| 00:20 | <reluctantwebdev> | Any humans here? |
| 00:20 | <jcranmer> | no |
| 00:20 | <TabAtkins> | Nah, just IRC bots. |
| 00:20 | <reluctantwebdev> | Darn. |
| 00:20 | <jcranmer> | we're all zombies here |
| 00:20 | <reluctantwebdev> | Oh well. |
| 00:20 | <bl4ckcomb_> | turing was wrong |
| 00:21 | <reluctantwebdev> | I’m fairly certain I can beat any web bot at whether it’s a bot or not, so I’ll stick around to see. :) |
| 00:21 | <reluctantwebdev> | So anyway, I had some thoughts about web development that I’d like to share. |
| 00:22 | <TabAtkins> | You said "I'd like to share". Tell me more about that. |
| 00:22 | <reluctantwebdev> | Hah! |
| 00:22 | <reluctantwebdev> | Good one! |
| 00:22 | <reluctantwebdev> | Moving on... |
| 00:23 | <reluctantwebdev> | The way I see it, all web pages have 3 things that need to be covered. |
| 00:23 | <reluctantwebdev> | Layout, content handling, and the content itself. |
| 00:24 | <reluctantwebdev> | When I’m trying (and failing) to make a web page, it’s usually because I want to put the layout down first. |
| 00:24 | <reluctantwebdev> | As I would prefer it, I’d pretty much rather have layout be its own document. That way, I can change the layout without changing the content, and without changing how that content is handled. |
| 00:25 | <bl4ckcomb_> | that's why there is CSS |
| 00:25 | <reluctantwebdev> | Not exactly. |
| 00:25 | <bl4ckcomb_> | html is structure and semantics |
| 00:25 | <reluctantwebdev> | Sure, CSS handles layout, but it also handles things completely unrelated to layout. |
| 00:25 | <reluctantwebdev> | Stylizing content has NOTHING to do with layout. |
| 00:25 | <reluctantwebdev> | Yes, I’m aware of what HTML purports to do. |
| 00:26 | <reluctantwebdev> | Frankly, in my opinion, it’s a content handler. |
| 00:26 | <reluctantwebdev> | Or the content itself, really. |
| 00:26 | <reluctantwebdev> | HTML is the content, CSS is the content handler, saying how the content should be presented. |
| 00:26 | <bl4ckcomb_> | html is the structure which contains content |
| 00:27 | <reluctantwebdev> | Only sort of. |
| 00:27 | <bl4ckcomb_> | I'm not sure what you mean by handler |
| 00:27 | <reluctantwebdev> | One could argue that paragraphs themselves are content. |
| 00:27 | <bl4ckcomb_> | html doesnt doe anything with content then saying its there |
| 00:27 | <reluctantwebdev> | Well, suppose you had a video. |
| 00:27 | <bl4ckcomb_> | *do |
| 00:27 | <reluctantwebdev> | But, when you put up the video on your website, you want to stylize it. |
| 00:28 | <reluctantwebdev> | So, you feed the video to your content handler, and the content handler adds snow, or sepia tones or whatever. |
| 00:28 | <reluctantwebdev> | Not that we can do that, currently; |
| 00:28 | <reluctantwebdev> | but it’s the same metaphor as stylizing text. |
| 00:29 | <bl4ckcomb_> | thats not the purpose of html. html will for instance define controls for video |
| 00:29 | <reluctantwebdev> | See, from where I’m sitting, CSS is trying to handle both the bottom of the document, the layout, and the middle of the document, BUT ONLY IF THE DOCUMENT IS TEXT. |
| 00:29 | <bl4ckcomb_> | not effects |
| 00:29 | <reluctantwebdev> | Yes, I know this. |
| 00:29 | <reluctantwebdev> | I’m saying that not respecting this whole “layout, content handler, content” hierarchy is flawed. |
| 00:30 | <reluctantwebdev> | For instance, what exists in place of CSS for video, or audio even? |
| 00:30 | <TabAtkins> | You haven't said anything about anything violating the hierarchy yet. |
| 00:30 | <TabAtkins> | And I can't parse that last sentence. |
| 00:30 | <reluctantwebdev> | There is no such thing. |
| 00:30 | <bl4ckcomb_> | reluctantwebdev, js can manipulate things |
| 00:30 | <reluctantwebdev> | Yes, maybe it can, but you wouldn’t want to stylize video with it. |
| 00:31 | <bl4ckcomb_> | video or the player? |
| 00:31 | <reluctantwebdev> | Then we’re just using a plugin again, bl4ckcomb. |
| 00:31 | <TabAtkins> | Oh, you want to style a video? FF has a (currently proprietary) ability to use SVG filters on HTML content via CSS. This should be specified in some manner in the future with the Filters Module for CSS. |
| 00:31 | <bl4ckcomb_> | not exactly |
| 00:32 | <reluctantwebdev> | The issue isn’t that I want to stylize a video. |
| 00:32 | <reluctantwebdev> | My point is, stylization should be done by the content handler, whatever that content handler is. |
| 00:32 | <TabAtkins> | Which is CSS, presumably. Which can stylize a video. Is there a problem? |
| 00:32 | <bl4ckcomb_> | you mean adding properties to a structure, reluctantwebdev ? |
| 00:33 | <reluctantwebdev> | What I’m trying to get it is that trying to get CSS to stylize everything is ridiculous. |
| 00:33 | <reluctantwebdev> | I also think that having CSS do layout is ridiculous, but that’s kind of another thing. |
| 00:33 | <TabAtkins> | You haven't made any such argument yet. You've so far just made a few observations that don't appear to be leading anywhere concrete. |
| 00:34 | <reluctantwebdev> | For instance, how would you stylize video of the World Cup to filter out those annoying horns? |
| 00:34 | <bl4ckcomb_> | you do it before you publish it |
| 00:34 | <TabAtkins> | You want to do advanced audio processing? Do that before you send the video to the browser. |
| 00:34 | <reluctantwebdev> | Yes, but you don’t specify the font before you publish text content to the web, do you? |
| 00:35 | <reluctantwebdev> | I’m pointing out that we have an inconsistency between how text and video are handled. |
| 00:35 | <TabAtkins> | Because specifying the font isn't computationally intensive and requiring of techniques beyond most webdevs. |
| 00:35 | <bl4ckcomb_> | video is non-readable content |
| 00:35 | <bl4ckcomb_> | the user agent must download it separately |
| 00:35 | <reluctantwebdev> | That all depends on how the video is encoded, Bl4ckcomb. |
| 00:36 | <reluctantwebdev> | What I really think is missing is that you are not treating the issues of layout, content handling, and the content itself as separate areas. |
| 00:36 | <reluctantwebdev> | Much like how WSIWYG stuff was crammed into HTML thoughtlessly. |
| 00:37 | <bl4ckcomb_> | huh? |
| 00:37 | <TabAtkins> | Nah, we're definitely treating them as separate areas. We're also practical about it, though. |
| 00:37 | <reluctantwebdev> | We can’t anticipate how the content of the future is going to be styled. |
| 00:37 | <bl4ckcomb_> | I'm not following at all |
| 00:38 | <reluctantwebdev> | bl4ckcomb, for instance, why are we changing a document whose main purpose is to stylize text in order to do the layout of the web page, which is a completely separate issue? |
| 00:39 | <TabAtkins> | The fact that CSS has properties for text styles and, separately, properties for document layout doesn't mean anything. |
| 00:39 | <reluctantwebdev> | I think it does. |
| 00:39 | <TabAtkins> | Your argument seems to be suggesting that you think the text-styling language and the document-layout language *must* be different things. |
| 00:39 | <TabAtkins> | Presumably because using a different syntax makes them more "independent"? |
| 00:40 | <reluctantwebdev> | My point is, I don’t see any rationale as to why they are the SAME thing. |
| 00:40 | <TabAtkins> | ...and? |
| 00:41 | <reluctantwebdev_> | Bah, disconnected. |
| 00:41 | <TabAtkins> | "I dont' see a good reason for X" isn't the same as "I see a good reason for not X". |
| 00:42 | <reluctantwebdev_> | I can understand, for instance, why HTML should be purely semantic. |
| 00:42 | <bl4ckcomb_> | reluctantwebdev, you could develop your own style language to style video's, but imo its better that css is extended for that purpose and certainly not html |
| 00:42 | <reluctantwebdev_> | I do not understand why that means that layout, which is a completely different issue from style, should be lumped into CSS. |
| 00:43 | <reluctantwebdev_> | Moreover, I don’t see why HTML should be the document web browsers look for if it isn’t going to be handling issues like layout. |
| 00:43 | <bl4ckcomb_> | reluctantwebdev, CSS does things that need to be seen or experienced as a human user. html gives everything that user agents (not only browsers) must know |
| 00:43 | <TabAtkins> | You're making an argument that the current design of the web platform appears somewhat arbitrary. |
| 00:43 | <TabAtkins> | The answer to that is "Yes, it is." |
| 00:43 | <reluctantwebdev_> | Exactly. |
| 00:44 | <TabAtkins> | But you aren't making an argument that it is bad, or that there is a better solution. |
| 00:44 | <reluctantwebdev_> | I’m making the argument that, in that people are at least somewhat logical, there should be a somewhat logical layout to the division of labor. |
| 00:44 | <reluctantwebdev_> | If only so that people can understand how to make web pages easier. |
| 00:45 | <reluctantwebdev_> | I’m also asking if there has been any effort thus far to actually make a web document structure with meaningful division of labor. |
| 00:45 | <TabAtkins> | You haven't made an argument that the current division is illogical, just that it's somewhat arbitrary. |
| 00:46 | <reluctantwebdev_> | If you want to change the layout of a file, you don’t want to change the style of the content. |
| 00:46 | <TabAtkins> | Again, CSS being both a styling and a layout language is an accident of history, but it's not so far been an actual problem. |
| 00:46 | <reluctantwebdev_> | Lumping those two things together is illogical. |
| 00:46 | <TabAtkins> | ...yes? |
| 00:46 | <TabAtkins> | Lumping the two such that changing one changed the other would be. |
| 00:46 | <reluctantwebdev_> | I’m merely stating that it’s a problem for ME. |
| 00:46 | <TabAtkins> | Having the two be separate features in a single language/document isn't illogical. |
| 00:47 | <reluctantwebdev_> | TabAtkins, having both features in a document only makes sense if that document contains all necessary information to display the web page. |
| 00:47 | <reluctantwebdev_> | I figure you either lump it all together, or you provide meaningful division of labor. |
| 00:48 | <reluctantwebdev_> | Either approach makes sense to me, and the status quo does not. |
| 00:48 | <TabAtkins> | That's one possible way to do it, yes. It's not a requirement. |
| 00:49 | <daedb> | If you want layout and style to be separate documents, just put them in separate css files and link both of them :p |
| 00:49 | <reluctantwebdev_> | One thing I wonder about is why bother having CSS style text at all, when we turn to outside sources for video anyway? |
| 00:49 | <TabAtkins> | I don't understand. |
| 00:50 | <reluctantwebdev_> | When we play videos, we embed a video player in our web page to handle the content. |
| 00:50 | <reluctantwebdev_> | Why are we bothering stylizing text? Why not just have an embeddable text displayer, capable of handling stylized content on its own? |
| 00:51 | <reluctantwebdev_> | The whole mess seems inconsistent. |
| 00:51 | <TabAtkins> | Because text is easy to display, and cheap bandwidth-wise to send over in a plain manner. |
| 00:52 | <bl4ckcomb_> | reluctantwebdev_, I still see videos, tables, images, ... as attachments to the text that is displayed |
| 00:52 | <reluctantwebdev_> | Is it really that much better than just calling something to display an RTF? |
| 00:53 | <TabAtkins> | Apparently, yes. |
| 00:53 | <reluctantwebdev_> | Based on... ? |
| 00:53 | <TabAtkins> | History of the web platform? |
| 00:54 | <reluctantwebdev_> | I recognize the need to maintain backwards compatibility. At the same time, that isn’t really evidence that it is better than an alternative. |
| 00:55 | <TabAtkins> | Indeed, there are better alternatives. There's no path going from here to there, though. |
| 00:56 | <reluctantwebdev_> | It seems to me, that, for the most part, HTML wants to evolve into merely a type of content, rather than the means by which that content is displayed, in much the same manner as a pure text file is content without formatting. |
| 00:57 | <reluctantwebdev_> | But I need to take content I already have, slap some style on it, and lay it out on the web page. |
| 00:58 | <reluctantwebdev_> | I may change the layout from time to time, I may change the style, and I may change the content. |
| 00:58 | <reluctantwebdev_> | But rarely will I ever modify two out of the 3 at once. |
| 01:01 | <reluctantwebdev_> | Not that there is an alternative that embodies this approach (for the web, anyway), but it seems like this is how web pages can, should, and often are, built; and it seems like the languages we use to code web pages aught to reflect that. |
| 01:01 | <TabAtkins> | You're still attempting to assert without evidence that it is better to have your three concerns handled by different languages. |
| 01:02 | <reluctantwebdev_> | In order to have evidence, I’d have to point to an industry that dynamically displays content on a regular bais. |
| 01:02 | <reluctantwebdev_> | Err, basis. |
| 01:02 | <reluctantwebdev_> | One that isn’t the Web. |
| 01:02 | <TabAtkins> | Not necessarily. You just have to, you know, make an argument. So far your argument appears to be "Well, I think it's cleaner this way." |
| 01:03 | <jcranmer> | you can do it that way if you use, say XSLT |
| 01:03 | <reluctantwebdev_> | I suspect there’s good reason I’ve never heard of XSLT. |
| 01:03 | <jcranmer> | what it comes down to is that things remain the way they are, I suppose |
| 01:04 | <jcranmer> | because people want them to be that way |
| 01:04 | <reluctantwebdev_> | No, if people wanted them to be that way, they’d be touting how great this division of labor is above anything else. |
| 01:04 | <jcranmer> | right now, you already have to effectively learn HTML, JS, and CSS |
| 01:04 | <reluctantwebdev_> | Exactly. That in itself is ridiculous. |
| 01:05 | <jcranmer> | your proposal would increase the number of languages to use |
| 01:05 | <reluctantwebdev_> | Maybe it would, but it makes more sense than kicking out JUST HTML. |
| 01:05 | <jcranmer> | JS for dynamic content, HTML for static content, CSS for styling, MAGIC for layout |
| 01:06 | <TabAtkins> | reluctantwebdev_: You're still not giving a reason why ripping out a chunk of CSS's functionality and giving it a different syntax would be an improvement. |
| 01:06 | <jcranmer> | the current answer seems to be using HTML to brush out the layout and CSS to actually fill it in |
| 01:06 | <reluctantwebdev_> | Because layout has its own issues that are not addressed by improving style attributes. |
| 01:07 | <jcranmer> | CSS already does most of the layout |
| 01:07 | <reluctantwebdev_> | Improving the ability to do layout has nothing to do with the ability to improve style, and deserves its own resources. |
| 01:07 | <jcranmer> | designing a new language would not give it its own resources |
| 01:07 | <jcranmer> | look at XHTML 2 :-) |
| 01:07 | <TabAtkins> | So are you now asserting that giving a chunk of CSS a different syntax will produce more resources for it? |
| 01:08 | <reluctantwebdev_> | I am merely asserting that the status quo is unacceptable and that someone should do something about it. |
| 01:08 | <TabAtkins> | You're asserting that, yes. So far without any evidence. |
| 01:08 | <jcranmer> | well, clearly, the CSS WG is interested in tackling layout |
| 01:08 | <jcranmer> | that's why you have advanced layout and flexbox modules |
| 01:09 | <reluctantwebdev_> | I think that web pages should have a sensible design methodology behind them, and right now, they don’t. |
| 01:09 | <jcranmer> | how not? |
| 01:09 | <TabAtkins> | You haven't made the argument that they don't. |
| 01:11 | <reluctantwebdev_> | Layout, content, and how the content is handled should all be on equal footing from one another, because they are all independent parts of the design process, with their own separate issues. |
| 01:12 | <reluctantwebdev_> | You could also argue that they are interdependent, but you could not argue that some are interdependent, and the others are not. |
| 01:12 | <TabAtkins> | And they are currently. You're trying to assert that, because two of them are addressed under the umbrella of a particular syntax, that they're not. You're wrong. |
| 01:12 | <jcranmer> | layout and content are very highly interdependent |
| 01:13 | <reluctantwebdev_> | That’s not true, jcranmer. |
| 01:13 | <jcranmer> | if I had a boilerplate copyright notice, I want that to be laid out in a specific location |
| 01:13 | <jcranmer> | adding new content impacts my layout |
| 01:13 | <reluctantwebdev_> | If the length of your content changes, either the layout must also change, or must be pre-configured to allow scrolling in the same space. |
| 01:13 | <jcranmer> | as I said, modifying your content at the very least requires you to consider the impact to layout |
| 01:14 | <reluctantwebdev_> | Right, but if you also increase the size of your font, then that changes the layout also. |
| 01:14 | <reluctantwebdev_> | Sometimes, you change the size of the font to suit the layout. |
| 01:14 | <reluctantwebdev_> | Yes, I misunderstood you initially. |
| 01:14 | <jcranmer> | clearly, they are not independent |
| 01:14 | <reluctantwebdev_> | No, not entirely. |
| 01:15 | <reluctantwebdev_> | But they still have independent steps in the design process. |
| 01:15 | <reluctantwebdev_> | You may go back for tweaking, but for the most part, the general shape stays the same. |
| 01:15 | <jcranmer> | not necessarily |
| 01:16 | <jcranmer> | well, I'm not a professional web developer |
| 01:16 | <reluctantwebdev_> | Neither am I, frankly. |
| 01:16 | <jcranmer> | or web designer |
| 01:16 | <reluctantwebdev_> | I’m just saying that the languages should conform to the process. |
| 01:17 | <jcranmer> | my understanding of the process is this |
| 01:17 | <reluctantwebdev_> | Programming languages may exist for the sake of interpretation by computer, but their FORM should exist for the people that use them. |
| 01:17 | <jcranmer> | you get a graphics designer to sketch out how the site should look |
| 01:17 | <jcranmer> | you then get the web developer to actually implement that design |
| 01:17 | <reluctantwebdev_> | The layout, more or less. |
| 01:18 | <reluctantwebdev_> | Yes, but that doesn’t break it down nearly enough. |
| 01:18 | <jcranmer> | either that, or the graphics designer just pulls up, say, Dreamweaver and the web developer groans |
| 01:18 | <reluctantwebdev_> | First, you come up with the layout, so that you can nail down usability. |
| 01:19 | <jcranmer> | in none of the design processes that I've observed has that really happened |
| 01:19 | <reluctantwebdev_> | Maybe not. :P |
| 01:19 | <jcranmer> | in any case |
| 01:19 | <reluctantwebdev_> | Then again, that could just be a symptom of the current division of labor. |
| 01:19 | <jcranmer> | there's no great clamor to change the way it's done |
| 01:19 | <jcranmer> | it's clearly good enough for now |
| 01:21 | <reluctantwebdev_> | Well, of course the people that can stomach it are content with the status quo... :P |
| 01:25 | <reluctantwebdev_> | Anyway, I figure that as long as the tools of web development have irrational divisions of labor, we’re going to be stuck with poor WSIWYG tools. |
| 01:27 | <jcranmer> | as long as WYSIWYG attempts to get pixel perfection, we're going to have crappy WYSIWYG tools |
| 01:28 | <bl4ckcomb_> | code generators are crappy by definition |
| 01:28 | <jcranmer> | not necessarily |
| 01:34 | <reluctantwebdev_> | Something that might demonstrate my point is to see how many otherwise modern web pages use CSS for styling but HTML for formatting. |
| 01:35 | <reluctantwebdev_> | I wouldn’t know how to get that, though. |
| 01:36 | <jcranmer> | that's not too hard |
| 01:36 | <jcranmer> | just scrape a load of pages |
| 01:37 | <reluctantwebdev_> | It’s my understanding that Google did a study, and that study is the basis of a lot of the HTML 5 decisions, by using what’s actually used in practice. |
| 01:41 | <GPHemsley> | It it valid for a runaway <a> to span multiple <li>s? |
| 01:41 | <GPHemsley> | (from a rendering perspective) |
| 01:56 | <MikeSmith> | GPHemsley: what does it mean, "runaway <a>" ? |
| 01:56 | <GPHemsley> | came across a situation where an </a> was accidentally written as </li> |
| 01:56 | <reluctantwebdev_> | I think he means without an ending tag. |
| 01:56 | <MikeSmith> | OK, so that's not valid |
| 01:56 | <GPHemsley> | thus, there was <li>Text before link <a>Runaway link</li> Text supposed to be after link</li><li>Another point altogether. Should this be linkified?</li> |
| 01:56 | <GPHemsley> | Not valid markup, obviously |
| 01:56 | <GPHemsley> | but what should the browser do in that situation? |
| 01:57 | <TabAtkins> | You'd have to look at the parser/tokenizer part of the spec. (I dunno.) |
| 01:57 | <GPHemsley> | :P |
| 01:58 | <MikeSmith> | GPHemsley: what TabAtkins said -- it should close the <a> element where the parsing algorithm says it should be closed |
| 01:58 | <MikeSmith> | and the rendering is then whatever you end up with in the DOM after that |
| 01:58 | <GPHemsley> | any tips (= links) to which part specifically I should be looking at? |
| 02:00 | <GPHemsley> | (Any idea what the non-HTML5 behavior was?) |
| 02:00 | <GPHemsley> | Pardon me for bringing my logic in with me for a moment, but I would think that the link should end with the </li>, no? |
| 02:00 | <MikeSmith> | the non-HTML5 behavior is likely pretty much what's in the spec |
| 02:02 | <GPHemsley> | hmm... it appears that what the browser did may be what the spec says (not that I'm surprised) |
| 02:02 | <GPHemsley> | http://dev.w3.org/html5/spec/the-end.html#misnested-tags:-b-p-b-p |
| 02:02 | <MikeSmith> | GPHemsley: you can use Live DOM Viewer to check this kind of stuff |
| 02:02 | <GPHemsley> | nah, I don't care that much :) |
| 02:03 | <MikeSmith> | OK, then I'l quit taking any more time to try to help you figure it out :) |
| 02:06 | <GPHemsley> | :P |
| 02:07 | <GPHemsley> | Well, I was just curious if I'd found a bug... and currently, there are two strikes against that |
| 02:07 | <GPHemsley> | ;) |
| 04:40 | <MikeSmith> | so I'm trying to serialize a DOM tree |
| 04:40 | <MikeSmith> | to another window |
| 04:40 | <MikeSmith> | I do var x = window.open(); |
| 04:41 | <MikeSmith> | then x.document.write(document.toString()); |
| 04:41 | <MikeSmith> | but I don't get the document content serialized out |
| 04:42 | <MikeSmith> | I instead get "[object HTMLDocument]" |
| 04:43 | <MikeSmith> | clearly I'm doing something wrong |
| 05:16 | <othermaciej> | hi all |
| 05:25 | <MikeSmith> | is there a restriction on being able to save the content of an about:blank window? |
| 05:25 | <MikeSmith> | save = user manually saves it to a file on filesystem |
| 05:27 | <MikeSmith> | my WebKit tells me, "The document "foo bar" could not be exported to "foo.html" |
| 05:28 | <MikeSmith> | and Chrome does not even give me a menu option for saving it |
| 05:28 | <MikeSmith> | and Gecko appears to not even let me write to the window to begin with |
| 08:27 | <zcorpan_> | MikeSmith: document.toString() is "[object HTMLDocument]" |
| 08:27 | <zcorpan_> | MikeSmith: maybe you want document.innerHTML, but that's not implemented anywhere |
| 08:28 | <jgraham> | If you don't care about things outside the <html> element, you could use .outerHTML on the root element |
| 08:29 | <zcorpan_> | i don't know what restrictions there are for about:blank documents |
| 08:30 | <hober> | have any of the major browser vendors expressed interest, one way or another, in implementing some form of distributed extensibility in text/html? |
| 08:30 | <hober> | besides msft, who have said they'd rather not: http://lists.w3.org/Archives/Public/public-html/2010Aug/0116.html |
| 08:30 | <jgraham> | hober: Microsoft did but then said that it shouldn't be in HTML5 |
| 08:31 | <hober> | jgraham: indeed |
| 08:31 | <micheil> | hey, anyone on the microsoft board of directors want to get IE9 onto XP? |
| 08:31 | <micheil> | (like, if anyone has some contacts or anything like that? :P ) |
| 08:31 | <hober> | micheil: I imagine that'll happen when safari 5 runs on mac os 10.3 |
| 08:32 | <jgraham> | I assume the Microsoft borad of directors don't hang out in #whatwg |
| 08:32 | <hober> | jgraham: not usually anyway |
| 08:32 | <zcorpan_> | hober: i think some vendors have expressed interest in implementing microdata |
| 08:32 | <zcorpan_> | which is some form of distributed extensibility in text/html |
| 08:32 | <micheil> | hober: slightly different I think, considering the os x updates happen pretty easily |
| 08:33 | <micheil> | jgraham: I had a feeling about that. |
| 08:33 | <jgraham> | I'm pretty sure some of the others must have said they didn't want to implement namespaces somewhere citable |
| 08:33 | <hober> | zcorpan_: true, but I think when people say "distributed extensibility" what they mean is "custom element names" |
| 08:33 | <jgraham> | But I couldn't tell you where |
| 08:34 | <zcorpan_> | hober: when some people say "distributed extensibility" what they mean is "xml namespaces in text/html" |
| 08:34 | <hober> | micheil: honestly, I've never upgraded an Mac OS X machine from 10.n to 10.n+1 |
| 08:34 | <hober> | micheil: the autoupdates get you from 10.n.m to 10.n.m+1 |
| 08:34 | <hober> | zcorpan_: indeed |
| 08:35 | <jgraham> | When some people say "distributed extensibility" they mean "the ability to write fbml templates and send them over the wire as valid HTML" |
| 08:35 | <hober> | jgraham: yeah, same here, but now that I'm writing the null CP it'd be nice to have citations |
| 08:35 | <micheil> | oh, well, at any rate, I think there'd be more XP machine out there then 10.3 machines |
| 08:35 | <jgraham> | Which makes no sense to me whatsoever |
| 08:35 | <hober> | jgraham: i think it boils down to microsoft word export-to-html o:foo elements |
| 08:38 | <zcorpan_> | http://diveintomark.org/archives/2009/10/05/distributed-unicorns-and-ponies |
| 08:42 | <hober> | zcorpan_: classic |
| 09:01 | <ashaw> | Didi anyone here see my CIE coleospace proposal on css-style, any coments? |
| 09:12 | <ashaw> | Did anyone here see my CIE colorspace proposal on css-style? any coments? |
| 09:12 | <erlehmann> | ashaw, still not |
| 09:12 | <annevk> | no need to ask questions twice |
| 09:13 | <annevk> | we have logs |
| 09:13 | <ashaw> | sorry, I was correcting my spelling. |
| 09:13 | <annevk> | the convention for that is s/coleospace/colorspace/ normally |
| 09:13 | <ashaw> | sorry. |
| 09:14 | <annevk> | no worries |
| 09:16 | <ashaw> | Is there any way to force integer operations in Javascript. |
| 09:16 | <ashaw> | ? |
| 09:16 | <ashaw> | As in crypto operations it is often a great disadvantage to use floats. |
| 09:17 | <ashaw> | .. implementing a library to do ECC over prime fields .. |
| 09:18 | <jgraham> | No |
| 09:19 | <ashaw> | could there be one in the future? |
| 09:19 | <jgraham> | But modern javascript engines are well optimised |
| 09:19 | <ashaw> | yes. |
| 09:19 | <jgraham> | So there is a good chance that if the number can be represented as an integer internally it will be |
| 09:20 | <ashaw> | but I use both the upper and lower half of the float so -- a*b = x+y |
| 09:20 | <annevk> | ojan_, it will need to be lowercase beforeinput though :) |
| 09:20 | <ashaw> | a, b, 26 bit integers. |
| 09:21 | <ashaw> | can be done without floats, but not. |
| 09:21 | <ashaw> | instead we need casts between floats and ints and other such anoyances |
| 09:22 | <annevk> | so do you have some code that you think ought to run faster? |
| 09:22 | <ashaw> | yep. |
| 09:22 | <ashaw> | just give me a sec. |
| 09:22 | <jgraham> | If you think a language limitation is mking you slow you need to post to es-discuss⊙mo |
| 09:23 | <annevk> | cryptography is supposedly part of these benchmarks that come out every other month or so |
| 09:23 | <ashaw> | no, this can be optimised. and public crypto is not. |
| 09:24 | <jgraham> | annevk: That is a bit unfair. There are like 3 serious javascript benchmarks |
| 09:24 | <ashaw> | inputs x, y integers < 23 bits. |
| 09:24 | <ashaw> | var a= x*y |
| 09:25 | <ashaw> | var temp = Math.floor(v/0x4000000) |
| 09:25 | <annevk> | euhm, a pointer would suffice |
| 09:25 | <ashaw> | ok |
| 09:25 | <annevk> | but yeah, es-discuss is the place |
| 09:29 | <ashaw> | pastie http://pastie.org/1162362 |
| 09:30 | <ashaw> | and no it is not a language limitation. |
| 09:30 | <MikeSmith> | zcorpan_: thanks |
| 09:31 | <MikeSmith> | but I took a look at how Robin Berjon's respec tool handles this, and I'm now noticing that I have exactly the same problem when I try to save content generated with respec |
| 09:33 | <jgraham> | MikeSmith: For something that was designed to work with HTML, DOM is surprisingly poor at letting you parse and serialize HTML |
| 09:34 | <MikeSmith> | jgraham, yeah, seems so |
| 09:35 | <ashaw> | is this case optimised at opera, as my testing says that it is very slow compared to the native code. |
| 09:35 | <ashaw> | in all browsers. |
| 09:52 | <gsnedders> | ashaw: Opera, like every other ES engine, stores numbers as int32 when possible |
| 09:53 | <gsnedders> | (By and large, there are a few cases where it's stored as a double regardless) |
| 09:53 | <ashaw> | I know, the problem is that in the temp variable that is used nowhere else, you cannot store it as an int32. |
| 09:54 | <ashaw> | did you look at the code |
| 09:56 | <gsnedders> | No, but I have now. |
| 09:58 | <gsnedders> | Theoretically you could notice that the only operation it is used in starts with ToInt32() and just use int32 maths for it |
| 09:59 | <gsnedders> | However, I'm somewhat dubious this is going to be a serious bottleneck anytime soon |
| 10:02 | <ashaw> | I meant my code |
| 10:05 | <gsnedders> | I meant your code too |
| 10:05 | <ashaw> | ah, where does the ToInt32 |
| 10:05 | <ashaw> | . |
| 10:06 | <ashaw> | come from, where I got this has no ToInt32() |
| 10:07 | <gsnedders> | ashaw: ToInt32 is the first operation applyed to both LHS and RHS of the & operator in the ES engine. |
| 10:07 | <gsnedders> | (Well, this is untrue, to the value of the LHS and RHS) |
| 10:08 | <ashaw> | ah hah. |
| 10:08 | <ashaw> | the reason this is a problem is this runs in the core loop of most Public key crypto code. |
| 10:11 | <gsnedders> | Hmm, most benchmarks I've seen involving crypto haven't looked too slow for most real-world usage |
| 10:13 | <hsivonen> | the terminology around run/execute/evaluate a script is sad |
| 10:13 | <ashaw> | 0.2s per encryption. |
| 10:13 | <hsivonen> | within the spec |
| 10:13 | <hsivonen> | and within Gecko |
| 10:13 | <hsivonen> | but especially across the spec and Gecko |
| 10:15 | <ashaw> | yes. |
| 10:15 | <ashaw> | but not fast enough to be useful as it could be. |
| 10:19 | <annevk> | hsivonen, attributes are unordered, right? |
| 10:25 | <jgraham> | annevk: YEs |
| 10:25 | <jgraham> | s/E/e/ |
| 10:28 | <annevk> | I guess for now I will comment out all the Attr stuff and then add the new things |
| 10:33 | <hsivonen> | annevk: they are logically unordered but .attributes must stay in a stable order |
| 10:34 | <hsivonen> | annevk: IIRC, in Gecko, the iteration order of attributes is the reverse of the source order and in IE the iteration order depends on how the attribute names get hashed |
| 10:35 | <hsivonen> | actually, it's the reverse of the source order in HTML and the source order in XML in Gecko |
| 10:36 | <hsivonen> | annevk: IIRC, Flash Player depends on the order in which the attributes of <embed> are passed to it |
| 10:37 | <hsivonen> | so is there now both IE9 beta and PP5? do they have different engine snapshots? |
| 10:40 | <jgraham> | I recall he same thing about the flash player now that you mention it |
| 10:42 | <MikeSmith> | does foolip not come on IRC any more? |
| 10:43 | <annevk> | isn't embed in a-z order or something? |
| 10:52 | <ashaw> | I have discovered a bug in the code that I pasted ( in my edits to the paste) |
| 10:52 | <ashaw> | new code at http://pastie.org/1162427 |
| 10:53 | <ashaw> | this code is run in a tight loop and in my ECC code 70% of time is in this code. |
| 10:53 | <ashaw> | is there any way to make this faster in any browsers. |
| 10:54 | <ashaw> | ECC: elliptic curve cryptography. |
| 10:55 | <jgraham> | Argh. The [ython logging module is insane |
| 10:55 | <jgraham> | *python |
| 10:55 | <jgraham> | insane |
| 10:56 | <jgraham> | Would it be so hard to write something like: |
| 10:56 | <jgraham> | logger = logging.logger(dest=sys.stdout, level="info") |
| 10:56 | <jgraham> | and get something that works? |
| 10:56 | <jgraham> | Rather than go through insane contortions |
| 11:08 | <timdown> | Hi all. I have some concerns about the text selections part of the HTML5 spec. I'd like to suggest some changes and am unsure of the most effective way of doing this. |
| 11:08 | <timdown> | Suggestions? |
| 11:13 | <Philip`> | timdown: The W3C Bugzilla is probably the best way for suggesting changes |
| 11:13 | <Philip`> | The WHATWG mailing list might be better if you want more discussion about it |
| 11:13 | <annevk> | timdown, changes that are not backwards compatible can probably not be made; new features are probably best done by convincing implementors |
| 11:15 | <timdown> | My concerns are all about the spec not being compatible with current browser implementations |
| 11:16 | <timdown> | I assume that's what you meant by backwards compatible, since previous versions of HTML have no text selection API |
| 11:18 | <annevk> | oh great |
| 11:18 | <annevk> | whatwg⊙wo would be best for that I think |
| 11:18 | <timdown> | I've filed a couple of bugs in Bugzilla, about a week ago, with no response. Is that normal? If so, I'll go quietly. |
| 11:19 | <annevk> | the editor is on a break |
| 11:19 | <timdown> | Ah, fair enough. |
| 11:19 | <annevk> | and bugs have taken up to three months before |
| 11:20 | <annevk> | there's lots of tiny issues everywhere so it takes a bit of time |
| 11:20 | <Philip`> | Mailing list posts have taken up to three years for a response |
| 11:20 | <timdown> | Great :) |
| 11:20 | <timdown> | OK, I'll be more patient. |
| 11:20 | <annevk> | Philip`, yeah, but not emails pointing out compatibility problems I think |
| 11:21 | <annevk> | timdown, what are the bug numbers? |
| 11:21 | <timdown> | 10583, 10624 |
| 11:22 | <timdown> | Actually that last one is only 3 days old, so I've exaggerated a bit. |
| 11:24 | <annevk> | Selection.toString() sounds like innerText |
| 11:24 | <timdown> | IE's innerText presumably |
| 11:25 | <timdown> | doesn't WebKit implement innerText as an alias for textContent? Or did I imagine that? |
| 11:25 | <Ms2ger> | It's not that simple |
| 11:25 | <annevk> | timdown, reportedly not |
| 11:26 | <Ms2ger> | innerText apparently even strips out the contents of script elements |
| 11:26 | <annevk> | timdown, e.g. just like in your example <script> is excluded |
| 11:26 | <annevk> | timdown, formatting is even taken into account |
| 11:26 | <timdown> | yes |
| 11:27 | <annevk> | textContent does none of that, it simply takes descendant Text nodes |
| 11:27 | <timdown> | I knew that innerText and textContent differed and that innerText stripped more out, but I forget the details. |
| 11:27 | <timdown> | Yes. |
| 11:28 | <timdown> | Much like Range.toString(). |
| 11:29 | <timdown> | Ah, maybe it's Opera that aliases innerText to textContent. |
| 11:30 | <annevk> | prolly |
| 11:31 | <annevk> | we should fix that, but ideally someone first writes down how |
| 11:34 | <hsivonen> | jgraham: I whined to Canonical through paid support about the jumpiness under load since the first half of August problem |
| 11:37 | <timdown> | annevk, yes. Trying to write down the rules it should follow was not very appealing so I didn't do it for the bug report. |
| 11:37 | <timdown> | 10624 I think is maybe more important |
| 11:39 | <jgraham> | hsivonen: Did anything happen? |
| 11:40 | <hsivonen> | jgraham: I whined a few minutes ago. let's see. |
| 11:40 | <hsivonen> | since I came back from vacation, I've been too busy with Gecko work to complain to Canonical about all the things that hinder my Gecko hacking |
| 11:41 | <kennyluck> | karlcow: re: Mozilla Audio data API, W3C launched the Audio Incubator Group for that -> http://www.w3.org/2010/04/audio/audio-incubator-charter |
| 11:42 | <karlcow> | kennyluck: ah thanks. I'll check that. |
| 11:42 | <karlcow> | the issue so far I have identified with Audio data API is that it is not a sound API, but an API to manipulate uploaded sounds. |
| 11:44 | <karlcow> | Canvas gives the ability to create 2D graphics, I wonder if it would be possible to create sounds directly. modulation, envelop, frequency, tone, etc. |
| 11:44 | <karlcow> | maybe I have missed something |
| 11:44 | <kennyluck> | karlcow: I believe it is a sound API. See http://schepers.cc/web-audio-goes-to-eleven |
| 11:44 | <karlcow> | I will read more carefully |
| 11:45 | <karlcow> | kennyluck: thanks again :) |
| 11:45 | <kennyluck> | (the example looks extremely interesting, but I don't yet have time to test it. Cause you need to compile some branch apparently) |
| 11:46 | <annevk> | Ms2ger, is this concept of ID/class attributes useful? |
| 11:47 | <annevk> | Ms2ger, I think we should just say that the id="" attribute defines the element-ID and the class="" attribute defines the element-classes |
| 11:47 | <Ms2ger> | On which elements? |
| 11:50 | <annevk> | all of them |
| 11:51 | <annevk> | failing that, it would be "attribute X" defines the ID and "attribute Y" defines the classes, depending on the markup language et al |
| 11:52 | <annevk> | no need to say anything at the Attr level since this is an Element thing |
| 12:14 | <annevk> | AttrExodus is quite a big undertaking; hopefully someone will implement it |
| 12:37 | <annevk> | done |
| 12:41 | <MikeSmith> | does Opera support CSS columns stuff? |
| 12:42 | <annevk> | not yet |
| 12:45 | <MikeSmith> | oh |
| 13:09 | <annevk> | Ms2ger, for setAttribute maybe "starts with xmlns" should be banned? |
| 13:10 | <Ms2ger> | Probably |
| 13:10 | <annevk> | and I guess setAttributeNS should have all the same restrictions as createElementNS |
| 13:11 | <annevk> | Peter`, got any closer on compareDocumentPosition? |
| 13:59 | <MikeSmith> | hmm, http://ajaxian.com/archives/simulating-hover-and-double-clicks-with-pure-css-on-mobile-devices is interesting |
| 13:59 | <MikeSmith> | though, '“dblclick” is a native JavaScript event' |
| 14:00 | <hsivonen> | oh. IE9 beta replaced IE8 |
| 14:00 | <hsivonen> | they didn't make that too clear ahead of installation |
| 14:00 | <hsivonen> | oh well |
| 14:01 | <hsivonen> | at least I still have IE8 for testing on XP |
| 14:04 | <hsivonen> | bah. the script error IE9 gives on this test case is not helpful: http://hsivonen.iki.fi/test/moz/sheet-blocking-script4.php |
| 14:26 | <miketaylr> | hsivonen: yeah you had to install ie9 preview 5 for a chromeless install that won't wipe out ie8 |
| 14:28 | <smaug____> | how do I file bugs on Opera? |
| 14:28 | <smaug____> | or where |
| 14:28 | <Philip`> | https://bugs.opera.com/wizard/ |
| 14:29 | <jgraham> | smaug____: Let me know the bug number you get back |
| 14:38 | <smaug____> | jgraham: "An error occurred while processing your bug report" :( |
| 14:40 | <smaug____> | jgraham: anyway, the bug is that :hover isn't consistent |
| 14:40 | <smaug____> | jgraham: http://mozilla.pettay.fi/moztests/hover.html |
| 14:40 | <smaug____> | jgraham: try to move mouse from parent document to iframe |
| 14:40 | <smaug____> | :hover state is kept properly |
| 14:41 | <smaug____> | if you move mouse outside window and then straight to the iframe, parent document doesn't get :hover state |
| 14:51 | <smaug____> | jgraham: any idea why I can't file bugs? |
| 15:01 | <hsivonen> | does anyone have an explanation of what exactly IE8 and IE9 are doing here: http://hsivonen.iki.fi/test/moz/sheet-blocking-script4.php ? |
| 15:01 | <hsivonen> | Does IE block the parser on <link> until the sheet has loaded? |
| 15:02 | <hsivonen> | AFAICT, my other tests in the sheet-blocking-script series suggest that IE generally continues parsing after <link> |
| 15:11 | <jgraham> | smaug____: (sorry my computer just died) |
| 15:11 | <jgraham> | smaug____: No idea why you can't file bugs |
| 15:11 | <jgraham> | I can ask |
| 15:20 | <jgraham> | smaug____: No, I'm told it should be working |
| 15:21 | <jgraham> | So erm, I'm not that useful |
| 15:43 | <hsivonen> | what's the deal with the blue outline in Opera at http://css-tricks.com/examples/ExpandingCaptionedImages/ ? |
| 15:43 | <hsivonen> | does outline: none; not suppress the outline in Opera? |
| 15:46 | <hsivonen> | eww. Chrome and Chromium don't anti-alias -webkit-transform rotations |
| 15:50 | <Rik`> | hsivonen: I think outline: 0; works better |
| 15:53 | <hsivonen> | Kudos to Opera for not visibly lowering the rendering quality during the transition |
| 15:54 | <hsivonen> | and for not rendering rotated text with hinting enabled (eww) |
| 15:56 | <hsivonen> | Rik`: at least Opera doesn't have the hideous outline when clicking. in a way, it makes sense not to let authors take the outline away when using the keyboard |
| 15:56 | <hsivonen> | Now that IE9 beta is out, I should probably do some community service and map out how the mode switching works |
| 15:58 | <Rik`> | outline: 0; seems to do nothing either so I guess you can't cancel it |
| 15:59 | <Rik`> | hsivonen: is it supposed to include IE6, 7 and 8 modes ? |
| 16:00 | <hsivonen> | Rik`: I think it has 5.5, 7, 8, 9 and something called 9 7 Compatibility View |
| 16:00 | <hsivonen> | dunno what the last one is |
| 16:01 | <Rik`> | how can they keep up with the maintenance ? |
| 16:01 | <hsivonen> | maybe the last thing is what happens if you press the compat mode button |
| 16:01 | <hsivonen> | maybe it means 5.5 or 7 depending on doctype |
| 16:02 | <hsivonen> | or something |
| 16:02 | <hsivonen> | oh, and I'd expect the 8 mode to have almost standards and standards within it |
| 16:02 | <hsivonen> | I wonder if the 9 mode also still has almost standards and standards |
| 16:02 | <hsivonen> | testing this is gonna be so much fun! |
| 16:03 | <hsivonen> | I think IE6 Standards Mode is the only mode that has been dropped from IE since 5.5 |
| 16:05 | <hsivonen> | awesome! There is indeed IE9 Standards and IE9 Almost Standards |
| 16:12 | <annevk> | yeah, so far you cannot disable our custom outline |
| 16:12 | <annevk> | we might add it though |
| 16:13 | <annevk> | IE9 Almost Standards? |
| 16:13 | <annevk> | so only quirks mode is incorrect? |
| 16:13 | <annevk> | i.e. quirks mode triggers the gazillion other modes |
| 16:15 | <hsivonen> | IE9 Almost Standards is what you get with an almost standards doctype and without X-UA-Compatible or any blacklists or overrides |
| 16:16 | <hsivonen> | zcorpan_, Philip`: I finally got around to reviewing Microsoft's IE8 flowchart from March, and I believe mine is more accurate |
| 16:21 | <hsivonen> | what's quirky, standards or almost standards hasn't changed since IE8 |
| 16:31 | <annevk> | I guess I'll move namespaceURI/localName/prefix from Node to Element and then request publication next week |
| 16:31 | <annevk> | prolly marked NodeExodus or some such |
| 16:31 | <annevk> | might as well put some of the bold ideas in there |
| 16:44 | <hsivonen> | hmm. does Microsoft not use the Public Suffix List? |
| 16:44 | <hsivonen> | iki.fi is on the list, but IE9 highlights "iki.fi" instead of "hsivonen.iki.fi" |
| 17:04 | <annevk> | hmm |
| 17:04 | <annevk> | did anyone else just get this email: 'Welcome to the "ietf-types" mailing list'? |
| 17:34 | <hsivonen> | if http://lostworldsfairs.com/moon/ is demoing WOFF, why does the font not look the same in Mac Minefield as it does in IE9? |
| 17:34 | <Philip`> | Because they have different font renderers? |
| 17:36 | <Philip`> | Or because Typekit does bad UA sniffing? |
| 19:38 | <hsivonen> | Philip`: I meant different on the level of showing a different font (from the OS font set, I think) |
| 21:01 | <MikeSmith> | jarib: you around? |
| 21:02 | <MikeSmith> | wanted to ask if you have interest in helping out with testing effort among Webapps, CSS, HTML, SVg WGs |
| 21:05 | <AryehGregor> | Oh, IE9 beta released. I'm really behind. (Well, it was only yesterday, but still.) |
| 21:06 | AryehGregor | doesn't have time to read the whole post |
| 21:07 | <MikeSmith> | me has not made time to look in detail either |
| 21:07 | <MikeSmith> | but seems like generally a really Good Thing |
| 21:08 | <MikeSmith> | and also seems good that they are immediately continuing on the next Platform Preview in parallel with the beta |
| 21:08 | <MikeSmith> | which seems like they are following the same dev model as other porjects |
| 21:09 | <MikeSmith> | (semi)stable release on one hand, dev releases on the other |
| 21:11 | <jarib> | MikeSmith: i do have interest. time might be more of an issue, though :) |
| 21:11 | <jarib> | anything in particular you're looking to do? |
| 21:11 | <MikeSmith> | jarib: just looking to have as much help with it as we can get -- especially testing-savvy people |
| 21:16 | <MikeSmith> | hey what happened to JonathanNeal? |
| 21:16 | <MikeSmith> | after his triumphant invention of the HTML5 gang sign, seems like he went quiet |
| 21:18 | MikeSmith | points jarib at DM |
| 21:21 | <paul_irish> | MikeSmith: he has a tendency to do that. :/ |
| 21:22 | <paul_irish> | he met a girl.. said he found his wife.. said he was quitting webdev.. took her to hawaii, where she dumped him. |
| 21:22 | <MikeSmith> | works better than e-mail |
| 21:22 | <paul_irish> | so now he's back sort of. kind of. sometimes. |
| 21:22 | <MikeSmith> | paul_irish: ah, oK |
| 21:22 | <MikeSmith> | didn't know about the draam |
| 21:23 | <MikeSmith> | I do empathize |
| 21:23 | <paul_irish> | aye |
| 21:23 | <paul_irish> | isnt there a style control thing where you can have an element tree ignore any cascading styles? |
| 21:31 | <MikeSmith> | TabAtkins: you were talking a while back about work on developing a test harness we could use for cross-spec testing |
| 21:32 | <MikeSmith> | but I don't remember what the outcome of that was |
| 21:34 | <jgraham> | MikeSmith: What do you mean "harness". I think I have confused things by using the word "harness" where "framework" would be more apropos |
| 21:34 | <jgraham> | TabAtkins was developing some thing for loading tests and recording the results |
| 21:34 | <jgraham> | But last I heard it only supported visual tests |
| 21:35 | jgraham | also isn't sure what the relationship to W3TestRunner is |
| 21:37 | <MikeSmith> | jgraham: I mean software that lets us run the tests across browsers and across OSes/platforms + a mechanism for collecting the results and storing them in a way that lets us generates reports of the results |
| 21:38 | <MikeSmith> | jarib: we have tests in different formats, I think |
| 21:38 | <MikeSmith> | e.g, reftests from Mozilla |
| 21:39 | <jgraham> | MikeSmith: Right, taht sounds like a harness |
| 21:39 | <MikeSmith> | jgraham: so what means "framework"? |
| 21:39 | <jgraham> | I have also used the word harness to mean "a framework for writing tests" |
| 21:39 | <MikeSmith> | ah |
| 21:40 | <jgraham> | MikeSmith: Like a javascript file that provides a set of functions for writing tests and (possibly) reporting the results (visually, to a harness, stc.) |
| 21:40 | <jgraham> | s/stc/etc/ |
| 21:40 | <MikeSmith> | ah, I see |
| 21:40 | <MikeSmith> | yeah |
| 21:42 | <jarib> | i'm guessing existing tests do not report results uniformly |
| 21:42 | <jarib> | so it's a question of whether tests should be rewritten/required to do that, or introduce another layer is needed to extract the results |
| 21:43 | <jarib> | s/is needed// |
| 21:43 | <MikeSmith> | ok |
| 21:44 | <jgraham> | jarib: For the HTML WG we seem, slowly, to be standardising around a single API for javascript tests |
| 21:44 | <jgraham> | Hopefully we will use reftests for most other things |
| 21:45 | <jgraham> | and then there will likely be a few visual tests for things that are inconvenient to automate |
| 21:46 | <jgraham> | If we can keep it that simple, making a harness to read the results is quite easy, although actually auomating reftests in a cross browser fashion isn't really possible yet |
| 21:46 | <jgraham> | *automating |
| 21:46 | <jarib> | what are reftests? |
| 21:46 | <jarib> | any links? |
| 21:46 | <jgraham> | Two pages, a test and a reference, should render identially |
| 21:46 | <jgraham> | *identically |
| 21:46 | <jarib> | aha |
| 21:47 | <jgraham> | http://wiki.csswg.org/test/reftest |
| 21:47 | <jarib> | have you considered using any of the existing JS testing frameworks? |
| 21:48 | <jarib> | (for the js tests, obviously) |
| 21:48 | <jarib> | i.e. http://code.google.com/p/js-test-driver/ would give you a lot for free |
| 21:48 | <TabAtkins> | paul_irish: No, no way to stop cascading. You could set every property to explicit values, if you felt like. |
| 21:49 | <TabAtkins> | MikeSmith: Yes, generally. I am right now bodging a previously-developer harness into working just for the CSSWG so we can get impl reports in time. |
| 21:49 | <MikeSmith> | ok |
| 21:49 | <MikeSmith> | good to hear |
| 21:50 | <TabAtkins> | But then I'll be writing one myself to work with the repo-slurper / review comment tracker I'm developing as well. |
| 21:50 | <TabAtkins> | I'm just being very slow at it, which is no good. The sooner I finish it, the sooner I can start actually hacking on webkit. |
| 21:52 | <TabAtkins> | So, I'll be building a few harnesses. First one is for self-describing tests, because they're the easiest. |
| 21:52 | <TabAtkins> | Then probably a js-test one using the framework that the testing WG is doing. |
| 21:52 | <TabAtkins> | Then reftest, if I can figure out how. |
| 21:52 | <jgraham> | "the testing wg"? |
| 21:53 | <TabAtkins> | Whatever you guys are. I forget the name. |
| 21:53 | <jgraham> | oh the HTML test people? |
| 21:53 | <TabAtkins> | Yeah. |
| 21:53 | <jgraham> | Ah, I see |
| 21:53 | <jgraham> | We kindof have a very basic harness the microsot contributed |
| 21:53 | <jgraham> | I wouldn't be sad to see it replaced though |
| 21:54 | <TabAtkins> | Well, you have some js functions that do test-y things that I can hook into. |
| 21:54 | <jgraham> | Yeah |
| 21:54 | <paul_irish> | TabAtkins: ok. found an old thread were dean edwards proposed sandboxing styles like that but i guess it never ended up making it in. |
| 21:54 | <TabAtkins> | paul_irish: Right. |
| 21:54 | <paul_irish> | thx |
| 21:54 | <jgraham> | But there is also a html file that loads tests in iframes and allows you to mark the result |
| 21:55 | <TabAtkins> | So what is that, a self-describing basically? |
| 21:55 | <jgraham> | I think it will hook into the js at some point in the near future |
| 21:55 | <jgraham> | TabAtkins: Yeah. The feature set at the moment is pretty poor |
| 21:55 | <jgraham> | e.g. for reftests if you can't actually provide automation (hard, requires proprietary APIs) |
| 21:56 | <jgraham> | you should at least provide a harness that allows you to view test, view ref, flip between the two, and mark the result |
| 21:56 | <TabAtkins> | Right, reftests are a problem until we have dependable ability to draw DOM elements into canvas. |
| 21:56 | <TabAtkins> | Yeah, that's the limit of what humans can do right now. |
| 21:58 | <jgraham> | Mostly I don't think this is a big problem because vendors are working on their own reftest runners |
| 21:58 | <jgraham> | Or have them already |
| 21:58 | <TabAtkins> | Sure. I think we all have the ability already, if we haven't built an actual runner for them. |
| 21:58 | <TabAtkins> | But you can't crowdsource a reftest runner. |
| 21:59 | <TabAtkins> | s/ a / with a / |
| 22:00 | <jgraham> | Well you don't really need to crowdsource, right? Or what do you envision crowdsourcing? |
| 22:00 | <TabAtkins> | Letting arbitrary people run tests themselves and report results. |
| 22:00 | <TabAtkins> | Like browserscope and the 20+ other similar efforts. |
| 22:01 | <TabAtkins> | The results are somewhat less trustworthy, but numbers can make up for that. |
| 22:02 | <jgraham> | It's not clear to me why it is interesting to do that |
| 22:02 | <jgraham> | I mean letting people run the tests is good for transparency |
| 22:02 | <jgraham> | But should have rather little other value |
| 22:03 | <jgraham> | Assuming that they are easy enough for vendors to run |
| 22:03 | <jgraham> | (CSS2.1 tests fail here) |
| 22:03 | <TabAtkins> | Yeah, that's the thing. |
| 22:03 | <TabAtkins> | If you can't automate, then the time necessary to run them can be prohibitive. |
| 22:03 | <TabAtkins> | But ordinary people donating some time to it is easier. |
| 22:07 | <jgraham> | Yeah, that is a point |
| 22:08 | <jgraham> | Really you want to avoid being in a situation where you are writing tests that vendors can't run economically |
| 22:08 | <TabAtkins> | Sure. That's why reftests were invented, to solve this problem for CSS. |
| 22:08 | <jgraham> | Because the value proposition to vendors of the tests is regression tracking |
| 22:33 | <MikeSmith> | karlcow: http://developer.apple.com/library/mac/#documentation/UserExperience/Conceptual/DictionaryServicesProgGuide/schema/schema.html#//apple_ref/doc/uid/TP40006152-CH4-SW1 |
| 22:33 | <MikeSmith> | "Dictionary Markup" for Apple Dictionary format |
| 22:34 | <MikeSmith> | great stuff |
| 22:34 | <MikeSmith> | includes a "gi" element for marking up gaiji |
| 22:34 | <MikeSmith> | and lots of other stuff |
| 22:35 | <MikeSmith> | seems like it could be worthy of standardizing on as a cross-app dictionary format |
| 22:41 | <othermaciej> | hi all |
| 22:41 | <TabAtkins> | yo |
| 22:41 | TabAtkins | drinks too much Diet Coke. He has a problem. |
| 22:41 | <othermaciej> | MikeSmith: so I did some checking on how bugmail would compare if we got new bug notifications instead of keyword notifications |
| 22:41 | <MikeSmith> | ok |
| 22:41 | <othermaciej> | MikeSmith: we'd get about twice as much bugmail per month, but a greater proportion of it would actually be relevant |
| 22:41 | <othermaciej> | I sent mail to the WG to see how people feel about that |
| 22:42 | <MikeSmith> | hai |
| 22:42 | <othermaciej> | TabAtkins: interested in what you think also, since you expressed the opinion that bugzilla currently hides things too much |
| 22:42 | <jgraham> | Could we get the bug mail for new bugs after some delay? |
| 22:42 | <jgraham> | Like 6 hours |
| 22:42 | <jgraham> | So obvious spam would be closed and never be sent to the list |
| 22:42 | <MikeSmith> | othermaciej, I still suggest considering to have the messages go to public-html-issue-tracking instead |
| 22:42 | <jgraham> | Might need some code |
| 22:43 | <MikeSmith> | having automated bug notifications go to the group's technical discussion list, for a group like this, seems less than ideal to me |
| 22:43 | <MikeSmith> | but I leave it up to the chairs to figure out what works best for the group |
| 22:43 | <TabAtkins> | othermaciej: Thanks! Commenting now. |
| 22:47 | <othermaciej> | MikeSmith: I have mixed feelings, but right now I feel that we are getting too little technical content on public-html rather than too much |
| 22:47 | <othermaciej> | MikeSmith: also, I was a bit skeptical of it for WebApps WG but it actually seems to work kind of OK |
| 22:47 | <othermaciej> | MikeSmith: I think work could go better if more people have a chance to comment on bugs before they are at the escalation stage |
| 22:48 | <MikeSmith> | all true |
| 22:49 | <MikeSmith> | though WebApps is a very different WG than the HTML WG |
| 22:50 | <MikeSmith> | othermaciej: my main concern is frankly about how it affects implementor participation on the list |
| 22:51 | <othermaciej> | MikeSmith: well, Jonas and Tab expressed an interest in seeing more data about new bugs |
| 22:51 | <othermaciej> | having looked at the past month of bugs, I would have also liked to see more of them |
| 22:51 | <MikeSmith> | and PhilipJ expressed the opposite |
| 22:52 | <MikeSmith> | and you are not going to hear from the implementors who have already checked out because of perceived noise |
| 22:52 | <jgraham> | FWIW I think it will be better than what we currently have |
| 22:52 | <jgraham> | But that is not a great endorsement |
| 22:52 | <MikeSmith> | they are not going to post to the list because they have already filtered it or whatever |
| 22:52 | <MikeSmith> | already ignoring it, I mean |
| 22:52 | <MikeSmith> | I am not saying that is absolutely what has already happened |
| 22:53 | <othermaciej> | well, whatwg@ gets *much* more mail per week than we do currently, and people don't check out of that |
| 22:53 | <othermaciej> | I think it's about the kind of content, not just the volume |
| 22:53 | <MikeSmith> | yes |
| 22:53 | <othermaciej> | but I'll be curious to hear phillipj's feedback about my suggestion |
| 22:53 | <MikeSmith> | yeah, me too |
| 22:53 | <MikeSmith> | he doesn't seem to be around on IRC these days |
| 22:54 | <MikeSmith> | he used to here on #whatwg regularly |
| 22:54 | <othermaciej> | I think keyword changes come off as extra spammy because people tend to do them in batches |
| 22:54 | <othermaciej> | so even though it's less volume than new bugs, it clusters more |
| 22:54 | <MikeSmith> | hmm, yeah, true |
| 22:54 | <MikeSmith> | that is the pattern |
| 22:55 | <TabAtkins> | MikeSmith: There's a difference between volume and noise. High-volume but low-noise can be hard to follow sometimes, but it's infinitely better than a high-noise list of somewhat lesser volume. |
| 22:56 | <MikeSmith> | TabAtkins: different people have very different perceptions about what they consider noise |
| 22:57 | <MikeSmith> | but we get into a philosophical discussion -- I think chatting with PhilipJ would help in this case |
| 22:57 | <TabAtkins> | Sure, but I suspect most people would agree that new bugs are much less noisy than random a11y keyword changes. |
| 22:57 | <TabAtkins> | Anyway, I sent my feedback. |
| 22:57 | <MikeSmith> | yeah, agreed |
| 23:56 | <gsnedders> | Who the hell drinks two bottles of vodka on their own in around an hour? |
| 23:56 | gsnedders | is kinda pissed off at having to deal with what flatmate inflicted on themself |