| 00:05 | Philip` | hears a cat making really quite disturbing noises outside his window |
| 00:07 | <Dashiva> | This suggests your window is a blog |
| 00:08 | <Philip`> | It has enough security vulnerabilities to be a blog |
| 00:08 | <Philip`> | Someone ought to invent transparent bricks - those would be much better at keeping intruders out |
| 00:51 | <Hixie> | Philip`: fair enough |
| 01:31 | <Hixie> | wow, that thread was easy to deal with |
| 01:31 | Hixie | switches to sentences around to fix 13 e-mails |
| 01:32 | <Hixie> | two |
| 01:32 | <Hixie> | Lachy: the attribute name was intentionally stupid so that people would suggest new names |
| 01:34 | <Philip`> | I think the attribute should be named "!" |
| 01:34 | <Hixie> | well that would be exciting |
| 01:34 | <Philip`> | so you can write <img! src=... alt=..> for an important image |
| 01:34 | <Hixie> | in SO many ways |
| 01:35 | <Hixie> | well you couldn't do _that_ |
| 01:35 | <Hixie> | you'd be creating an <img!/> element |
| 01:35 | <Philip`> | Whoops |
| 01:36 | <Hixie> | but i do like the idea of <img ! src='photo' alt=> |
| 01:36 | <Hixie> | 'er |
| 01:36 | <Philip`> | That's spoilt my plan :-( |
| 01:36 | <Hixie> | but i do like the idea of <img ! src='photo' alt=Photo> |
| 01:37 | <Hixie> | the only things that end a tag name are >, /, whitespace, and EOF |
| 01:37 | <Philip`> | Hmm, it's even well-formed XML |
| 01:37 | <Philip`> | Oh |
| 01:37 | <Philip`> | No it's not |
| 01:37 | <Philip`> | I was accidentally testing it as text/html instead |
| 01:38 | Philip` | thought that seemed a bit odd |
| 01:59 | <Philip`> | annevk: I don't know if I've seen "interoperability" defined anywhere, so I'll make something up and say that interoperability is when authors can write pages that work in all UAs and developers can write UAs that work on all pages |
| 02:00 | <Philip`> | Then the linked-image case is interoperable regardless of how UAs handle <a><img alt=""></a>, because authors can write <a><img alt="Button label"></a> (which will work in all UAs) and developers can auto-generate some link text for <a><img alt=""></a> (which will work on all pages) |
| 02:02 | <Philip`> | (Not all authors *will* do that, and not all UAs *will* do t'other, but that isn't a lack of interoperability and it won't have negative effects beyond that page or that UA) |
| 02:05 | <Hixie> | andersca: spec updated to handle all feedback so far, though not what we discused earlier |
| 02:07 | <andersca> | Hixie: awesome - I'm fixing the case where you cancel the load of the implicit resource |
| 02:07 | <Hixie> | that's in the spec now |
| 02:08 | <andersca> | oh, cool |
| 02:08 | andersca | reloads |
| 02:11 | <andersca> | Hixie: excellent |
| 02:11 | <andersca> | Hixie: what about step 2 of "If the resource being loaded was not loaded from an application cache, but it was loaded using HTTP GET or equivalent" |
| 02:12 | <andersca> | Hixie: if the implicit resource fails to load |
| 02:13 | <Hixie> | that's handled by step 21 of the application cache update process |
| 02:16 | <andersca> | Hixie: what if the cache manifest didn't change, then step 21 won't be reached |
| 02:18 | <Hixie> | andersca: oh you mean if this is a new implicit entry but the cache is already there... |
| 02:18 | <Hixie> | andersca: hmm |
| 02:18 | <andersca> | yes |
| 02:18 | <Hixie> | andersca: sure, will fix that |
| 02:18 | <andersca> | Hixie: what should happen in that case? |
| 02:18 | <Hixie> | i guess it just doesn't get added |
| 02:18 | <andersca> | yeah |
| 02:19 | <andersca> | and there will be no error, since there's no update |
| 02:23 | <andersca> | Hixie: maybe the document should no longer be associated with the cache in that case |
| 02:36 | <Hixie> | andersca: fixed the infinite loop |
| 02:36 | <andersca> | Hixie: excellent |
| 04:24 | <Hixie> | http://www.w3.org/2002/09/wbs/40318/dcp/ (as yet unannounced, i think, so don't vote on it yet) is going to be interesting. i wonder why that particular e-mail was chosen as a summary, for instance. and what weird effects there will be in casting the question in such a binary way. |
| 04:36 | <Hixie> | Lachy, annevk, anyone: do you remember how to get cvs access to dev.w3.org? |
| 04:37 | <heycam> | Hixie, i think you need to mail an ssh public key of yours to a Team person |
| 04:37 | <Hixie> | oh they still don't have a form for this? |
| 04:37 | <Hixie> | ok |
| 04:38 | <heycam> | not afaik |
| 04:38 | <Hixie> | do you remember what kind of key they want? |
| 04:38 | <Hixie> | 1024 bit rsa? |
| 04:38 | <heycam> | don't know if they needed any specific key size / alg |
| 04:38 | <Hixie> | i seem to recall there was some problem with the first key i sent |
| 04:38 | <Hixie> | but this was years ago |
| 05:08 | <MikeSmith> | Hixie, heycam : only criterion for the key is that it needs to be SSH2 |
| 05:08 | <MikeSmith> | SSH protocol 2 |
| 05:09 | <Hixie> | k thanks |
| 05:09 | <MikeSmith> | it can rsa or dsa of any length |
| 07:11 | <annevk> | Philip`, you start getting content that relies on a particular UA behavior which is the problematic bit |
| 07:11 | <annevk> | Philip`, so it seems better to make it non-conforming |
| 07:59 | zcorpan_ | considers setting up a filter to mark "alt" emails as read |
| 08:02 | <annevk> | hmm, all i tried to say was that HTML5 is not against "accessibility" and now i get ranted for that :( |
| 08:03 | <annevk> | seems like quite a hopeless case indeed |
| 09:26 | <jgraham_> | hsivonen: Thank you for pointing out that accessibility is clearly not a "cornerstone of the web" for any reasonable use of the metaphor. |
| 09:27 | <jgraham_> | People really have to distinguish between what is actually true and what they would like to be true |
| 09:27 | <hsivonen> | indeed |
| 09:46 | <Hixie> | important="" isn't really the right name... we need something that means either: |
| 09:46 | Hixie | ponders |
| 09:47 | <Hixie> | i don't even know really want we want to convey |
| 09:47 | <Philip`> | You want to convey that the alt text is a description rather than an alternative presentation of the image |
| 09:47 | <Philip`> | (I think) |
| 09:48 | <Philip`> | Uh |
| 09:48 | <Philip`> | s/presentation/representation/ |
| 09:49 | <Hixie> | not even really a description |
| 09:49 | <Philip`> | Hmm, but it should also convey that the image is non-decorative and no textual representation can exist, since those are the only cases where it's allowed |
| 09:50 | <Hixie> | we want to convey that the alt is that the image is known to be not decorative, and that alt is giving the category of image in question, not a replacement representation |
| 09:50 | <Hixie> | a textual representation might be able to exist, we just don't know what it is |
| 09:50 | <Philip`> | (I mean a description of the image as an object (e.g. "Photo"), rather than a description of the content of the photo) |
| 09:50 | <Hixie> | e.g. in the case of a photo uploaded by a blind user or hsivonen's image tool |
| 09:52 | <jgraham_> | If the text was a description of the image it could go in title |
| 09:52 | <jgraham_> | Oh you made that distinction |
| 09:54 | <Philip`> | Does it have to be an attribute, rather than e.g. <img src=... alt="Important image: Photo">? |
| 09:54 | <Philip`> | (which would not be machine-extractable information, but I'm not sure if machines need to extract this information) |
| 09:55 | <jgraham_> | What's the rationale for having this attribute at all? How would UAs use it? What problems would it fix? |
| 09:55 | <hsivonen> | Philip`: doesn't the heading and preceding paragraph in the image report already make the point that the images are important for the function of the tool? |
| 09:56 | <hsivonen> | Philip`: if a user can't use it but ends up wading through the content, how does "Important image: Photo" a dozen times help? |
| 09:57 | <jgraham_> | In particular how would this attribute deal with the problem of authoring tools adding bogus content to pass validation? |
| 09:58 | <Hixie> | the idea is that the attribute would have one effect on UAs |
| 09:59 | <Hixie> | namely, when the <img> element is being rendered as text instead of as an image, the image would be rendered in a way that clearly indicates that there was an image there |
| 09:59 | <Hixie> | instead of just silently replacing the image with text |
| 09:59 | <Hixie> | so <p>Hello <img src=world.png alt="World"> how are <img important src=you.png alt="Word">?</p> would render as something like: |
| 10:00 | <Philip`> | hsivonen: The alternative (i.e. <img src=...>) means the UA will have to apply heuristics, so it might ignore all the images or read all their filenames, or might ignore all spacer.gifs since they look decorative, which seems like a worse user experience than just telling the user that there's an image (as would happen with alt="Image") |
| 10:00 | <Hixie> | Hello World how are [IMAGE: Word]? |
| 10:01 | <Hixie> | the idea being that you can still convey some sort of information, just not enough to act as a replacement |
| 10:01 | <hsivonen> | Philip`: are conceivable content-based heuristics worse? |
| 10:01 | <Philip`> | hsivonen: Users you can't see images would still find the image report tool useful to make sure they've added alt attributes to all their images (and if they've missed some then they could ask someone for suitable text) |
| 10:01 | <Hixie> | e.g. <img alt=Photo> is useless, but <img important alt=Photo> at least lets the user know there's an image there. |
| 10:01 | <Philip`> | s/you/who/ |
| 10:01 | <hsivonen> | Philip`: are there notable UAs that read file name in the case of V.nu? |
| 10:01 | <jgraham_> | Hixie: So in fact this doesn't really help the AT case (since AFAIK they do that anyway), it just helps the text browser case |
| 10:02 | <Hixie> | jgraham_: I don't think current behaviour in ATs is really indicative of ideal behaviour in ATs, even in the face of existing legacy markup. |
| 10:02 | <hsivonen> | Philip`: yeah, but for that use case you only need to check if a particular heading appears in the report possible counting the rows in the table after it |
| 10:02 | <Hixie> | jgraham_: my limited experience suggests that ATs are horrificly bad at this point, and not in a manner gated by HTML's limitations. |
| 10:03 | <Philip`> | hsivonen: Lynx does |
| 10:03 | <Philip`> | hsivonen: e.g. http://validator.nu/?doc=http%3A%2F%2Fnews.bbc.co.uk&showimagereport=yes says "[_44646536_34e73f52-7689-477a-86d7-7d309cf9a480.jpg] Carlos Tevez, Ryan Giggs and Cristiano Ronaldo celebrate Manchester United's Premier League title victory From line 2817, column 73; to line 2817, column 344" |
| 10:03 | <jgraham_> | Hixie: Sure. But nevertheless they tend to supply this information with or without an additional attribute |
| 10:04 | <hsivonen> | Philip`: eww. is that the default now? |
| 10:04 | <Philip`> | hsivonen: I don't think I've set any non-default options |
| 10:04 | <Hixie> | jgraham_: possible |
| 10:04 | <Hixie> | jgraham_: but they don't have to |
| 10:05 | <Hixie> | jgraham_: especially if we provide this |
| 10:05 | <jgraham_> | Hixie: I'm just not sure what having an additional attribute buys you here |
| 10:05 | <Hixie> | jgraham_: yeah well neither am i really |
| 10:05 | <jgraham_> | Is this substantially different from the @noalt case, but with a semanticised name? |
| 10:05 | <Hixie> | jgraham_: i'm just trying to find a technically good solution to the problem that is being posited to exist with the current text |
| 10:06 | <Hixie> | jgraham_: noalt was defined to have no behaviour at all, and was defined to only be valid if alt= was missing |
| 10:06 | <Hixie> | jgraham_: so yes |
| 10:06 | <jgraham_> | I thought noalt was supposed to indicate a "critical content" image for which no alt text was available? |
| 10:07 | <Hixie> | yes but it was not defined to allow alt="" to be specified with it |
| 10:07 | <Philip`> | hsivonen: Any half-decent content-based heuristics ought to ignore <img src=spacer.gif>, which is bad in this case since it unexpectedly hides some of the image report's images from the user (when they might often want to know about the image so they can copy-and-paste the URI or something) - as the content producer you know more about whether the image is ignorable than the UA can know, and you have to use the alt attribute to communicate that knowled |
| 10:07 | <Hixie> | you couldn't do <img alt=photo noalt> |
| 10:07 | <Philip`> | s/$/ge/ |
| 10:09 | <hsivonen> | Philip`: hmm. alt='image' would be truly ironic, but it would indeed make things better in Lynx |
| 10:09 | <jgraham_> | Hixie: So the @importantimage case would require alt and hence not address some of the cases where that is harmful e.g. a photo dragged into a WYSIWYG editor |
| 10:09 | <hsivonen> | but not in Safari+VO |
| 10:09 | <Hixie> | jgraham_: I don't know if it should require alt, necessarily |
| 10:10 | <Hixie> | jgraham_: but either alt, or important, or both, would need to be specified |
| 10:10 | <Philip`> | hsivonen: It would help in Opera+VO too, since that ignores images with no alt |
| 10:10 | <zcorpan> | Philip`: if the images in the image report were links, would UA still ignore spacer.gif? |
| 10:10 | <Philip`> | hsivonen: (at least while I'm believing that it's bad for the images to be ignored) |
| 10:10 | <Hixie> | jgraham_: though in the case of a wysiwyg editor, nothing that's been proposed so far would help, since you have no way to know if it's decorative or not |
| 10:10 | <Hixie> | jgraham_: and the spec currently doesn't allow you to not know that |
| 10:11 | <zcorpan> | Philip`: <a href=http://www.example.org/spacer.gif><img src=http://www.example.org/spacer.gif></a>; is different from <img src=http://www.example.org/spacer.gif>; |
| 10:11 | <Hixie> | fundamentally the problem with important="" boils down to the fact that you have to define the behaviour for when you have neither alt="" nor important="" |
| 10:11 | <jgraham_> | Hixie: The safe assumption is that it's not decorative |
| 10:12 | <Hixie> | jgraham_: not necessarily -- it would be annoying as all hell to have "image" read out a billion times |
| 10:12 | <Philip`> | zcorpan: The UA shouldn't ignore the whole link, but I don't see why it couldn't ignore the spacer.gif and treat that pattern the same as <a href></a> and convert it to text as "Empty link" |
| 10:12 | <jgraham_> | Hixie: It basically seems like missing alt would map to important |
| 10:12 | <Hixie> | ...and (continuing my earlier thought) you therefore end up making the default the same as including important="", and the attribute becomes mostly useless, unless we consider saying the kind of image to be a significant win |
| 10:13 | <Hixie> | right |
| 10:13 | <jgraham_> | s/missing alt/misgging alt and important/ |
| 10:13 | <jgraham_> | s/gg/ss/ |
| 10:13 | <Philip`> | s/ss/s/ |
| 10:13 | <zcorpan> | Philip`: why would it be useful to do that? |
| 10:13 | <jgraham_> | Philip`: :-p |
| 10:14 | <Philip`> | zcorpan: Which "that"? The ignoring-the-image bit, or the rendering-empty-links-nonemptily bit? |
| 10:14 | <zcorpan> | Philip`: ignoring-the-image |
| 10:14 | <jgraham_> | Hixie: For the case where "image would be read out repeatedly, the user could just stop reading the page or engage some sort of skip images mode or something. |
| 10:14 | <Philip`> | zcorpan: <a href=...><img src=spacer.gif><img src=spacer.gif><img src=spacer.gif></a> |
| 10:15 | <zcorpan> | Philip`: i don't see your point |
| 10:15 | <Philip`> | zcorpan: You wouldn't really want to consider all the images to be significant - the only significant thing is that there's a link there |
| 10:15 | <Hixie> | jgraham_: at which point you've just assumed the images are decorative |
| 10:15 | <zcorpan> | Philip`: are there pages with 3 spacer.gifs in links? |
| 10:15 | <Philip`> | zcorpan: I don't know :-p |
| 10:15 | <Hixie> | zcorpan: almost certainly |
| 10:16 | <jgraham_> | Hixie: Only all remaining images. And it puts the user, rather than the author's editor, in control of their user experience |
| 10:16 | <Hixie> | jgraham_: you're assuming a competent user |
| 10:16 | <Hixie> | jgraham_: but sure |
| 10:16 | <Hixie> | jgraham_: i'm not sure that's the useful default, though |
| 10:17 | <Hixie> | jgraham_: depends which happens more often |
| 10:17 | <Philip`> | Users have a more direct incentive to act competently than authors do :-) |
| 10:17 | <jgraham_> | Basically I think that assuming images for which no information is available are decorative is worse information loss than assuming that they are important and risking the possibility that they are not |
| 10:17 | <Hixie> | jgraham_: you could also imagine a UA/AT saying "this page contains 45 unlabeled images." at the start of the read. |
| 10:17 | <jgraham_> | Hixie: Yeah |
| 10:17 | <Hixie> | jgraham_: (and then not saying where they are) |
| 10:19 | <zcorpan> | i think that noalt/importantimage/etc would be used incorrectly so much that the user experience will be the same or better if such attributes are ignored |
| 10:19 | <zcorpan> | frankly i think that alt is used incorrectly so much that UAs are better off doing image analysis to decide whether the image should be ignored or not |
| 10:20 | <jgraham_> | It would be good if tool authors could be convinced that accessibility is an important problem that needs to be solved by human-led analysis and so encouraged to provide tools for step-through analysis of the document's accessibility |
| 10:20 | <Hixie> | zcorpan: sure, sadly we're not there yet |
| 10:21 | <jgraham_> | If I were an accessibility expert I would spend my time doing that rather than trying to make one accessibility issue get proxied by a syntax requirement that can be caught by a validity check |
| 10:22 | <jgraham_> | (especially given how poor the proxy turns out to be) |
| 10:22 | <Hixie> | the accessibility people doing real work aren't visible in the standards process |
| 10:22 | <Hixie> | i know a lot of them at google |
| 10:23 | <Hixie> | you likely will only rarely hear of them |
| 10:24 | <jgraham_> | Hixie: That's a pity. I mean it's good that they're doing real work but bad that that takes up so much time they can't contribute some sanity to standards |
| 10:25 | <Hixie> | yeah well |
| 10:25 | <Hixie> | the same goes for most engineers, frankly |
| 10:26 | <jgraham_> | Yeah that's pretty noticable |
| 10:26 | jgraham_ | should go |
| 10:26 | <Hixie> | later |
| 10:33 | Philip` | really doesn't like how his laptop's power cable has a loose connection |
| 10:33 | <Philip`> | Every time I move it, I worry about whether that's the last time I'll be able to charge my battery |
| 10:34 | <Hixie> | my dell had that |
| 10:34 | <Philip`> | It'd be nicer if things either worked or didn't work, rather than dangling on the edge |
| 10:34 | <Hixie> | and then one day smoke came out of it |
| 10:34 | <Philip`> | Oh, lovely |
| 10:34 | <Hixie> | they fedexed me a new one overnight |
| 10:36 | <Lachy> | Hixie, re your question about ssh and dev.w3.org, didn't you already have access? How have you been uploading the spec without it? |
| 10:36 | <Hixie> | not for me |
| 10:36 | <Lachy> | oh |
| 10:36 | <Hixie> | i got my access years and years ago |
| 10:37 | <Hixie> | hehay, i didn't know about rfc 4735 |
| 10:37 | <Hixie> | example/* and */example media types |
| 10:43 | <hsivonen> | I wonder where the 'acrimonious tone' came from... |
| 10:47 | <Philip`> | (£30 for a new AC adapter? It's just a dumb lump of plastic and metal :-( ) |
| 10:54 | <Lachy> | my understanding of important="" was that it would indicate that the image is important content and that the user should attempt to find out more about it through additional heuristics or image analysis, if the alt wasn't sufficient. But then I'm not sure how it's any different from <figure><img src="..."><legend>...</legend></figure> |
| 11:03 | <hsivonen> | looks like the meaning of "bolted on" is not agreed upon http://lists.w3.org/Archives/Public/wai-xtech/2008May/0099.html |
| 11:05 | <hsivonen> | I expect there's a Latin term for argument by shame |
| 11:05 | <Lachy> | I don't understand how it could be interpreted differently from being a feature that is not part of the actual content, but rather included along side it to address some issue |
| 11:06 | <hsivonen> | Lachy: if you twist your point of view far enough, an <img> element is just text with an image alternative |
| 11:06 | <Hixie> | uh |
| 11:06 | <Hixie> | wow |
| 11:07 | <Hixie> | that's quite the e-mail |
| 11:07 | <hsivonen> | Hixie: no offense to you intended, but what the spec used to say wasn't really how authors use <img> |
| 11:07 | <Hixie> | hsivonen: indeed |
| 11:07 | <Hixie> | that's why it changed :-) |
| 11:08 | <Hixie> | that e-mail reminds me of religous zealot e-mail |
| 11:08 | <Philip`> | hsivonen: What the spec says about almost all of HTML isn't how authors really use it |
| 11:08 | <Hixie> | "'god does not exist.' you are going to hell! the world is not perfect, but it definitely has a god!" |
| 11:09 | Philip` | sees that HTML 1.0 allowed but did not require ALT |
| 11:09 | <hsivonen> | Philip`: I think the syntax definition needs some more authoring practice alignment with border=0, etc. |
| 11:09 | <Philip`> | (at least judging by its '< A HREF="Go">< IMG SRC ="Button"> Press to start</A>' example) |
| 11:10 | <Philip`> | ...although actually I'm just looking at an old draft |
| 11:10 | <Hixie> | Philip`: html1? |
| 11:10 | <Hixie> | Philip`: i thought "proposal" (the first version of html) didn't have <img> at all |
| 11:10 | <Hixie> | Philip`: and that mark anderseen invented it for mosaic |
| 11:10 | <Hixie> | is that wrong? |
| 11:10 | <Hixie> | or do you mean html+ |
| 11:10 | hsivonen | defers replying, tries to get out of the alt tarpit by working on MathML parsing |
| 11:12 | <Philip`> | http://www.w3.org/MarkUp/draft-ietf-iiir-html-01.txt |
| 11:12 | <Philip`> | Google says that's "html 1.0", so who am I to disagree? |
| 11:12 | <hsivonen> | aside: I find that most of my work-related unhappiness comes from dealing with accessibility-related issues. And I think accessibility is a good thing. |
| 11:13 | <hsivonen> | I mean the alt thread and the aria-foo vs. aria:foo thing |
| 11:13 | <Philip`> | (http://www.ietf.org/rfc/rfc1866.txt seems to have optional ALT too) |
| 11:14 | <Lachy> | alt wasn't required until HTML 3.2, I think |
| 11:14 | <Lachy> | or maybe it was 4.0 |
| 11:14 | <hsivonen> | I'd *so* walk away from the alt stuff if it wasn't about what people want my product to do |
| 11:14 | Philip` | is quite happy working on canvas stuff since it's so clearly inaccessible that nobody even cares any more |
| 11:16 | <Hixie> | wow, i've never seen this draft |
| 11:16 | <Hixie> | this adds a whole version of html to my previously assumed versions |
| 11:18 | <hsivonen> | Hixie: do you mean the IIIR draft? it has been cited on the WHATWG list. |
| 11:19 | <Hixie> | interesting |
| 11:19 | <Hixie> | must've not yet gotten to that e-mail, or assumed that e-mail to be referring to something else |
| 11:19 | <Hixie> | rel=interested is a hilarious (unscalable) idea |
| 11:20 | <hsivonen> | Hixie: I think I cited it in an <i> vs. <em> email |
| 11:20 | <Lachy> | what is rel=interested supposed to mean? |
| 11:20 | <Hixie> | oh i probably assumed it was just a copy of proposal, then |
| 11:20 | <Hixie> | Lachy: see near the end of that draft |
| 11:22 | <Lachy> | I don't see how that's not scalable. How hard can it be to notify any website that links to mine with rel=interested, when my page is updated? ;-) |
| 11:22 | <Lachy> | I suppose it's a like a poorly designed version of pingback |
| 11:23 | <Hixie> | wow, alt=" really is optional in that draft |
| 11:23 | <Hixie> | so optional that the draft doesn't even count it! |
| 11:23 | <Hixie> | "[IMG] has two attributes: SRC [...] ALIGN [...] ALT [...]" |
| 11:24 | <zcorpan> | obviously image alignment is more important than text alternative |
| 11:25 | <Lachy> | yeah, alt wasn't added later in RFC 1866, but still optinal until HTML 4.0. I guess that proves alt really is bolted on. |
| 11:25 | <Hixie> | man this spec is so vague it's ridiculous |
| 11:25 | <Hixie> | anyway |
| 11:25 | <Hixie> | back to making hml5 |
| 11:25 | <Hixie> | html5 |
| 11:27 | zcorpan | looks up expansions for hml |
| 11:27 | <zcorpan> | Human-Machine Language 5? |
| 11:27 | <zcorpan> | Hellfire Missile Launcher? |
| 11:28 | <zcorpan> | or Help Me Lord |
| 11:36 | <virtuelv> | Hixie: you were looking for a security model for File I/O? |
| 11:37 | Philip` | wonders why it's bad to use <progress> for a static gauge |
| 11:39 | <Lachy> | Philip`, because using progress sugests that the value will change in the future, and because common progress widgets UI looks different from a gauge |
| 11:40 | <Hixie> | virtuelv: ? |
| 11:40 | <virtuelv> | Lachy: yes, but if I say have a project tracker, and it says that some task is 80% complete |
| 11:40 | <Hixie> | virtuelv: in what context? |
| 11:40 | <virtuelv> | Hixie: the proposal sent to the web api mailing list? |
| 11:40 | <Hixie> | virtuelv: ah, yes, what about it? |
| 11:41 | <virtuelv> | I have a workable idea for it, I think |
| 11:41 | <Lachy> | virtuelv, then it's not really static. It may just take a while before it progresses beyond 80% |
| 11:41 | <Hixie> | let the wg know :-) |
| 11:41 | <virtuelv> | Lachy: the point is that it's not going to progress during the lifetime of the page |
| 11:42 | <Hixie> | Lachy: a progress bar might make sense for something that is a task competion amount (maybe.) but it's not appropriate for a disk space utilisation meter, e.g. |
| 11:42 | <Lachy> | Hixie, yeah, I know |
| 11:42 | <Hixie> | er, that was meant for Philip` |
| 11:43 | <Lachy> | virtuelv, that doesn't really matter. It's still a progress indicator |
| 11:43 | <hsivonen> | as a practical matter, it matters that progress bar is animated to look busy on OS X and Ubuntu |
| 11:43 | <virtuelv> | Lachy: agreed, but it was somewhat unclear from Philip`s question |
| 11:44 | <hsivonen> | so if the project isn't completing while the users is waiting, you may want a gauge anyway |
| 11:46 | <Lachy> | are there non-animated progress indicators available in OSX or Ubuntu? |
| 11:46 | <hsivonen> | I don't know |
| 11:50 | <Lachy> | maybe we should have some way to indicate whether the progress bar is relatively static or dynamic, where a dynamic one is expected to update relatively quickly, and a static may stay the same for a while. |
| 11:51 | <Lachy> | and if there are non-animated progress bars available, they would be used for the static progress bars. |
| 11:53 | <hsivonen> | isn't a non-dynamic progress bar close enough to gauge for all practical purposes? |
| 11:53 | <Lachy> | I don't know. Maybe. |
| 11:56 | <Lachy> | Apple's HIG shows both active and inactive fill progress bars http://developer.apple.com/documentation/UserExperience/Conceptual/OSXHIGuidelines/XHIGControls/chapter_19_section_5.html#//apple_ref/doc/uid/TP30000359-TPXREF208 |
| 11:57 | <Lachy> | but it doesn't explain when each one should be used. |
| 11:58 | <Hixie> | inactive = the window isn't focused |
| 11:58 | <Lachy> | oh, ok |
| 11:59 | <Hixie> | <meter> = capacity indicator on that page |
| 11:59 | Philip` | doesn't like forms that try to automatically capitalise addresses and result in "King'S College" |
| 12:00 | <Lachy> | so I guess my static/dynamic idea is no good. Ignore it. |
| 12:00 | <Hixie> | (we don't have relevance indicators or rating indicators or discrete capacity indicators or asynchronous progress indicators, as defined by that page) |
| 12:00 | <Hixie> | (though maybe we should) |
| 12:01 | <hsivonen> | I thought <meter> was meant to be a relevance indicator, too |
| 12:04 | <Lachy> | for the rating indicator in that HIG page, would that be best done <input type=range>, and somehow styled with XBL? |
| 12:04 | <Hixie> | hsivonen: yeah, in html5 <meter> acts as both |
| 12:04 | <Hixie> | Lachy: no idea |
| 12:04 | <Hixie> | 4am. bed time. |
| 12:11 | <Hixie> | before i go to bed |
| 12:11 | <Hixie> | does anyone want to give a talk on html5 at @media 2008 in london at the end of may? |
| 12:11 | <Hixie> | Philip`? jgraham_? |
| 12:15 | <Hixie> | don't all reply at once :-) |
| 12:16 | Hixie | goes to bed |
| 12:26 | <BenMillard> | if JGraham goes and my travel is paid, we could show how HTML5 table accessibility has improved through research, prototypes and analysis |
| 12:26 | <Lachy> | hmm. Only 16 days to go till @media. I don't think I could put together a presentation and get approval and funding to go in that short time. |
| 13:33 | <hsivonen> | any suggestions on how to best handle *not* acknowledging the self-closing flag? |
| 13:34 | <hsivonen> | hmm. perhaps I should use break instead of early return and fall to a late return that inspects whether the flag has been cleared |
| 13:35 | <hsivonen> | yay for goto programming |
| 13:39 | <zcorpan> | goto is the shit. opera supports it in javascript too |
| 13:46 | <hsivonen> | if I reset the insertion mode in MathML, what do I set the secondary mode to? |
| 13:48 | takkaria | adds to the spam of public-html |
| 13:50 | <hsivonen> | takkaria: http://lists.w3.org/Archives/Public/wai-xtech/2008May/0098.html |
| 13:52 | <Lachy> | takkaria, this is what at least one blind photographer does http://my.opera.com/oedipus/albums/ |
| 13:53 | <Lachy> | although, the alt attributes used repeat the tags, that's a problem with my.opera, which is out of his control |
| 14:05 | <takkaria> | hsivonen: that goes part of the way but doesn't quite address it |
| 14:08 | <hsivonen> | takkaria: it seems, though, that the WCAG 2.0 framers have come to the conclusion that there are use cases that should be allowed to exist even if they aren't fully accessible |
| 14:12 | <takkaria> | mm |
| 14:17 | <Philip`> | Hixie: @media seems too trendy for me :-p |
| 14:17 | <Philip`> | and I don't think I'd be any good at that kind of thing |
| 14:17 | <hsivonen> | Lachy: switching services is not out of the users control and advocates seem to be suggesting getting rid of bad tools, i.e. switching tools |
| 14:18 | <takkaria> | good lord, misunderstandings are easy to come by these days |
| 14:18 | <Philip`> | You could ask the toolmaker to improve the tools - switching away is not the only possible form of feedback |
| 14:19 | <hsivonen> | Philip`: sure. my point is that it comes down to choosing what you value more if the tool maker doesn't fix |
| 14:59 | <Lachy> | wow, Rob seemed to completely sidestep the issue instead of answering the question http://lists.w3.org/Archives/Public/public-html/2008May/0263.html |
| 15:17 | <takkaria> | there has to be some middle ground here somewhere such that the a11y people and the critical-content-altless people can stop talking past each other |
| 15:19 | <Philip`> | Perhaps all the middle ground has been turned into no man's land |
| 15:24 | <hsivonen> | the middle ground seems to be giving up on the distinction between alt not known and purely decorative and using alt='' for both cases |
| 15:26 | <zcorpan> | hsivonen: i think that could be reasonable, given that there is a non-trivial amount of content images on the web that have alt='', and so users would miss out if UA ignored them |
| 15:27 | <zcorpan> | (i don't have data but i've seen content images with alt='' more than content images with good alt, i think) |
| 15:27 | <hsivonen> | zcorpan: of course, the next question is: if no alt and alt='' mean the same thing, why bother with alt='' except for backwards compat and in order to have more error cases |
| 15:28 | <zcorpan> | hsivonen: indeed, i'd be fine with making no alt and alt='' equivalent and have alt optional :) |
| 15:30 | Philip` | thinks data would help, to determine roughly what "non-trivial amount" and "more" mean |
| 15:30 | <zcorpan> | Philip`: indeed |
| 15:31 | <zcorpan> | i might be completely off and hence my conclusions wrong |
| 15:38 | <takkaria> | hmm, webcams that take a snapshot every x minutes and upload it to the web are like blind photographers but remove humans from the equation entirely |
| 15:38 | <takkaria> | I think that might turn into my new example of choice |
| 15:39 | <hsivonen> | takkaria: whoever sets up the Webcam knows what general direction it is pointed at |
| 15:39 | <Philip`> | What if the webcam is mounted on a robot that is controlled remotely by visitors to the web site? |
| 15:40 | <takkaria> | hsivonen: not on a laptop |
| 15:41 | <takkaria> | I should start collecting these cases somewhere |
| 15:42 | <hsivonen> | the MathML stuff has an admirably small impact on tree construction code |
| 15:42 | <hsivonen> | most of the changes were related to adding a namespace parameter all over |
| 15:42 | <hsivonen> | which thankfully is the kind of stuff Eclipse is really good at |
| 15:45 | <mpt> | takkaria, http://www.geonet.org.nz/volcano/volcams/whiteisland/whiteisland-recent.html |
| 15:45 | <Lachy> | what we really need for the alt discussion to get somewhere is a study that looks at the reasons why people use alt="" or alt="bogus, validator friendly content" or whatever. The difficulty is in figuring out how to perform such a study |
| 15:46 | <Lachy> | maybe a carefully crafted survey, asking web developers a bunch of questions about the issue would work |
| 15:47 | <hsivonen> | Lachy: no, this is the kind of stuff where you have go look for revealed preferences from actual actions, because if you ask, people lie |
| 15:47 | <takkaria> | mpt: thanks |
| 15:47 | <Lachy> | yeah, but the problem is, how do you unambiguosly determine someone's reasons from simply looking at their output? |
| 15:48 | <Philip`> | Lachy: You can't, but you can't determine that by any other mechanism either |
| 15:48 | <mpt> | interview them afterwards |
| 15:48 | <mpt> | Welcome to sociology! |
| 15:49 | takkaria | attempts to pin down some concrete proposals rather than wordy long-paragaph replies |
| 15:49 | <Philip`> | Also, your assumption that people have reasons is not clearly valid |
| 15:52 | <Lachy> | Philip`, people always have reasons for doing something, even if that reason is that they didn't care and just picked one alternative randomly |
| 15:58 | <Lachy> | if I were to do a presentation on HTML5 at @media, which topics do you think people would like to hear about? |
| 15:58 | <Lachy> | it looks like the presentations are an hour long, so I could fit a lot in |
| 16:02 | <takkaria> | heh, moving the goalposts is popular sport on public-html |
| 16:03 | hsivonen | expects the famous 'axiomatic proof' to show up |
| 16:06 | <takkaria> | I wish I had the time to write a proper reply to Rob today, but it'l have to wait til tomorrow |
| 16:06 | <Lachy> | hsivonen, http://ln.hixie.ch/?start=1041735552&count=1 ? |
| 16:07 | <hsivonen> | Lachy: yeah |
| 16:08 | <Lachy> | fortunately, Shelbey hasn't joined the group yet. But who knows, she might. |
| 16:09 | <virtuelv> | hm, interesting image chosen for Hixie's last log entry |
| 16:11 | <Lachy> | virtuelv, the image is somewhat random (though it doesn't update too frequently). |
| 16:15 | <hsivonen> | can foreign elements ever participate in foster-parenting? |
| 16:23 | <zcorpan> | hsivonen: you mean in e.g. <table><math> ? |
| 16:30 | <zcorpan> | "you're not reading my questions." "you're not reading my answers." |
| 16:30 | zcorpan | wonders why he's still reading |
| 16:33 | <hsivonen> | zcorpan: yes |
| 16:34 | <hsivonen> | zcorpan: umm no |
| 16:34 | <hsivonen> | I mean stuff that occurs 'in foreign content' |
| 16:34 | <zcorpan> | ah |
| 16:40 | <virtuelv> | hsivonen: are your XTech slides anywhere online? |
| 16:42 | <hsivonen> | virtuelv: http://hsivonen.iki.fi/xtech2008/ but the server is not responding right now |
| 16:42 | <hsivonen> | actually it is responding but very slowly |
| 16:43 | <virtuelv> | hsivonen: that's ok, I just need something to paste into my trip report |
| 16:49 | aroben | wonders why it's <dialog> instead of <dialogue> |
| 16:52 | <hsivonen> | yay. pointing out a use case is undermining the work of the WG |
| 16:53 | <gavin> | wow |
| 16:54 | <hsivonen> | Hixie: I'm now even more convinced that 'in foreign content' should be a flag and not an item in the mode enumeration |
| 16:54 | <hsivonen> | for efficient impl |
| 16:55 | <hsivonen> | do we want to have the self-closing flag on end tags for the purpose of tokenizer tests? |
| 17:23 | <gsnedders> | Hixie: it isn't just atom:content that allows it. It's that, and Atom XHTML Text Constructs. |
| 17:24 | <hober> | i.e. atom:title, atom:summary, atom:content, atom;rights |
| 17:24 | gsnedders | was being lazy and not looking |
| 17:24 | <gsnedders> | But it's probably better to generically define it |
| 17:25 | <gsnedders> | atom:subtitle too |
| 17:25 | <hober> | right, by referencing the text construct dfn |
| 17:25 | <virtuelv> | hsivonen: you did a lightning talk too? |
| 17:26 | <hsivonen> | virtuelv: yeah |
| 17:27 | <virtuelv> | url? |
| 17:27 | <hsivonen> | virtuelv: the same directory |
| 17:28 | <hsivonen> | http://hsivonen.iki.fi/xtech2008/html5-parsing-lightning.pdf |
| 17:28 | <virtuelv> | ty |
| 17:28 | <virtuelv> | I never looked at the directory, since I assumed it was a direct link |
| 19:18 | <othermaciej> | ok the accessibility crowd needs to get their story straight |
| 19:19 | <othermaciej> | how can they cite each other's emails as supporting evidence when they are saying directly contradictory things? |
| 19:19 | <othermaciej> | am I losing my mind? |
| 19:24 | <Lachy> | othermaciej, which mails are you referring to? |
| 19:25 | <othermaciej> | Lachy: Robert Burns saying that alt is not supposed to be used for photos, and others citing his arguments |
| 19:25 | <othermaciej> | am I totally lacking in reading comprehension skills because that seems like the opposite of what AI54 says |
| 19:26 | <rammic> | Can anyone answer a stupid spec question for me? (4.7.5.1. Changes to the networking model) would seem to imply that if a cached application cannot be reached due to network error (downtime?), then the browser should load the cache version- right? |
| 19:28 | <Lachy> | yeah, I thought Rob's argument was a bit strange |
| 19:29 | <Lachy> | rammic, I'll check the spec and see how I interpret it. |
| 19:29 | <rammic> | Thanks. |
| 19:31 | <andersca> | Hixie? |
| 19:32 | <Lachy> | rammic, I believe you're right. Though, I could be wrong, since I'm not very familar with that section of the spec. |
| 19:34 | <rammic> | Okay. I'm playing around with the Firefox implementation and found it wasn't working as expected... I hope failback is intended (otherwise kills several use cases). |
| 19:37 | Philip` | loves C, particularly when his program crashes with a non-existent backtrace because a function is somehow making %esp offset by 4 when it's called |
| 19:41 | <othermaciej> | ok it is starting to bug me a little that Steven Faulkner doesn't seem to even be trying to spell my name correctly |
| 19:45 | <Lachy> | najiec, I think it is you who is misspelling your name :-) |
| 19:59 | <Philip`> | (Yay for undeclared functions, and C guessing that it probably just takes int arguments and returns an int or something and who cares if that's going to crash your program) |
| 20:05 | <othermaciej> | -Werror |
| 20:05 | <Philip`> | Legacy codebase with hundreds of existing warnings :-p |
| 20:07 | <Philip`> | and I want my fork to differ as little as possible from the original version, to avoid future pain when merging changes, so I don't want to fix all the warnings in my copy |
| 20:09 | <gsnedders> | Philip`: I ended up trying that once. I rewrote all but one method in the end. :P |
| 20:19 | <annevk> | it's amusing that when html4all people are confused with HTML5, HTML5 is blamed |
| 20:19 | <annevk> | but when their propposal is confusing, the reviewer is blamed |
| 21:17 | <hsivonen> | argh. now armchair lawyering about copyright metadata is brought in |
| 21:21 | <Lachy> | I think alt text used in these Examples that Gez linked to would be better marked up as captions. http://emacspeak.sourceforge.net/raman/ |
| 21:22 | <Lachy> | although, I guess they are links, so maybe not |
| 22:12 | <Hixie> | andersca: here now |
| 22:12 | <Hixie> | andersca: note btw there were some changes made yesterday that you will want to see |
| 22:13 | <andersca> | Hixie: oh? |
| 22:13 | <jgraham_> | othermaciej_: Steve consistently calls me "Graham" for some reason I don't quite understand |
| 22:14 | <Hixie> | andersca: yeah, look at the html5 tracker and at the recent checks with the webkit icon |
| 22:14 | <andersca> | Hixie: ice! |
| 22:14 | <andersca> | nice even |
| 22:15 | <Hixie> | andersca: as far as i know that resolves all the outstanding issues with that part of the spec |
| 22:19 | <gsnedders> | jgraham_: Would you rather myself and Philip` look at you at once? |
| 22:20 | <jgraham_> | gsnedders: That's unnerving. This is just faintly insulting (which is a pity because I think Steve is actually a nice guy) |
| 22:27 | <gsnedders> | Lachy: re your tweet: the solution to issue 42. :P (or, seriously, why defining error handling actually helps anyone when all the browsers are close enough anyway) |
| 22:28 | <Lachy> | awsome! Someone actually reads my twitter feed :-) |
| 22:29 | <gsnedders> | Lachy: Feel honoured. I follow very few :) |
| 22:36 | Philip` | wonders how much of http://googleblog.blogspot.com/2008/05/moving-to-unicode-51.html is the effect of Google's indexing system becoming less US-centric over time |
| 22:36 | <Hixie> | MikeSmith: help me out here. what would i need to do to get http://www.w3.org/2000/11/DOM-Level-2-errata updated? |
| 22:45 | <soypunk> | blah, I wish http://twitter.com/replies were visible for other users so that besides seeing my own I could also do http://twitter.com/lachy/replies |
| 22:45 | <soypunk> | i guess this is as good as it gets: http://quotably.com/Lachy/statuses/810546483 |
| 22:51 | <mcarter> | in my pursuit of implementing cross-sub-domain SSE and TCPConnection for IE, I've noticed that sometimes switching document.domain causes some weird illegal access errors, but then calling CollectGarbage(); right after switching the domain sometimes fixes those errors. I don't suppose anyone might possibly know why that is? |
| 22:52 | <Hixie> | o_O |
| 22:54 | <Lachy> | soypunk, why do you need to be able to see replies to me? |
| 22:54 | <soypunk> | well replies to your inquiry... |
| 22:55 | <Lachy> | FWIW, you're the only one to respond via twitter so far |
| 22:55 | <Hixie> | i think for explaining <meter> better we should have some pictures of gagues |
| 22:55 | <Hixie> | gauges |
| 22:55 | <Hixie> | anyone wanna take screenshots of gauge controls? :-) |
| 22:55 | <soypunk> | Lachy: are you collecting this info for the authoring guide, a follow-up ALA piece, or other? |
| 22:55 | <Lachy> | other |
| 22:56 | <Lachy> | for 2 things actually |
| 22:56 | <smedero> | should really keep my html-wg/whatwg nicks the same. |
| 22:56 | <Lachy> | 1. An interview I'm doing on thursday evening for a podcast, and a presentation that jgraham and I are doing at @media |
| 23:19 | <roc> | I guess I'd better alert dcamp to those offline spec changes |
| 23:20 | <Hixie> | ooo, i didn't realise mozilla had an implementation of the offline stuff that was in sync with the spec |
| 23:20 | <Hixie> | my bad |
| 23:20 | <roc> | you didn't? |
| 23:20 | <roc> | Dave rewrote it all to be in line with the spec |
| 23:20 | <roc> | well, hopefully |
| 23:20 | <Hixie> | i thought the mozilla stuff was an old implementation that was removed for ff3 |
| 23:20 | <roc> | the old implementation was removed ... and a new one inserted :-) |
| 23:21 | <Hixie> | aah :-) |
| 23:21 | Hixie | updates the markers in the spec |
| 23:22 | <roc> | http://starkravingfinkle.org/blog/2008/05/firefox-3-offline-app-demo-part-2/ |
| 23:22 | <roc> | > The biggest changes involve dropping support for our own <link rel="offline"> mechanism and supporting WHATWG manifests and application cache. |
| 23:22 | <Hixie> | nice |
| 23:23 | <othermaciej_> | is there a web interface that I can use to look at checkins of HTML5? |
| 23:23 | <roc> | I think you did know this at one point, since you wrote some tests and they failed on Firefox :-). but I think Dave fixed them |
| 23:23 | <Hixie> | othermaciej: yes, see the link at the top of the spec |
| 23:23 | <Hixie> | roc: ooooo, that rings a bell, yes |
| 23:24 | <Hixie> | man i need to keep better track of who does what |
| 23:24 | <othermaciej> | Hixie: thanks! |
| 23:25 | gsnedders | improves Hixie's bell |
| 23:32 | <vlad_> | annevk: ping |
| 23:32 | <Hixie> | ('course it's 00:37 in his time zone, so he might not be around) |
| 23:33 | <vlad_> | damn you |
| 23:40 | <Philip`> | vlad_: http://html5.googlecode.com/svn/trunk/web-apps-tracker/web-apps-tracker - patches welcome ;-) |
| 23:45 | <hendry> | for vim users http://svn.natalian.org/projects/html5/html.vim |
