.html'}, ['Click Me! Yay bad link text!']); |
| 02:18 | <othermaciej> | your idea is that strings work in the children list? |
| 02:18 | <Hixie> | yeah |
| 02:18 | <othermaciej> | I kinda like that |
| 02:18 | <othermaciej> | the children list could just be rest args instead of an array |
| 02:18 | <othermaciej> | would save the need to allocate an array to call this API |
| 02:18 | <Hixie> | i'd like strings to implictly create text nodes anywhere nodes are accepted |
| 02:18 | <Hixie> | rest args works for me except that often you'll have an array |
| 02:18 | <Hixie> | so we'd need to support both |
| 02:19 | <othermaciej> | Element.create('svg:a', {href: 'example.html'}, 'Click Me! Yay bad link text!', Element.create('svg:image'….)) |
| 02:19 | <Hixie> | (think of <ol> where your child list might be an array to which you .push()ed a lot of <li>s) |
| 02:19 | <othermaciej> | (actually that may be illegal SVG but you get the idea) |
| 02:19 | <othermaciej> | you can let arrays be in the rest args and unpack them to one level |
| 02:19 | <othermaciej> | then you can have a fixed list followed by an array |
| 02:19 | <Hixie> | lgtm |
| 02:19 | <othermaciej> | or array in the middle |
| 02:20 | <Hixie> | ship it |
| 02:21 | <TabAtkins> | That sounds really good. |
| 02:23 | <TabAtkins> | othermaciej: You gonna write it up? If you don't have time, I'll do it. |
| 02:24 | <othermaciej> | writing it up briefly |
| 02:24 | <TabAtkins> | kk |
| 02:30 | <othermaciej> | sent |
| 07:21 | <hsivonen> | :-( looks like manu-db filed the absence of some RDFa misfeatures as bugs against Microdata on Saturday |
| 07:23 | <Hixie> | feedback isn't bad |
| 07:43 | <hsivonen> | Hixie: in other news, I find it strange that you didn't just go ahead and fix http://www.w3.org/Bugs/Public/show_bug.cgi?id=12267 as a no-brainer |
| 07:43 | <annevk> | I guess the other thing we could do is have new Element("html", "img") |
| 07:43 | <annevk> | well |
| 07:43 | <annevk> | that might as well be new HTMLElement("img") |
| 07:43 | <annevk> | so I guess that's not the way to go |
| 07:44 | <hsivonen> | annevk: ooh. radical. are factories no longer fashionable for the DOM? |
| 07:44 | <Hixie> | hsivonen: that bug is the furthest thing from a no-brainer that i've had to deal with all year |
| 07:44 | <Hixie> | hsivonen: you sure you got the right number? |
| 07:44 | <annevk> | hsivonen, yeah, somewhat |
| 07:44 | <annevk> | hsivonen, e.g. we introduced new Event(...) |
| 07:46 | <hsivonen> | Hixie: yes, I got the right number |
| 07:46 | <Hixie> | hsivonen: it's not a no-brainer. i don't think anyone else has ever suggested that it is. |
| 07:46 | <Hixie> | i mean just look at roc's last comment. |
| 07:48 | <annevk> | Oh, there's a mailing list proposal for Element.create() |
| 07:50 | <hsivonen> | annevk: is proposing factory patterns for stylistic similarity with legacy? |
| 07:51 | <hsivonen> | annevk: now that we no longer believe in Web DOM having to work with separate interfaces and multiple independent backing implementations in the same process, is there a technical reason for needing factories anymore? |
| 07:59 | <zcorpan> | dudes, did you check prior art before designing element creation api? |
| 07:59 | <zcorpan> | prior art being jquery and other libraries |
| 08:00 | <Hixie> | i assumed anne would do that, personally |
| 08:03 | <annevk> | hsivonen, the reason given was that it would not actually create an object of type Element |
| 08:03 | <hsivonen> | annevk: oh, right |
| 08:04 | hsivonen | needs to wake up properly |
| 08:04 | <annevk> | me too |
| 08:04 | <hsivonen> | so Alice and Bob seem to come from A and B. What's the story behind Nadia and Dirk? |
| 08:05 | <annevk> | I'm sort of afraid people are only going to implement new features of DOM Core and not care much about aligning on the legacy bits |
| 08:05 | <annevk> | That's what happened with XMLHttpRequest |
| 08:07 | <hsivonen> | I had assumed Nadia and Dirk were just the TAG's favorite scenario actors, but I just saw Nadia show up in an accessibility scenario, too |
| 08:08 | <hsivonen> | are Nadia and Dirk from some seminal book about scenarios that I haven't read? |
| 08:09 | <annevk> | Maybe everyone just copies from each other? |
| 08:23 | <Hixie> | othermaciej: since nodes get adopted silently now anyway, you basically never need to explictly create a node in another doc |
| 08:24 | <othermaciej> | Hixie: hmm, good point |
| 08:24 | <annevk> | WebUSB https://bugzilla.mozilla.org/show_bug.cgi?id=674718 |
| 08:25 | <annevk> | I wonder if that is the right abstraction layer. Does not seem likely USB will be around forever... And there's lots of other ways to connect hardware too... |
| 08:25 | <Hixie> | using the word "steal" in the initial description seems like a bad idea in that context ;-) |
| 08:25 | <Hixie> | (given the many security implications here) |
| 08:33 | <hsivonen> | hmm. the bug description doesn't go in too much detail about use cases, so it's hard to assess if the abstraction is right |
| 08:49 | <annevk> | Hixie, so if we do Element.create() I guess we should have appendChild("string"), insertBefore("string", child) etc. as well for consistency? |
| 09:34 | <hsivonen> | when I saw the URL http://www.html5la.com/ I thought someone had set up an analog of MPEG-LA |
| 09:59 | <annevk> | is there a bug on Gecko on implementing Event constructors? |
| 09:59 | <smaug____> | IIRC no |
| 09:59 | <annevk> | I'll file one |
| 10:00 | <annevk> | would be nice if they got traction so init*Event can die in a fire |
| 10:00 | <smaug____> | I won't implement them before I know how to handle mouse events |
| 10:00 | <annevk> | What is the problem with mouse events? |
| 10:01 | <smaug____> | well, which all properties can be initialized |
| 10:01 | <smaug____> | and to which document/window is the event bound to |
| 10:01 | <smaug____> | since that affects to pageX/Y values |
| 10:02 | <annevk> | all properties can be initialized |
| 10:02 | <hsivonen> | doesn't JS |new| always occur in the context of a script global object i.e. window? |
| 10:02 | <annevk> | pageX/Y have defaults? |
| 10:02 | <annevk> | smaug____, same as for Image() |
| 10:03 | <smaug____> | if you initialize only clientX/Y, pageX/Y are calculated from those values |
| 10:03 | <hsivonen> | (I suppose there might be uses for a parent document creating event objects associatied with a child window, though) |
| 10:03 | <annevk> | Workers might work differently |
| 10:03 | <annevk> | smaug____, wow weird |
| 10:04 | <smaug____> | annevk: indeed |
| 10:04 | <smaug____> | and the web relies on that behavior |
| 10:04 | <annevk> | smaug____, for init*Event it does |
| 10:04 | <annevk> | smaug____, event constructors can work more sanely |
| 10:05 | <smaug____> | that is true |
| 10:05 | <annevk> | DOM 3 Events has nothing on this of course |
| 10:05 | <smaug____> | though, it would create yet another inconsistent API |
| 10:05 | <smaug____> | DOM 3 events doesn't specify pageX/Y |
| 10:05 | <annevk> | right |
| 10:05 | <smaug____> | it should be in cssom-view if somewhere |
| 10:06 | <annevk> | cssom-view just had MouseEvent as a temporary stopgap |
| 10:06 | <annevk> | mouse events should really be defined in one place |
| 10:07 | <smaug____> | so should Event, but somehow it is now in two places :p |
| 10:07 | <annevk> | yeah, I think DOM 3 Events should drop it and rename itself to UI Events or some such |
| 10:08 | <annevk> | I think keeping event constructors consistent with themselves is worth more than keeping them consistent with init*Event |
| 10:08 | <annevk> | especially with weird legacy proprietary behavior and especially since with event constructors you can set pageX/Y yourself |
| 10:09 | <smaug____> | pageX/Y behavior is not proprietary |
| 10:09 | <annevk> | which spec? |
| 10:09 | <smaug____> | I mean,it is, IIRC, implemented by all |
| 10:10 | <smaug____> | though, haven't tested that for years |
| 10:10 | smaug____ | should go back to think about mutation event replacement |
| 10:11 | <annevk> | sure, it needs to become part of a spec |
| 10:11 | <smaug____> | there still isn't a good enough proposal :( |
| 10:11 | <annevk> | for mutation events? |
| 10:11 | <smaug____> | yeah |
| 10:11 | <smaug____> | er, proposal for the replament |
| 10:11 | <smaug____> | replacement |
| 10:12 | <annevk> | how does the replacement Mozilla initially proposed deal with Comment/ProcessingInstruction modifications? |
| 10:13 | <annevk> | kind of a pity progress on mutation event replacement halted a bit with so much conflicting requirements |
| 10:13 | <smaug____> | comment would be handled like textnodes |
| 10:13 | <annevk> | I can add any agreed on proposal to DOM Core at this point I think |
| 10:14 | <smaug____> | unfortunately there isn't such yet |
| 10:16 | <smaug____> | in my initial implementation PI would be handled like textnode too |
| 10:17 | <annevk> | that would make sense I think |
| 10:17 | <annevk> | treat PI as CharacterData except it does not implement any of its methods |
| 10:18 | <smaug____> | I wonder if PI could inherit CharacterData |
| 10:18 | <annevk> | yes it can, but you was one of those arguing against that |
| 10:18 | <annevk> | were* |
| 10:19 | <smaug____> | that is how it is implemented in Gecko, we just expose different interface |
| 10:19 | <smaug____> | annevk: I was? |
| 10:19 | <smaug____> | that is possible |
| 10:19 | <annevk> | http://www.w3.org/Bugs/Public/show_bug.cgi?id=12205 |
| 10:20 | <annevk> | oh no |
| 10:20 | <annevk> | you were not |
| 10:20 | <smaug____> | http://www.w3.org/Bugs/Public/show_bug.cgi?id=12205#c10 |
| 10:21 | <JirkaK> | hi, just noticed <ol reversed> in spec. What are usescases for this bizzare attribute? |
| 10:21 | <annevk> | we could make it inherit and add another interface between CharacterData and Text/Comment |
| 10:22 | <annevk> | that exposes the modification methods |
| 10:22 | <Ms2ger> | JirkaK, lists that go backwards |
| 10:23 | <hsivonen> | https://plus.google.com/109022289777044064966/posts/j43r9owxqry ...and the violent language like "punch" enters the discussion about Microdata and RDFa |
| 10:24 | <Ms2ger> | Wow, that took long |
| 10:24 | <JirkaK> | Ms2ger, well, yes. but is it worth new attribute? It's impossible to render such list progressively and feels more like styling thing anyway |
| 10:24 | <hsivonen> | s/ the// |
| 10:25 | <Ms2ger> | See http://blog.whatwg.org/reverse-ordered-lists, for example |
| 10:26 | <heycam> | Hixie, s/pedancy/pedantry/ in your recent commit message, no? :) |
| 10:26 | <Zeddy> | i have a problem conserning the cache manifest, im building a webpage for mobile devices which lets the user download all the required files for offline use, but the content can be in different languages depending on the url mapviewer.php?lang=1 ?lang=2 and so on. This results in double downloading if the user visits both languages, which obviously is the way it should work.. but i would like |
| 10:26 | <Zeddy> | to delete all cached files when the user visits another language, otherwise the amount of cached data will exeed the limitations that mobile browsers have |
| 10:26 | <Zeddy> | http://www.kokonniemi.fi/guide/index.php/view/map/1/1/1 <- finnish language |
| 10:27 | <Zeddy> | http://www.kokonniemi.fi/guide/index.php/view/map/1/2/1 <- english language |
| 10:27 | <Zeddy> | visit the first and you have 1.4 Mb cached, then visit the second and you have 2.8 Mb cached |
| 10:28 | <Ms2ger> | And they're all comments from foolip |
| 10:28 | <erlehmann> | Zeddy, how can it be SO MUCH? |
| 10:28 | <erlehmann> | i mean … text? |
| 10:29 | <Zeddy> | what text? |
| 10:31 | <JirkaK> | Of course, I did some research on <ol reversed>. I think that's more harmful then useful, it's not backward compatible -- it will show wrong numbers in old browsers. I'm thinking about filling bug for removing it. But before that I would like to know whether I haven't miss anything. |
| 10:38 | <hsivonen> | I suppose social norms say that I should be OK with Nathan announcing that he's going to punch people who ask about use cases when he says it right after I've asked him about use cases. |
| 10:39 | <hsivonen> | yet, I still think it's sad that discussion has come to this (though I don't really expect him to hit me if within reach) |
| 10:40 | <foolip> | hsivonen, I read it as "please don't discuss specs with me" |
| 10:40 | <foolip> | which is easy enough to not do |
| 10:40 | <annevk> | Julian even +1'd it |
| 10:41 | <foolip> | but yes, I also interpreted as a response to you, hsivonen, which is not OK |
| 10:41 | <zcorpan> | speaking of <ol reversed>, css still doesn't have an equivalent feature yet right? |
| 10:42 | <annevk> | zcorpan, don't think so |
| 10:42 | <annevk> | zcorpan, reversed counters would be pretty awful |
| 10:43 | <Ms2ger> | That makes it rather surprising the CSSWG hasn't added it yet |
| 10:46 | <annevk> | ouch |
| 10:48 | <jgraham> | hsivonen: "Violence is the last refuge of the incompetent" and all that. I took it as a sign that he can't actually justify anything he wants by reason and so should be ignored. |
| 11:01 | <annevk> | so we had Microdata and RDFa and now there's two new proposals plus a bunch of change requests? |
| 11:01 | <annevk> | 927 is the new 386 |
| 11:02 | <hsivonen> | annevk: indeed |
| 11:04 | <jgraham> | Maybe there should be a short test before you file a bug to ensure that you have read the right comics |
| 11:05 | <Ms2ger> | Would that stop them? |
| 11:05 | <jgraham> | No, but the idea of testing people on comic knowledge before they can file bugs amuses me |
| 11:05 | <hsivonen> | jgraham: see also https://plus.google.com/114490712483753086051/posts/RgEF6RBXH6D |
| 11:06 | <jgraham> | hsivonen: Heh |
| 11:12 | <foolip> | annevk, the RDF crowd have begun looking at microdata in detail, which I would consider a good thing |
| 11:12 | <annevk> | Yeah I guess |
| 11:13 | <foolip> | unless all their suggestions are rejected and they just more pissed off... |
| 11:13 | <foolip> | get |
| 11:13 | <annevk> | I'm with the last paragraph (before the addendum) of http://hsivonen.iki.fi/schema-org-and-communities/ |
| 11:15 | <foolip> | certainly, I expect that search engines would give little weight to hidden microdata and only use if it seems to match up with everything else |
| 11:23 | <annevk> | there is now some Pokemon stuff in my WHATWG spark |
| 11:23 | <foolip> | aren't sparks just google news feeds for a keyword? |
| 11:23 | <foolip> | they seem extremely useless to me |
| 11:27 | <hsivonen> | I think I saw someone saying that sparks are Semantic |
| 11:28 | <Zeddy> | did anyone perhaps answer my question while i was away? |
| 11:29 | <Ms2ger> | hsivonen++ |
| 11:31 | <foolip> | hsivonen, I think you mean this: http://semanticweb.com/thoughts-on-google-plus-the-magic-isnt-social-its-semantic_b21852 |
| 11:32 | <hsivonen> | foolip: yeah, that was what I was thinking about, though I didn't actually read the post fully. just saw the title and someone else's link to it |
| 11:33 | <foolip> | hsivonen, https://plus.google.com/102122664946994504971/posts/NRo1igagUGy |
| 11:42 | <Zeddy> | what happends if a domain has 9Mb cached, and then it starts caching something else, will the old caches be deleted when the 10Mb limit is reached or will it just stop caching the new stuff? O_oi |
| 11:42 | <zcorpan> | Hixie: regarding version history, sometimes it's handy to check the old snapshots in http://www.whatwg.org/specs/web-apps/ and do inline search |
| 11:43 | <zcorpan> | Hixie: i think you should publish such snapshots maybe once a year or so and link to them from the spec |
| 11:43 | <zcorpan> | Hixie: svn history is too much if you want to know which year canvas was added |
| 11:44 | <foolip> | zcorpan, would filtering the commit log per section solve the problem? |
| 11:45 | <zcorpan> | foolip: how would you do that? |
| 11:45 | <foolip> | zcorpan, by running the spec splitter on each commit and seeing which sections changed |
| 11:46 | <foolip> | I was working on a script to do that some time ago, but got bored |
| 11:46 | <zcorpan> | oooh. interesting |
| 11:46 | <zcorpan> | that would be pretty cool |
| 11:46 | <foolip> | It would help me not get surprised by changes to the video section that happened on seemingly unrelated commits |
| 11:46 | <zcorpan> | but i'd still want year snapshots |
| 11:46 | <jgraham> | Pretty sure I had a similar idea and got similarly bored :) |
| 11:47 | <foolip> | ok |
| 11:47 | <foolip> | I think I ended up filing bugs on the spec about how the outlining algorithm worked instead... :-/ |
| 11:47 | <foolip> | (because I tried using it) |
| 11:48 | <zcorpan> | maybe the right solution is to have the multipage versions in svn |
| 11:49 | <Ms2ger> | There are svn blame snapshots at every thousandth commit somewher |
| 11:49 | <Ms2ger> | +e |
| 11:49 | <foolip> | I just use my git conversion and bisect on that |
| 11:52 | <gsnedders> | zcorpan: So snapshots are a good idea? :P |
| 11:53 | <annevk> | they are nice for spec archeology as version history is not always great |
| 11:53 | <foolip> | would having web access to every single commit be acceptable? |
| 11:53 | <annevk> | yeah |
| 11:54 | <annevk> | I was looking into that at some point |
| 11:54 | <foolip> | I guess that we can't recreate old versions as the toolchains have changed though, the spec splitter notably |
| 11:54 | <annevk> | I guess if you ignore external dependencies changing it might be okay |
| 11:55 | <Zeddy> | i just can't believe html5 cache manifest is missing a method for clearing the cache. |
| 11:59 | <annevk> | you can use HTTP for that |
| 12:09 | <zcorpan> | gsnedders: yes |
| 12:10 | <zcorpan> | gsnedders: i don't mind if the snapshotting is automated |
| 13:07 | <MikeSmith> | AryehGregor: about bugzilla and e-mail, please ping me when you're around |
| 13:20 | <annevk> | whoa MikeSmith da man |
| 13:20 | <annevk> | MikeSmith, Traversal can be removed from http://platform.html5.org/ |
| 13:20 | MikeSmith | takes a look |
| 13:23 | <MikeSmith> | annevk: OK, removed |
| 13:23 | <annevk> | cool |
| 13:23 | <annevk> | it's in DOM Core now |
| 13:31 | <Zeddy> | :(( |
| 13:31 | <Zeddy> | html5 has failed me |
| 13:32 | <cygri> | hi foolip, around? a Live Microdata question |
| 13:32 | <foolip> | cygri, always at your service! |
| 13:32 | <cygri> | i though that this microdata snippet would produce a property: http://pastebin.com/cr2bz5b2 |
| 13:33 | <foolip> | it does, if you look at the JSON |
| 13:33 | <foolip> | perhaps you wonder why there's no Turtle? |
| 13:33 | <cygri> | no, i'm looking at the json and don't see a property |
| 13:34 | <cygri> | it shows an item with zero properties |
| 13:34 | <cygri> | at http://foolip.org/microdatajs/live/ |
| 13:34 | <foolip> | I see this: http://pastebin.com/5hDEPzK9 |
| 13:34 | <Philip`> | Maybe your browser's parser moves <meta> into the <head>? |
| 13:34 | <foolip> | cygri, are you using Firefox? |
| 13:35 | <foolip> | if so, Philip` is likely correct |
| 13:35 | <cygri> | safari. why would this make a difference? |
| 13:35 | cygri | tries other browsers |
| 13:35 | <foolip> | some browsers move <meta> during parsing |
| 13:35 | <foolip> | they shouldn't if they have an HTML5 parser though |
| 13:36 | <foolip> | not sure which "some browsers" are at the moment |
| 13:36 | <cygri> | but the markup is just form input in this case, no? |
| 13:36 | <foolip> | no, it's parsed in an iframe and then the resulting DOM is used |
| 13:37 | <foolip> | since I didn't feel like implementing an HTML parser in JavaScript... |
| 13:37 | <foolip> | and for the preview tab of course |
| 13:37 | <zcorpan> | hsivonen has already done that |
| 13:38 | <cygri> | ah i see. good point |
| 13:38 | <cygri> | it works in chrome |
| 13:38 | <cygri> | that explains it. good to know that meta in the body isn't quite safe to use yet. |
| 13:39 | <cygri> | thanks! |
| 13:39 | <foolip> | perhaps I should feature-detect this bug and warn about it |
| 13:40 | <Philip`> | Or look for any <meta itemprop>/<link itemprop> in <head>, so you only warn if the input has triggered that bug? |
| 13:41 | <cygri> | Philip`++ |
| 14:13 | <foolip> | cygri, can you try http://foolip.org/microdatajs/live/ now to see if you get a warning about the bug? |
| 14:13 | <foolip> | I'll just always warn for it, since I have no way of knowing that the author didn't actually put <meta itemprop> in <head> |
| 14:14 | <Zeddy> | Can i use Session variables in the manifest file to get dynamic manifest files but with the same URI ? |
| 14:16 | <annevk> | what are session variables? |
| 14:16 | <annevk> | if you make your manifest dynamic you lose all benefit it has |
| 14:18 | <cygri> | foolip, i can't see any change in safari |
| 14:18 | <Zeddy> | annevk, please do explain |
| 14:19 | <Zeddy> | that makes no sense whatsoever |
| 14:19 | <cygri> | foolip, but in firefox i do see the warning |
| 14:19 | <foolip> | cygri, does is <meta> moved to <head> in http://software.hixie.ch/utilities/js/live-dom-viewer/saved/1091 using safari? |
| 14:19 | <foolip> | if not, then it must be some other bug |
| 14:20 | <Zeddy> | i have a mobile guide for mobile phones, where the user can select in what language they want their guide to be in, i only want to cache the selected language data |
| 14:20 | <Zeddy> | each language can have different images and ofcourse text |
| 14:21 | <cygri> | foolip, yes the meta is inside head in safari, but i still don't see the warning |
| 14:21 | <zcorpan> | hmm. <video></video> is valid but <audio></audio> is not |
| 14:21 | <zcorpan> | wonder if that's intentional. Hixie ^ |
| 14:21 | <foolip> | cygri, very odd, I'll have to find myself a machine with safari to test |
| 14:21 | <MikeSmith> | zcorpan: yeah, I wondered about that too |
| 14:21 | <foolip> | cygri, which version? |
| 14:22 | <cygri> | foolip, Version 5.0.5 (6533.21.1) |
| 14:22 | <foolip> | ok, will check tonight |
| 14:22 | <cygri> | foolip, wait a sec, migth be a js caching thing |
| 14:23 | <foolip> | that would explain it :) |
| 14:23 | <annevk> | Zeddy, if you change the manifest the UA will fetch all its listed resources again |
| 14:26 | <cygri> | foolip, yes that's it |
| 14:26 | <cygri> | it works now |
| 14:26 | <foolip> | cygri, yay :) |
| 14:26 | <cygri> | had to blow away the cache |
| 14:26 | <foolip> | so I guess Safari 5 is the newest? |
| 14:27 | <cygri> | i think so |
| 14:27 | <foolip> | a big red warning for all, then! |
| 14:27 | <cygri> | safari's js error console actually shows this: |
| 14:27 | <cygri> | <meta> is not allowed inside <body>. Moving <meta> into the <head>. |
| 14:27 | <cygri> | i should have checked there first :-) |
| 14:28 | <foolip> | well, looks like they need to update their parser |
| 14:28 | <foolip> | file a bug :) |
| 14:39 | <jgraham> | I wonder if the html5lib tests have been updated for that case… maybe I missed it |
| 14:39 | <cygri> | foolip, i tried to file a bug. it's too hard. looks like i have to file it against webkit, and i have no idea what version of webkit safari is using. i should probably install a safari/webkit nightly build, but that's too much work right now. |
| 14:39 | <cygri> | next time i come across some safari issue, i'll figure out how to install a nightly and do the right thing, i promise ;-) |
| 14:41 | <brucel> | WHATWG types: has there been any thoughts on extending fragment indentifiers to allow CSS selectors so you can link to arbitrary places on 3rd party documents (like http://simonstl.com/articles/cssFragID.html) ? |
| 14:44 | <beverloo> | I think there was a thread about that a year ago |
| 14:44 | <annevk> | every other year there's one :) |
| 14:45 | <brucel> | does it ever go anywhere? (eg, it it worth going through the logs?) |
| 14:45 | <annevk> | nah |
| 14:45 | <brucel> | fairy nuff |
| 14:46 | <annevk> | not much interest basically |
| 14:46 | <annevk> | there's more interesting stuff to work on |
| 14:48 | <foolip> | cygri, they probably know that they're not using an HTML5 parser, it's OK |
| 14:48 | <asmodai> | annevk: Will Opera ever support more of MathML for HTML 5? |
| 14:49 | <annevk> | I don't really know much about our MathML support |
| 14:49 | <annevk> | except that I don't think we should have implemented it using some sort of CSS profile that kind of, but not really, works |
| 14:49 | <asmodai> | annevk: Seems to parse it and display it roughly, but just plain text, nothing involving a real math look |
| 14:50 | <hsivonen> | cygri: Safari 5.0.5 has the old parser |
| 14:50 | <hsivonen> | cygri: 5.1 has big internal changes, though I haven't checked if it has the HTML5 parser |
| 14:51 | <hsivonen> | cygri: 5.1 is the latest Safari release |
| 14:51 | <annevk> | asmodai, that's what I meant, yes |
| 14:51 | <asmodai> | annevk: *nod* |
| 14:52 | <asmodai> | annevk: It would be so great to have more browsers support it |
| 14:52 | <annevk> | asmodai, make sure it is Acid tested |
| 14:52 | <hsivonen> | cygri: it seems 5.1 has the same parser as Chrome 8 and newer |
| 14:52 | <asmodai> | I know one group of folks that would love to be able to use more math stuff: gamedevs |
| 14:52 | <annevk> | asmodai, or design some kind of Math test, dunno |
| 14:52 | <asmodai> | annevk: mmm |
| 14:52 | <gsnedders> | asmodai: Basically we suppotr http://www.w3.org/TR/mathml-for-css/ |
| 14:53 | <gsnedders> | *support |
| 14:53 | <jgraham> | Well ther are plenty of people who would use maths if they could |
| 14:54 | <hsivonen> | maybe Opera will allocate resources to MathML once IE adds MathML support |
| 14:54 | <asmodai> | jgraham: Oh of course |
| 14:54 | <asmodai> | jgraham: Just that I have seen them pick up on web stuff quickly. |
| 14:55 | <asmodai> | A lot of them are looking at how to use recent web tech to replace some tools/utilities they have now that only run on desktop. |
| 14:55 | <jgraham> | hsivonen: Maybe. It's not exactly something that people are clamouring for at the moment so it's not hard to see why it doesn't get priority |
| 14:55 | <hsivonen> | mathml support looks like burden for browsers as long as e.g. wikipedia doesn't use it |
| 14:56 | <jgraham> | I think people are pretty happy to do maths either as images (becase c+p doesn't work all that well anyway) or using e.g. MathJAX |
| 14:56 | <asmodai> | You see MathJAX pop up all over the place, yea |
| 14:56 | <jgraham> | e.g. I think Kahn Acadamy uses the latter |
| 14:56 | <gsnedders> | hsivonen: Yeah, which does make the CSS-expressable subset seem like a reasonable thing to impl |
| 14:57 | <asmodai> | although I like the layout stuff you can do with MathML |
| 14:57 | <jgraham> | gsnedders: Well it is easy. Which is nice. It kinda asucks compared to a proper implemntation, which is bad. |
| 14:57 | <asmodai> | e.g. some of the things I got in http://www.in-nomine.org/~asmodai/3d.html at the beginning |
| 14:58 | <hsivonen> | MathML is the kind of thing that is really hard to justify to someone who is making decisions about how to allocate paid effort in browser development |
| 14:58 | asmodai | makes a mental note to expand on that page again in the coming weeks. |
| 14:58 | <gsnedders> | jgraham: Indeed. |
| 14:59 | <hsivonen> | MathML needs to be contributed by someone who is on a math-specific mission and manages to get the patches accepted before someone stops to think about the ongoing maintenance of that code |
| 14:59 | <gsnedders> | hsivonen: Well, unless the paid effort is paid for, which I guess is why we have our impl |
| 14:59 | <cygri> | hsivonen, ok thanks, that's great to know. i'm still on os x 10.6.7, seems like they don't make safari 5.1 available for that. |
| 14:59 | <hsivonen> | gsnedders: do you mean an Opera embedding customer commissioned the CSS level support? |
| 15:00 | <gsnedders> | hsivonen: I don't know, but I guess so, because as you say, it's hard to justify paid development of. |
| 15:00 | <hsivonen> | cygri: Safari 5.1 is available for 10.6.8 |
| 15:00 | <hsivonen> | cygri: and 10.6.8 is available for 10.6.7 :-) |
| 15:01 | <cygri> | hsivonen, yes i know. i'm one reboot away from that |
| 15:01 | <cygri> | but you know, in 2011, rebooting and restarting all programs is just as much work as reinstalling was in 1998 ;-) |
| 15:01 | <wilhelm> | How do mathematicans usually type complicated equations or functions? Are visual editors used, or do people type TeX or MathML directly? |
| 15:01 | <asmodai> | A lot use TeX. |
| 15:02 | <asmodai> | For MathML I used to use my hand-written code, but switch to firemath extension for firefox |
| 15:02 | <asmodai> | If I can take the formulae writing professors at the university I work for as an example of mathematicians. |
| 15:02 | <jgraham> | wilhelm: Mathematicians use LaTeX |
| 15:02 | <jgraham> | Biologists use MS Word |
| 15:03 | <jgraham> | (to make a hopefully-not-too-unfair generalisation) |
| 15:04 | <jgraham> | Insane people type MathML directly |
| 15:04 | <gsnedders> | hsivonen: Ah, the answer seems to actually be someone implemented it in their spare time. |
| 15:04 | <hsivonen> | who typesets in Mathematica? |
| 15:04 | <wilhelm> | Right. I've dealt with one biologist and one professor emeritus in mathematics so far regarding this issue. |
| 15:04 | <wilhelm> | (That's wearing my web developer hat, not my browser vendor hat.) |
| 15:04 | <jgraham> | People typeset in mathematica? |
| 15:05 | <asmodai> | The funny part is |
| 15:05 | <asmodai> | Microsoft Windows 7 has the math input panel |
| 15:05 | <asmodai> | which places data on the clipboard that is mathml apparently |
| 15:05 | <asmodai> | http://dpcarlisle.blogspot.com/2010/01/mathml-on-clipboard.html |
| 15:07 | <wilhelm> | So from what age can one expect mathematicans to know LaTeX? (Can I just wait for the emeritus to die and be replaced by someone younger, or do I have to implement some fancy editor for these people? :) |
| 15:07 | <asmodai> | <?xml version="1.0" encoding="utf-16" ?> |
| 15:07 | <asmodai> | <m:math xmlns:m="http://www.w3.org/1998/Math/MathML"; xmlns:MicrosoftMathRecognizer="http://schemas.microsoft.com/mathrecognizer"; > |
| 15:08 | <wilhelm> | (I'm doing some consulting work for an old-fashioned encyclopædia moving from paper to web. :) |
| 15:09 | <karlushi> | hixie: Re http://www.w3.org/Bugs/Public/show_bug.cgi?id=13452 itemid |
| 15:09 | <karlushi> | It is not about redefining itemid but about clarifying the role. Your statement for closing is true but completely orthogonal to what was suggested. |
| 15:09 | <karlushi> | There is also still a false statement in the spec. |
| 15:09 | <karlushi> | "The exact meaning of a global identifier is determined by the vocabulary's specification." |
| 15:09 | <karlushi> | something along "the specific types of uris usable as a global identifier is determined by the vocabulary's specification." is more exact. |
| 15:09 | <jgraham> | wilhelm: In my very limited experience people old enough to not know LaTeX also have secretaries (or grad students) to typeset for them |
| 15:11 | <wilhelm> | jgraham: Ah. How convenient. |
| 15:11 | <jgraham> | Things I have learnt today: very early Laura Marling has a slight Kate Nash/Lilly Allen twang |
| 15:20 | <wilhelm> | jgraham: How do people studying physics or chemistry (or any other relevant subject) type their formulas? Are they more similar to biologists or mathematicans? (c: |
| 15:22 | <gsnedders> | wilhelm: I believe my gf's answer to that (chem, FWIW) is "on paper" :) |
| 15:22 | <jgraham> | gsnedders: Well she isn't a grad student |
| 15:22 | <jgraham> | So isn't writing papers |
| 15:22 | <jgraham> | (probably) |
| 15:23 | <gsnedders> | True. |
| 15:23 | gsnedders | finds someone older to chase after |
| 15:24 | <jgraham> | wilhelm: 435 |
| 15:24 | <wilhelm> | … 435? |
| 15:24 | <jgraham> | Just replace the axis label with "probablity of using LaTeX" |
| 15:25 | <jgraham> | And reduce the gap to mathematicians |
| 15:25 | <jgraham> | http://xkcd.com/435/ |
| 15:25 | <wilhelm> | … Ah. (c: |
| 15:35 | <jgraham> | http://www.enisa.europa.eu/act/application-security/web-security/a-security-analysis-of-next-generation-web-standards |
| 15:37 | <annevk> | yeah not all of those are actual problems |
| 15:38 | <annevk> | like those raised on web messaging and CORS are not |
| 15:38 | <annevk> | afaict |
| 15:38 | <annevk> | I wonder if they are in need of funding |
| 15:41 | <MikeSmith> | everybody's in need of funding |
| 15:43 | <hsivonen> | wow. site figuring out which parts of a video served by the site the user watched is listed as a threat |
| 15:44 | <annevk> | the web messaging issue was at best editorial but is actually just author-facing text vs requirements confusion |
| 15:45 | <annevk> | the issue with CORS the issue is that they do not quite get how "bad" the situation with arbitrary content bodies already is |
| 15:46 | <karlushi> | everybody's in need of fonding |
| 15:46 | <karlushi> | Apple dict is interesting on that topic. |
| 15:46 | <karlushi> | fond |fänd| |
| 15:46 | <karlushi> | adjective [ predic. ] ( fond of) |
| 15:46 | <karlushi> | having an affection or liking for : I'm very fond of Mike |
| 15:47 | <jgraham> | Oh, I thought you typoed fondling |
| 15:47 | <karlushi> | I should have |
| 15:47 | <karlushi> | it would have been more me. |
| 15:47 | <karlushi> | but I would have to replace MikeSmith by a dog |
| 15:47 | <karlushi> | Apple dict: "stroke or caress lovingly or erotically : the dog came over to have his ears fondled" |
| 15:48 | <karlushi> | I had no idea that there was zoophilia in Apple dict |
| 15:49 | <MikeSmith> | heh |
| 15:50 | <MikeSmith> | dog fondling |
| 15:50 | <jgraham> | I'm sure MikeSmith would like having his ears stroked too |
| 15:51 | wilhelm | covers his eyes. |
| 15:52 | <MikeSmith> | heh |
| 15:52 | AryehGregor | pings MikeSmith about Bugzilla and e-mail |
| 15:52 | <MikeSmith> | aha |
| 15:53 | <MikeSmith> | AryehGregor: so I will set up the bugzilla component |
| 15:53 | <MikeSmith> | you still not getting mail as expected? |
| 15:53 | <Ms2ger> | Oh, if you're doing that, I'd like one too ;) |
| 15:54 | <MikeSmith> | Ms2ger: for DOM Range? |
| 15:55 | <Ms2ger> | Parsing |
| 15:55 | <Ms2ger> | http://html5.org/specs/dom-parsing.html |
| 15:56 | <AryehGregor> | jgraham, a math postdoc who taught a summer research thing I was in told me that he had colleagues who used MS Word for typesetting. Apparently you can do it without driving yourself completely insane, if you memorize all the keyboard shortcuts. |
| 15:56 | <AryehGregor> | Although I assume it's much more limited. |
| 15:56 | <Ms2ger> | Madness |
| 15:57 | <AryehGregor> | (Incidentally, that math postdoc is now some sort of faculty at the University of Tokyo, and just e-mailed me in the last day telling me that the research I did with him that summer has been accepted for publication by the Journal of Number Theory. Awesome.) |
| 15:57 | <AryehGregor> | "Monomial Maps on P^2 and their Arithmetic Dynamics" |
| 15:58 | <jgraham> | AryehGregor: Oh. Wow. OK. My experience is admittedly only directly limited to a single astrophysics department which didn't have computers that would even run Word |
| 15:58 | AryehGregor | wonders how much of the resulting paper he'll actually understand |
| 15:58 | <jgraham> | Although I don't think it was atypical of the physical sciences as a whole |
| 15:58 | <annevk> | DOM Range already has a component |
| 15:58 | <AryehGregor> | I'm pretty sure LaTeX is the only format that journals will universally accept, though. It's definitely the standard one. |
| 15:58 | <jgraham> | AryehGregor: Congratulations |
| 15:58 | <gsnedders> | AryehGregor: The ECMAScript spec is a Word doc, though published as a PDF |
| 16:00 | <jgraham> | Also, I think that arXiv is LaTeX-only |
| 16:00 | <hsivonen> | AryehGregor: was your statement about the universal acceptance of LaTeX meant to be qualified my a field like math or physics? |
| 16:00 | <AryehGregor> | hsivonen, I was talking about math. |
| 16:01 | <AryehGregor> | Physics and computer science seem to be similar, though, in my somewhat limited and mostly indirect experience with them. |
| 16:01 | <annevk> | MikeSmith, while you are in Bugzilla "Access Control" can be renamed to "CORS" |
| 16:01 | <jgraham> | (arXiv is how physics, CS, and so on route around the brokenness of journals) |
| 16:01 | <AryehGregor> | And math. |
| 16:02 | <AryehGregor> | Did Perelman ever publish his proof of the geometrization hypothesis in an actual journal? |
| 16:02 | hsivonen | was unaware of CS seriously routing around the brokenness of journals |
| 16:02 | <hsivonen> | jgraham: do universities these day routinely give publication credit for arXiv when applying for funding or a raise? |
| 16:03 | <AryehGregor> | (apparently it's the geometrization conjecture, not hypothesis, whatever) |
| 16:03 | <jgraham> | Well I am not so sure about CS. They are somewhat weird because most publications are "conference proceedings" |
| 16:03 | <Ms2ger> | AryehGregor, I'm not sure you could generalize much from what Perelman does |
| 16:04 | <MikeSmith> | annevk: OK |
| 16:04 | <jgraham> | hsivonen: Since I never did either I wouldn't know. But people would put work that was on arXiv but not yet accepted by a journal on their CV |
| 16:04 | <MikeSmith> | Ms2ger, AryehGregor - will get those added and ping you once they're set up |
| 16:04 | <Ms2ger> | Ta |
| 16:04 | <jgraham> | Of course, they would still *submit* to a journal |
| 16:04 | <AryehGregor> | Ms2ger, you mean most world-famous mathematicians don't live on bread-and-water diets in their parents' house in Russia? |
| 16:05 | <AryehGregor> | MikeSmith, thanks. |
| 16:05 | <Ms2ger> | AFAIK, Wiles doesn't |
| 16:05 | <jgraham> | But the reality is that arXiv is how you get other people in the field to read what you have done |
| 16:05 | <Ms2ger> | Are there any other world-famous mathematicians? |
| 16:05 | <hsivonen> | jgraham: how can one submit to a journal if you've already published in arXiv? |
| 16:05 | <AryehGregor> | If by "world-famous" you mean "famous among mathematicians worldwide", then yes, lots. |
| 16:06 | <AryehGregor> | Even among non-mathematicians, some people have heard of, say, Terence Tao. |
| 16:06 | <AryehGregor> | Probably some others too. |
| 16:06 | <jgraham> | hsivonen: There is generally some handwaving about it being OK to distribute preprints |
| 16:06 | Ms2ger | googles |
| 16:06 | <AryehGregor> | Among mathematicians, sheesh, anyone who's written a moderately popular textbook. |
| 16:06 | <jgraham> | Even though you have to do all the copyright assignment rubbish |
| 16:07 | <jgraham> | (handwaving in the agreement you sign I mean) |
| 16:10 | <hsivonen> | jgraham: what about in terms of novelty? I've heard worries about even having data on a slide deck on the Web if the data hasn't been published in a journal yet |
| 16:11 | <AryehGregor> | MikeSmith, my post yesterday to the thread here "RFC from implementers on Element.innerText", from ayg⊙an, still hasn't appeared in the archives: http://lists.w3.org/Archives/Public/public-html/2011Aug/thread.html |
| 16:13 | <AryehGregor> | MikeSmith, I'm currently receiving public-html list mail to ayg⊙an, though. |
| 16:13 | <zewt> | the archives seem to randomly totally lose mails |
| 16:13 | <zewt> | though i don't recall if it's the w3 archives or whatwg ones that i've seen that on |
| 16:13 | <zewt> | probably the latter |
| 16:14 | <jgraham> | hsivonen: I have never heard anything like that |
| 16:14 | <AryehGregor> | jgraham, really? I have. Lots of academics are scared of others stealing their ideas. |
| 16:15 | <AryehGregor> | Historically there were lots of mathematicians who wouldn't publish things until they had worked out all the interesting implications themselves, so that they'd get all the credit. |
| 16:15 | <jgraham> | Well I have heard paranoia about being scooped of course |
| 16:15 | <AryehGregor> | Like Newton, Gauss, etc. |
| 16:15 | <jgraham> | Mathematics is a bit different from science here I think |
| 16:15 | <AryehGregor> | But if you put it on arXiv, of course, and someone else publishes first and doesn't credit you, you can complain to the journal and prove that you made it public first. |
| 16:16 | <AryehGregor> | I mean, you have to give attribution to stuff you find on arXiv too, if you used it. |
| 16:16 | <jgraham> | Indeed |
| 16:17 | <AryehGregor> | I'm guessing this isn't as much of an issue for experimental stuff, because most of the work is often in the physical work rather than the ideas. |
| 16:17 | <AryehGregor> | I'd expect theoretical physics to not be so different from math here, though. |
| 16:18 | <annevk> | AryehGregor, did you read the W3C email about posting to the archives and go to the website and hit accept? |
| 16:18 | <annevk> | AryehGregor, you should have gotten one |
| 16:19 | <AryehGregor> | Yes, I did. |
| 16:19 | <MikeSmith> | AryehGregor: I just checked and you message was in the approval queue |
| 16:19 | <MikeSmith> | not sure why |
| 16:20 | <MikeSmith> | but I made it go forth |
| 16:20 | <AryehGregor> | MikeSmith, thanks a lot. |
| 16:20 | <MikeSmith> | if it happens again, just let me know and I'll check |
| 16:20 | <MikeSmith> | there are many things about this mail system that still confuse me |
| 16:21 | <smaug____> | huh, strange things http://blog.chromium.org/2011/08/new-websocket-protocol-secure-and.html |
| 16:21 | <smaug____> | "no further breaking changes are expected" o_O |
| 16:23 | MikeSmith | steps out to get some grub |
| 16:23 | <annevk> | I like how they point to the W3C edition of the spec and the WHATWG mailing list |
| 16:25 | <jgraham> | I like how version -10 of the spec sends a Sec-WebSocket-Version header of 8 |
| 16:29 | <annevk> | http://www.w3.org/TR/WD-xml-link-970406 did not know about this |
| 16:29 | <annevk> | precursor to XLink! |
| 16:39 | AryehGregor | sets up Gmail to send mail via Gmail's SMTP servers instead of Gmail's SMTP servers, so that it appears to come from aryeh.name instead of gmail.com |
| 16:40 | <AryehGregor> | Which is not only pointless, it means that Gmail is now storing my password in plaintext. Hurrah. |
| 16:46 | <AryehGregor> | But no more distracting Sender: headers! |
| 16:46 | Philip` | doesn't think he's ever read a CS paper via arXiv |
| 16:47 | <dglazkov> | good morning, Whatwg |
| 16:47 | <Philip`> | Most people seem to provide their own articles as PDFs on their own site, so you can access most stuff for free |
| 16:48 | jgraham | is worried that dglazkov has delusions of being Robin Williams |
| 16:48 | <jgraham> | Philip`: But how do you find new stuff? |
| 16:49 | <dglazkov> | jgraham: fine, I'll stop good morningning whatwg |
| 16:50 | <Philip`> | jgraham: Look at the sites of relevant conferences, which list all their papers |
| 16:50 | <Philip`> | then they either link to a PDF or you Google for the title to find one |
| 16:53 | <Philip`> | Journal publication sounds extremely slow (though I have no experience with that), but conferences are usually like six months from first submission until the actual conference, so there's not much incentive to work around the delay by posting not-yet-published papers on other sites |
| 17:23 | <manu-db> | hsivonen: :-( looks like manu-db filed the absence of some RDFa misfeatures as bugs against Microdata on Saturday |
| 17:24 | <manu-db> | Call them what you will - they were requirements for structured data initiatives |
| 17:24 | <manu-db> | and seeing as how people were telling me that somebody should file them, but nobody way, I just took it upon myself to file the bugs so they're at least on record. |
| 17:25 | <manu-db> | I didn't expect that they'd result in changes to the spec |
| 17:26 | <manu-db> | and that seems to be true since almost all of them are being closed as WONTFIX without waiting for feedback |
| 17:26 | <AryehGregor> | Waiting for feedback from whom? |
| 17:26 | <AryehGregor> | You can reopen resolved bugs if you have new information you'd like to present in response to the resolver's comments. |
| 17:27 | <AryehGregor> | It says that somewhere in the resolution boilerplate. |
| 17:27 | <manu-db> | Yes, but that's needlessly combative... |
| 17:27 | <zewt> | well, it's very very bad bug-reporter-relations to close a bug without giving a chance to respond |
| 17:27 | <manu-db> | exactly |
| 17:27 | <zewt> | among the worst things you can do, really |
| 17:27 | <manu-db> | that's my point, exactly - thank you zewt |
| 17:28 | <scor> | I'd call it rude |
| 17:28 | <manu-db> | At least w/ the bugs that hsivonen and Hixie filed against RDFa - we had weeks of debate and discussion - recorded in the public logs - before deciding to close the bugs |
| 17:28 | <AryehGregor> | I guess it depends on your assumptions about how the bug tracker is to be used. |
| 17:28 | <annevk> | dglazkov, please don't! |
| 17:28 | <jgraham> | bugs are a stupid place for this kind of conversation |
| 17:29 | <zewt> | doesn't matter. if you close bugs like that, you're *going* to piss off reporters, and that can be avoided |
| 17:29 | <AryehGregor> | In HTML's case, open bugs automatically wind up on Hixie's todo list, and remain there until resolved. |
| 17:29 | <dglazkov> | annevk: ok. |
| 17:29 | <dglazkov> | :) |
| 17:29 | <AryehGregor> | So generally "open" = "awaiting action from Hixie". |
| 17:30 | <AryehGregor> | RESOLVED NEEDSINFO is supposed to mean "we can't do anything about this until the reporter answers our questions", but maybe it doesn't convey that very well. |
| 17:30 | <jgraham> | zewt: That wan't really my point |
| 17:30 | <AryehGregor> | I generally use NEEDSINFO if I think there's some reasonable chance that the reporter will be able to clarify, and WONTFIX if I'm confident that we don't want to fix the bug regardless. |
| 17:30 | <Philip`> | AryehGregor: RESOLVED doesn't really convey the impression that the issue is currently unresolved and awaiting further discussion |
| 17:31 | <zewt> | "resolved" says "we're done here" to reporters |
| 17:31 | <AryehGregor> | No, it doesn't, but I don't configure the W3C bug tracker. |
| 17:31 | <AryehGregor> | Or the HTMLWG policies that say how to use it. |
| 17:31 | <manu-db> | yes, the issue is "RESOLVED", AryehGregor |
| 17:32 | <AryehGregor> | I've sometimes avoided RESOLVED NEEDSINFO and just left the bug open when I'm confident the commenter will actually respond with the info, but I don't do that consistently. |
| 17:32 | <Philip`> | If people can agree on changes that would significantly improve matters, I assume it wouldn't be too hard to change the tracker and/or policies to cope with that |
| 17:32 | <manu-db> | I'd rather have a bug listed as OPEN NEEDSINFO - or BLOCKED NEEDSINFO |
| 17:32 | <AryehGregor> | Anyway, please do feel free to reopen any of the bugs if you feel you have further information to provide, regardless of what RESOLVED sounds like it means. |
| 17:32 | <Philip`> | so it's probably a useful discussion |
| 17:33 | <AryehGregor> | I do agree with your point about the demoralizing effect of the terminology here, but you'd have to complain to MikeSmith or someone about that, I can't add new options for what I can do to bugs. |
| 17:33 | <AryehGregor> | Anyway, I'm not going to be doing much bug-resolving anymore, it's taking away too much time from spec editing. |
| 17:35 | <manu-db> | It's not necessarily demoralizing for me, AryehGregor - but it really did sound like you were just closing the bug and not listening (I forgot which one precisely), but if you hadn't told me your intent - my base assumption is that you're not interested at all in addressing the concern. |
| 17:36 | <manu-db> | that's what RESOLVED means to me... |
| 17:36 | <manu-db> | I'll go back and review when I have some time and attempt to explain in more detail. |
| 17:36 | <AryehGregor> | Generally when I close NEEDSINFO I say something like "I don't understand, please explain X". If I close WONTFIX, then yes, I'm basically saying I think I understand what you're saying and I disagree and I don't think you're going to get me to agree. |
| 17:36 | <AryehGregor> | (but you're still welcome to try) |
| 17:37 | <wilhelm> | Opera's long-dead, horribly forked Bugzilla had a CUSTWAIT status (as in, more information from the customer is needed if we are to proceed with this bug). That prevented such bugs from cluttering developers' todo-lists. |
| 17:37 | <AryehGregor> | (or rather you're welcome to try getting Hixie to agree, since I probably won't be resolving bugs much anymore) |
| 17:37 | <manu-db> | I have little expectation that I'll convince any of the Microdata proponents about most of those features - but I just wanted to ensure that these issues were documented somewhere. |
| 17:38 | <zewt> | a lot of trackers have a "needs info, and the bug will switch to closed after a while otherwise" state |
| 17:38 | <manu-db> | at the very least, it protects Microdata from CR bugs... you can show people that there was a discussion about these features and you chose to not support them. |
| 17:38 | <manu-db> | I don't think we have that for these features vs. Microdata (except, perhaps, buried across the WHATWG list) |
| 17:39 | wilhelm | wonders what the primary usecase for http://dvcs.w3.org/hg/rdf/raw-file/5187c2033a91/rdf-turtle/built-xhtml/FPWD.html is. |
| 17:40 | <manu-db> | I don't think Hixie will agree with many of these bugs - he has a specific idea of what Microdata is for and doesn't seem that interested in supporting some of these other use cases that RDFa supports... which is fine - it's creating uber confusion w/ implementers, but there is only so much that can be done about that. |
| 17:40 | <zewt> | wilhelm: dunno, but its title sure makes me think of Logo |
| 17:41 | <manu-db> | TURTLE is helpful when expressing RDF in a fairly compact form |
| 17:41 | <scor> | wilhelm: you should ask in #swig |
| 17:42 | timeless | finishes ATAG20 |
| 17:42 | <timeless> | (yuck) |
| 17:42 | <manu-db> | we haven't found a use for it at our company - except when explaining the type of output a particular RDF processor should generate |
| 17:42 | <manu-db> | We tend to just use JSON-LD: http://json-ld.org/ |
| 17:42 | <manu-db> | and work in JSON |
| 17:43 | <manu-db> | which removes the need to work w/ TURTLE/SPARQL/etc. - it works for us, but other people find TURTLE as a good 'glue' RDF syntax. |
| 17:45 | <manu-db> | wilhelm: Go here http://ow.ly/5ThZK and click on the "TURTLE" tab - it's easier to read than the JSON |
| 17:45 | <AryehGregor> | MikeSmith, there's no way to change the e-mail address of a Bugzilla account, is there? |
| 17:46 | <wilhelm> | manu-db: Interesting. Thanks. |
| 17:58 | <AryehGregor> | TabAtkins, problem solved! http://aryeh.name/tmp/editing/editing.html#the-forecolor-command |
| 17:58 | AryehGregor | is proud of himself |
| 18:00 | <annevk> | how can currentColor be true at that point? |
| 18:00 | <annevk> | and how do you define valid CSS color? |
| 18:01 | <annevk> | or was this about the "comment" boxes on the side? |
| 18:01 | <annevk> | I should review this draft at some point... |
| 18:02 | <AryehGregor> | And you know what the best part is? It renders identically in all browsers, and I only tested initially in Chrome. |
| 18:02 | <AryehGregor> | <3 <3 standards |
| 18:02 | <AryehGregor> | annevk, it was about the first "Comments" box there, namely how to do the thing where it overflows to the left. |
| 18:02 | Philip` | gets confused by the "2." adjacent to the second "Comments" button, until realising that it comes from the <li> on the other side of the screen |
| 18:02 | <Philip`> | (in Opera 11.50) |
| 18:02 | <AryehGregor> | Yeah, I noticed that too. |
| 18:02 | <AryehGregor> | Looks like an Opera bug. |
| 18:03 | <AryehGregor> | It's fine in Chrome and Firefox. |
| 18:03 | <AryehGregor> | IE has a different bug there, it seems. |
| 18:03 | <Philip`> | So by "renders identically in all browsers" you mean except for Opera and IE? |
| 18:03 | <AryehGregor> | annevk, value can be currentColor if you do document.execCommand("forecolor", false, "currentColor"). As for defining valid CSS color, I don't (yet). |
| 18:03 | <AryehGregor> | Philip`, I was referring the the comment box, with the overflowing-to-left effect. |
| 18:04 | <AryehGregor> | Including the one-pixel-high absolutely positioned divs that make it looks like the border stretches out to accommodate the extra width of the table. |
| 18:05 | <AryehGregor> | Looks pixel-perfect between browsers. |
| 18:05 | <TabAtkins> | AryehGregor: Oh man, covering up the border with extra elements. Clever, but nasty. |
| 18:05 | <AryehGregor> | Notice the extra <div style=clear:right></div> I had to add, too. |
| 18:05 | <AryehGregor> | Hey, it works. |
| 18:06 | <AryehGregor> | (without the middle div, the last div is still positioned above the float) |
| 18:06 | <Ms2ger> | That looks like something Wikipedians would do |
| 18:06 | <AryehGregor> | I was never part of that insane-inline-style-hacks Wikipedia crowd, in case you were wonderingg. |
| 18:06 | <AryehGregor> | wondering. |
| 18:06 | <AryehGregor> | I don't think there's any way to do it without adding extra elements. ::before and ::after aren't enough, are they? |
| 18:07 | <Ms2ger> | I was more of a "use classes!" kind of person back then |
| 18:09 | <TabAtkins> | AryehGregor: They... might be. |
| 18:20 | <Hixie> | wow |
| 18:20 | <Hixie> | the w3c are actually going to fork the whatwg spec of their own accord |
| 18:21 | <Hixie> | so much for the "we don't want specs to fork" bullshit |
| 18:21 | <Hixie> | (webrtc wg) |
| 18:22 | <AryehGregor> | I thought we all agreed long ago that they only didn't want non-W3C organizations to fork W3C specs, and are fine with the W3C forking its own specs or anyone else's. |
| 18:22 | <AryehGregor> | At least they're consistent, give them that. |
| 18:22 | <AryehGregor> | Is anyone going to pay attention to the fork? |
| 18:23 | <Hixie> | AryehGregor: they'll certainly pay attention once i start citing it as precedent for forking while ignoring copyrights being fine |
| 18:24 | <Hixie> | (good news, everyone! w3c is saying by their actions that we can fork their specs and ignore the copyright!) |
| 18:25 | <AryehGregor> | If they're ignoring the license, it's only insofar as they didn't think to include some boilerplate somewhere, so it's not exactly comparable to ignoring a license that forbids redistribution at all. |
| 18:25 | <erlehmann> | i thought specs were permissively licensed? |
| 18:25 | <erlehmann> | or is it more like “verbatim copying allowed” ? |
| 18:25 | <erlehmann> | (i have no idea, sorry) |
| 18:26 | <Hixie> | erlehmann: w3c specs are (c) w3c "all rights reserved" |
| 18:26 | <erlehmann> | o.0 |
| 18:26 | <Hixie> | (whatwg specs are under a licence that allows reuse, though doesn't allow copyright reassignment) |
| 18:26 | <zewt> | (c) "we own the web" |
| 18:27 | <Hixie> | AryehGregor: i highly doubt they are intending to include any boilerplate |
| 18:27 | <Hixie> | AryehGregor: since pubrules requires that the spec says (c) w3c |
| 18:27 | <zewt> | even if it's a felony? |
| 18:27 | <erlehmann> | so they have replaced “© Copyright 2004-2011 Apple Computer, Inc., Mozilla Foundation, and Opera Software ASA.” ? |
| 18:27 | <AryehGregor> | You could make them change it if you complained, you know. |
| 18:27 | <Hixie> | they haven't done anything yet as far as i know |
| 18:27 | <erlehmann> | oh well |
| 18:27 | <Hixie> | AryehGregor: i intend to |
| 18:27 | <AryehGregor> | Have fun. |
| 18:27 | <Hixie> | once there's something to complain about :-) |
| 18:28 | <Hixie> | for now i'm just telling the person who is doing it that doing so violates w3c's policies |
| 18:28 | <manu-db> | who is doing that? |
| 18:28 | <timeless> | AryehGregor: generally bugzilla lets you change your email address... |
| 18:28 | <AryehGregor> | timeless, how? I didn't see an option. |
| 18:29 | <Hixie> | manu-db: webrtc wg |
| 18:29 | <manu-db> | the entire WG made a resolution to replace the copyright? |
| 18:29 | <timeless> | AryehGregor: typically https://bugzilla/userprefs.cgi?tab=account#new_login_name |
| 18:29 | <manu-db> | Link to the resolution? |
| 18:29 | <manu-db> | or discussion? |
| 18:30 | <zewt> | heh, random dripping sarcasm on webapps |
| 18:30 | <AryehGregor> | timeless, I don't see it here: http://www.w3.org/Bugs/Public/userprefs.cgi?tab=account |
| 18:30 | <AryehGregor> | Just password and real name, not login name. |
| 18:30 | <zewt> | infrequent enough compared to some lists that it stands out when it happens |
| 18:31 | <Hixie> | manu-db: nothing public currently |
| 18:31 | <manu-db> | I'd be surprised if they did what you are saying - seems like a fairly clear-cut copyright issue that could result in a law suit. |
| 18:32 | <timeless> | AryehGregor: then it's been disabled/removed |
| 18:32 | <timeless> | the feature exists in bugzilla upstream |
| 18:32 | <erlehmann> | manu-db, more like a cease and desist. |
| 18:32 | <timeless> | contact Team |
| 18:32 | <erlehmann> | right? |
| 18:32 | <manu-db> | At least, I haven't seen a WG be that stupid... ever. |
| 18:33 | <zewt> | iirc willful copyright misrepresentation is a felony in the US--could get worse than a lawsuit |
| 18:33 | <timeless> | AryehGregor: given the way permissions work @w3, it would make sense not to allow users to change accounts directly |
| 18:34 | <timeless> | zewt: http://www.lib.purdue.edu/uco/CopyrightBasics/penalties.html |
| 18:35 | <Hixie> | manu-db: i was surprised myself. but hopefully my e-mail will help them understand the problem and fix it. |
| 18:35 | <timeless> | i'm sure i don't want to know |
| 18:35 | <manu-db> | erlehmann, yes - that comes first - but one could argue that just by publishing a copy of that, they've caused irreparable harm to Apple, Opera and Mozilla. It's all lawyer BS at that point, but copyright violations like this can have a number of consequences: |
| 18:35 | <timeless> | is this rdfa-in-html? |
| 18:36 | timeless | goes looking for webrtc |
| 18:36 | <manu-db> | damages/profits awarded, $200 - $150K per violation, attorneys fees/costs, injunction, impounding of the website, jail time for infringer |
| 18:37 | <manu-db> | which is why I think that this isn't going to go anywhere - all that WG needs is to get a whiff of the rammifications and they'll back down. I'd chalk it up to "not knowing how copyright works" rather than willful infringement. |
| 18:37 | <AryehGregor> | It's obviously not willful, it was some people who weren't paying attention and/or didn't know what they were doing and will be fixed immediately if someone complains to the right people. |
| 18:38 | <manu-db> | timeless, no it's not rdfa-in-html :) |
| 18:38 | <timeless> | ok, sorry :) |
| 18:39 | <TabAtkins> | manu-db: Some of that only comes into play with criminal copyright infringment, which has much stricter requirements (intended to limit it to people, say, printing books and selling them without agreement from the copyright holder). |
| 18:39 | <manu-db> | TabAtkins: yes - that's true. However, lawyers usually aren't shy with pushing the law as far as they can. |
| 18:40 | <AryehGregor> | When it comes to criminal law, private lawyers have no say in enforcing it. You'd have to get the DA to go along with it. |
| 18:40 | <AryehGregor> | Private people can't file criminal charges. |
| 18:40 | <AryehGregor> | So it's irrelevant unless you think the government would care, which they absolutely would not. |
| 18:40 | <zewt> | well, in the US, anyway |
| 18:40 | <Hixie> | MikeSmith: is feedback to the -comments list still being turned into bugs? |
| 18:41 | <AryehGregor> | Plus none of it is relevant because they'd fix it immediately and clearly have been acting in good faith, so the judge is not going to be pleased with the plaintiff if they bring it to court after it caused them no damage and their complaint was promptly and fully addressed. |
| 18:41 | <zewt> | apparently the police enforce copyright in Japan (or some parts of; Kyoto?) on their own |
| 18:41 | <zewt> | which to me sounds completely insane |
| 18:41 | <AryehGregor> | That happens in the US too for criminal copyright infringement. |
| 18:41 | <AryehGregor> | But that's only very severe cases. |
| 18:42 | <zewt> | but not to that degree--they're acting against eg. p2p users directly |
| 18:42 | <zewt> | which is a strictly civil matter here |
| 18:47 | <AryehGregor> | Yes, true. |
| 18:47 | <timeless> | zewt =~ s/eg./e.g./g |
| 18:47 | <timeless> | (sorry, i'm 99 |
| 18:47 | <timeless> | % done complaining about ATAG20) |
| 18:48 | <hsivonen> | MikeSmith: congrats for http://es5.github.com/ being used to find a bug in the HTML5 parser in Gecko |
| 18:48 | <timeless> | in the html5 parser? nice |
| 18:48 | <timeless> | bug-url? |
| 18:48 | <hsivonen> | timeless: https://bugzilla.mozilla.org/show_bug.cgi?id=675499 |
| 18:51 | <timeless> | spiffy |
| 18:52 | timeless | chuckles |
| 18:52 | <timeless> | someone had: |
| 18:52 | <timeless> | .hgrc: [alias] |
| 18:52 | <timeless> | push = push --something |
| 18:52 | <timeless> | diff = diff --something |
| 18:52 | <timeless> | that eventually caused stack overflow |
| 18:53 | <timeless> | because the aliases were applied iteratively (once for each invocation of *something*) |
| 18:55 | <timeless> | (hg help config says that's a bad idea fwiw, so it was technically documented as don't do that..) |
| 18:57 | <hober> | Hixie: their intent to fork was made public a few days ago: http://lists.w3.org/Archives/Public/public-webrtc/2011Jul/0166.html |
| 18:58 | <erlehmann> | manu-db, i was of the impression the whatwg are nice guys. |
| 18:59 | <sicking> | Hixie: is there a reason why MessageEvents sent by WebSockets doesn't have a .origin? |
| 18:59 | <timeless> | Hixie: hey |
| 18:59 | <timeless> | http://www.whatwg.org/specs/web-apps/current-work/webrtc.html |
| 18:59 | <timeless> | s/specification that define the/specification that defines the/ |
| 18:59 | <timeless> | Hixie: ^ :) |
| 19:01 | <Ms2ger> | File a bug ;) |
| 19:01 | <annevk> | Hixie, that's chaired by a Google guy no? |
| 19:02 | <annevk> | Hixie, wonder if it'll turn out as lovely as the WebSocket protocol |
| 19:02 | <Ms2ger> | Probably |
| 19:27 | <annevk> | to be fair |
| 19:27 | <annevk> | browsers are implementing WebSocket as is |
| 19:27 | <annevk> | though I sort of think that's also because it wasn't really worth the fight |
| 19:31 | <zewt> | heh |
| 19:32 | <zewt> | posts saying "no, big waste of time" are, ironically, the biggest wastes of time |
| 19:32 | <annevk> | I meant more fighting for an alternative simpler version |
| 19:32 | <Michael> | Is there a rest api anywhere to query the HTML specs? |
| 19:33 | <MacTed> | cygri - latest (regular) Safari is Version 5.1 (6534.50) ... so they may well have already fixed the issue you hit earlier |
| 19:34 | <annevk> | Michael, nope |
| 19:34 | <cygri> | MacTed, that's right, that's what hsivonen told me as well. looks like 5.1 is the first version that actually includes an HTML5 parser. |
| 19:34 | <Michael> | darn. |
| 19:37 | <hsivonen> | Safari 5.1 usage share is growing more slowly than I had expected |
| 19:37 | hsivonen | expected System Update to replace Safari versions almost overnight |
| 19:37 | <hsivonen> | apparently not |
| 19:38 | <hsivonen> | partly the fault of Leopard and 5.1 not being available for Leopard |
| 19:51 | <sicking> | Hixie: (i think i asked elsewhere, but I can't find the message now, appologies if it's a repeat) is there a reason why websockets doesn't set .origin on the message events |
| 19:54 | <hober> | hsivonen: check 5.0.6 for leopard |
| 19:59 | <Hixie> | hober: that e-mail doesn't suggest they're going to use the same text, just fork the functionality |
| 19:59 | <Hixie> | sicking: what would the origin be? |
| 20:00 | <Hixie> | sicking: by definition if you can connect to a websocket, it's treated as same-origin |
| 20:01 | <gsnedders> | hsivonen: Yeah, Apple seem to be having a harder time getting people upgrading now. |
| 20:01 | <gsnedders> | hsivonen: I wonder if Leopard will still get security updates. |
| 20:02 | <timeless> | AryehGregor: hey |
| 20:02 | <AryehGregor> | timeless, hey. |
| 20:02 | <timeless> | what kind of caching headers are you using on http://aryeh.name/spec/editing/editing.html |
| 20:02 | <timeless> | my browser restarted today |
| 20:03 | <timeless> | but it showed me the july 26 version of the doc |
| 20:03 | <timeless> | a reload got me the july 29 version |
| 20:03 | <timeless> | s/the/a/g |
| 20:03 | AryehGregor | sees Cache-Control:must-revalidate, max-age=0 |
| 20:03 | <AryehGregor> | July 29 is the most recent published one. |
| 20:03 | <Hixie> | if i wanted to add a warning about clickjacking attacks to the html spec, anyone have any suggestions on where i should put it? |
| 20:03 | <timeless> | i guess that tabcandy doesn't honor that |
| 20:03 | <AryehGregor> | The WIP one is http://aryeh.name/tmp/editing/editing.html, but that's changing as I work on it in real time. |
| 20:03 | <timeless> | Hixie: near <iframe>s? |
| 20:04 | <AryehGregor> | I'm moving zillions of <!-- comments --> to be viewable without looking at the source, which is taking a while. |
| 20:04 | timeless | loads http://www.whatwg.org/specs/web-apps/current-work/multipage/the-iframe-element.html#the-iframe-element |
| 20:06 | <Hixie> | timeless: why near iframes? that's just one way to do clickjacking, it's not where an author would look for how to avoid getting p0wned, surely |
| 20:07 | <Ms2ger> | Rendering :) |
| 20:07 | <timeless> | http://www.whatwg.org/specs/web-apps/current-work/multipage/dom.html#security-document ? |
| 20:07 | <Hixie> | hmm |
| 20:07 | <Hixie> | i'm thinking maybe the intro section |
| 20:08 | <Hixie> | maybe under "A quick introduction to HTML" |
| 20:08 | timeless | shrugs |
| 20:08 | <timeless> | i'd search for it under a security section |
| 20:08 | <AryehGregor> | Do you have some kind of general security advice section? |
| 20:08 | <timeless> | http://www.whatwg.org/specs/web-apps/current-work/multipage/browsers.html#security-nav |
| 20:08 | <AryehGregor> | Why provide only this security advice and not other security advice too? |
| 20:08 | <Hixie> | AryehGregor: this is basically what this will become, i thikn |
| 20:08 | <timeless> | AryehGregor: he has 5 or so |
| 20:09 | <Hixie> | AryehGregor: i was going to mention a few other things too |
| 20:09 | <Hixie> | if i put it in the intro section |
| 20:09 | <timeless> | maybe 20 |
| 20:09 | <AryehGregor> | It makes sense to include advice about XSS, CSRF, etc. before you start on weird stuff like clickjacking. |
| 20:09 | <timeless> | yeah, clickjacking should not be the first thing to draft |
| 20:09 | <Hixie> | yeah |
| 20:09 | <timeless> | start w/ XSS |
| 20:10 | <timeless> | since the version of clickjacking that you seem to mean is actually a derivation from XSS |
| 20:11 | <timeless> | I guess after quick introduction wouldn't be a bad place |
| 20:11 | <timeless> | `Security Concerns` |
| 20:14 | <timeless> | > DOM trees contain several kinds of nodes, in particular a DOCTYPE node, elements, text nodes, and comment nodes. |
| 20:14 | <timeless> | Hixie: s/elements/element nodes/ ? |
| 20:14 | <Hixie> | *shrug* |
| 20:14 | <Hixie> | it's right either way |
| 20:15 | <timeless> | the change makes the sentence more consistent.. |
| 20:16 | <Ms2ger> | I'd file a tracker issue |
| 20:16 | <timeless> | does that require a bugzilla account? |
| 20:16 | timeless | isn't sure how to get one of those |
| 20:17 | <Hixie> | let's focus on actual bugs and not on whether sentences might be slightly more polished please :-P |
| 20:17 | <Hixie> | typos are one thing, i don't mind fixing those |
| 20:17 | <Ms2ger> | Hixie, the pedants must win! |
| 20:17 | <timeless> | your document is too long! and you're making it longer! ignore the fact that i wanted to make it a word longer ;-) |
| 20:18 | <Ms2ger> | Like me |
| 20:18 | <Hixie> | but when the sentence is accurate, correct, precise, grammatically sound, spelt correctly, scans well, and merely suffers from a slight inconsistency, let's just let it slide |
| 20:18 | <AryehGregor> | timeless, I consistently use all node types as standalone nouns, like "Comment" or "Element", except "Text", where I say "Text node". |
| 20:19 | <AryehGregor> | Like "If node is a Comment, do X. If node is a Text node, do Y." |
| 20:19 | <AryehGregor> | Or "If node is a Text, Comment, or ProcessingInstruction node, do Z." |
| 20:19 | <AryehGregor> | Because "If node is a Text" sounds silly. |
| 20:19 | <timeless> | AryehGregor: that would be a reasonable strategy, but from context, it isn't the one Hixie is using :) |
| 20:19 | <Hixie> | yeah if a bug does get filed on that particular thing, i'm just removing the last "node", not adding one after "elements" |
| 20:19 | <AryehGregor> | Evidently not. |
| 20:19 | <timeless> | Hixie: i'm fine w/ that fix :) |
| 20:20 | <Hixie> | cross-site scripting is essentially when you don't escape user input in html; sql injection is the same with sql; what do you call the similar thing where you don't validate user input on the server and so allow out-of-bounds input to be stored in the db? |
| 20:20 | <AryehGregor> | Hixie, "not a vulnerability by itself |
| 20:20 | <AryehGregor> | " |
| 20:21 | <timeless> | lol |
| 20:21 | <AryehGregor> | MediaWiki, for instance, doesn't escape anything before storing in the DB, as a rule. |
| 20:21 | <AryehGregor> | It trusts text content in the DB as untrusted. |
| 20:21 | <timeless> | Hixie: failure to enforce boundary constraints |
| 20:21 | <AryehGregor> | All escaping is done as close to output time as possible, so it's easy to audit. |
| 20:21 | <AryehGregor> | It's much harder to escape at input time, since input can come from a zillion different places that you have varying degrees of control over. |
| 20:21 | <timeless> | AryehGregor: what if there isn't `output` in the normal sense? |
| 20:21 | <timeless> | such as a join? |
| 20:22 | <AryehGregor> | Then what vulnerability is there? |
| 20:22 | <Ms2ger> | /nick Robert'); DROP TABLE Students;-- |
| 20:22 | <Hixie> | i mean like checking that an "age" field is >=0 |
| 20:22 | <AryehGregor> | By "output" I mean "use". |
| 20:22 | <AryehGregor> | Hixie, store it in the database as an unsigned integer? :) |
| 20:22 | <AryehGregor> | But okay. |
| 20:22 | <timeless> | s/0/18/ |
| 20:22 | <Hixie> | sure there's lots of solutions but what is the vulnerability called? |
| 20:22 | <timeless> | or 21 if you agree w/ Nancy Reagan.. |
| 20:22 | <AryehGregor> | "not validating user input" |
| 20:23 | <timeless> | failure to enforce boundary constraints ... |
| 20:23 | <Hixie> | yeah that's all i could come up with too |
| 20:23 | <Hixie> | i was hoping there was a buzzword like "sql injection" or "xss" |
| 20:23 | <timeless> | not everything is cool enough to get a buzzword |
| 20:25 | <timeless> | i think `failure to santize <whatever>` is probably the closest to a buzzword, but i'm not sure that matches what you want |
| 20:25 | <AryehGregor> | I don't know of a way to phrase it positively (without "failing", "not", etc.) |
| 20:25 | <timeless> | you can review cwe to see `failure to santize` appearing a bunch of times |
| 20:25 | <timeless> | `improper input validation` |
| 20:25 | <timeless> | doesn't technically have a negative :) |
| 20:25 | <timeless> | that's from CWE-74 |
| 20:26 | <AryehGregor> | "Input validation" is definitely the term for the thing you should be doing. |
| 20:26 | <AryehGregor> | For the thing you shouldn't be doing, "improper input validation" sounds like as good a term as any. |
| 20:26 | <AryehGregor> | It's not an exploit by itself, so it doesn't get a buzzword. |
| 20:26 | <AryehGregor> | Well, I guess it can be a minor exploit by itself. |
| 20:27 | <AryehGregor> | Like if you have maxlength and users can just remove it in their favorite web development tool and submit stuff that's whatever length they want, that might be classified as a security vulnerability, insofar as it involves users doing something with your site that they aren't supposed to be able to. |
| 20:27 | <Philip`> | One occasional problem is storing a user's IP address in a database, when you get it from some function that understands X-Forwarded-For, when not running behind a proxy that sets X-Forwarded-For, so malicious users can set it to whatever they want and trick you into storing the wrong IP address |
| 20:28 | <Ms2ger> | So, Bert thinks newline normalization in attributes is the biggest problem when parsing HTML with an SGML parser? |
| 20:28 | <Philip`> | which is sort of related to storing wrong information in a database, but separate from constraint validation (which any self-respecting SQL database ought to be enforcing automatically) |
| 20:29 | <timeless> | Philip`: in theory you're supposed to validate the source against a list of trusted proxies |
| 20:30 | <timeless> | Ms2ger: that means bert doesn't think html as sgml has many problems? |
| 20:30 | <timeless> | (like Comments or NetTags ?) |
| 20:30 | <Philip`> | In practice people forget to do that |
| 20:30 | <Ms2ger> | I wouldn't dare to claim I understand Bert |
| 20:30 | <timeless> | heh |
| 20:31 | <AryehGregor> | Philip`, Wikipedia was actually hit by a subtle version of that once. |
| 20:31 | <timeless> | AryehGregor: nice |
| 20:31 | <AryehGregor> | Squid did understand X-Forwarded-For and sanitized it, but the PHP variables used to access it treated X_Forwarded_For the same as X-Forwarded-For, and Squid didn't. |
| 20:31 | <timeless> | nice |
| 20:32 | <AryehGregor> | MediaWiki has a whitelist of trusted XFF providers that it uses, including various major ISPs in addition to its own internal Squids and Varnishes. |
| 20:32 | <timeless> | > Each element in the DOM tree is represented by an object, and these objects have APIs so that they can be manipulated. |
| 20:32 | <timeless> | Ms2ger, AryehGregor : what does `manipulated` mean to you? |
| 20:33 | <AryehGregor> | It's kind of vague. |
| 20:33 | <timeless> | to me, i think `modified`, but perhaps my dictionary is flawed |
| 20:33 | <Ms2ger> | Change any kind of internal state, I guess |
| 20:33 | timeless | underlines Ms2ger's use of the word `Change` |
| 20:33 | <AryehGregor> | Basically "modified", yeah. |
| 20:33 | <AryehGregor> | Very general. |
| 20:34 | <Ms2ger> | :) |
| 20:34 | <timeless> | but we all selected a word meaning `modify` |
| 20:34 | <timeless> | which to me is somewhat exclusive of the concept `to inspect` |
| 20:34 | <Ms2ger> | Mm |
| 20:36 | <AryehGregor> | I guess. |
| 20:37 | <timeless> | i'll take an Mm and an I guess :) |
| 20:50 | timeless | is confused |
| 20:50 | <timeless> | annevk: you around? |
| 20:57 | <annevk> | not really |
| 20:59 | <annevk> | but |
| 20:59 | <annevk> | DOM Core defines "element" to mean "Element node" |
| 21:06 | <timeless> | annevk: my question to you was different |
| 21:06 | <timeless> | (but it seems my irc connect is too flaky for such purposes) |
| 22:04 | <AryehGregor> | Can anyone explain to me why in Firefox (7.0a2), list items with the comments button all have an extra line break at the start? http://aryeh.name/tmp/editing/editing.html#split-the-parent |
| 22:05 | <TabAtkins> | I don't see an extra linebreak in current aurora. |
| 22:05 | <AryehGregor> | Hmm. |
| 22:05 | <AryehGregor> | Oh, Ctrl-F5 fixed it. |
| 22:05 | <AryehGregor> | Okay, good. |
| 22:06 | AryehGregor | has no idea why, though |
| 22:06 | <AryehGregor> | Opera places the list marker in a quite silly place. |
| 22:06 | <TabAtkins> | List marker placement is completely un-interoperable. |
| 22:07 | <AryehGregor> | Yeah, but putting it in front of the first child of the list when that child is right-floated is just silly. |
| 22:07 | <TabAtkins> | Firefox and Opera generally use one model, with significant differences. IE and Webkit generally use a second model, with significant differences. |
| 22:07 | <gsnedders> | TabAtkins: is this in Opera 12? |
| 22:07 | <TabAtkins> | I'm speccing the IE behavior, because it's simple and sane. |
| 22:07 | <TabAtkins> | gsnedders: It's in "whatever I have on my windows box, which keeps itself fairly updated". |
| 22:08 | <AryehGregor> | Hmm, now IE9 is behaving like Firefox did before I did Ctrl-F5. |
| 22:08 | <gsnedders> | TabAtkins: Ah, no, that's not out in a public build yet. Presto 2.9.192 finally fixes the <ul><li></ul> non-marker issue. |
| 22:08 | <gsnedders> | TabAtkins: And fixing that caused a lot of change of code that produces markers. |
| 22:09 | <gsnedders> | TabAtkins: Also, you defining what CSS properties affect list-markers? |
| 22:09 | <gsnedders> | (like font styling) |
| 22:10 | <TabAtkins> | gsnedders: All of them. ::marker is identical to ::before, basically. |
| 22:10 | <AryehGregor> | Ah. |
| 22:14 | AryehGregor | now is very confused |
| 22:14 | <TabAtkins> | Hm? |
| 22:14 | <AryehGregor> | IE9 is still creating the extra line breaks. |
| 22:14 | <AryehGregor> | I haven't found a minimal test case yet that distinguishes. |
| 22:15 | AryehGregor | concludes he's too tired to do that right now anyway |
| 22:15 | <timeless> | AryehGregor: network gremlins? |
| 22:15 | <AryehGregor> | I should really say "the extra vertical whitespace". |
| 22:15 | <AryehGregor> | Doesn't seem network-related. |
| 22:15 | <TabAtkins> | Ooh, I suspect it's related to the float interfering with phantom line boxes. |
| 22:16 | <TabAtkins> | The IE9 model, explained in terms of the Lists spec, which it doesn't actually implement but might as well... |
| 22:16 | <TabAtkins> | Means that ::marker is an abspos element using a special positioning scheme. Abspos elements leave behind a "placeholder" box in their normal position, which is a 0x0 inline element. |
| 22:17 | <TabAtkins> | Sometimes, if a linebox would only be generated due to a placeholder, CSS suppresses the linebox. |
| 22:17 | <TabAtkins> | It's possible that the float is interfering with this. |
| 22:17 | <timeless> | sounds yummy |
| 22:17 | <TabAtkins> | s/inline element/inline box/ |
| 22:17 | timeless | goes movie hunting |
| 22:17 | <timeless> | (seems like a better idea) |
| 22:18 | <timeless> | good luck storming the castke |
| 22:18 | <timeless> | s/k/l/ |
| 22:21 | <TabAtkins> | Hmm, but I also can't reproduce anything with that. |
| 22:24 | <zewt> | sleep well, and dream of large women |