/irc-logs / freenode / #whatwg / 2008-08-25 / end

Options:

# Session Start: Mon Aug 25 00:00:00 2008
# Session Ident: #whatwg
# [00:00] * Quits: hasather (n=hasather@90-231-107-133-no62.tbcn.telia.com) (Read error: 110 (Connection timed out))
# [00:06] <hsivonen> what's the deal with the new delicious.com wanting me to sign in again and again and putting autocomplete=off in the login fields?
# [00:06] <Hixie> zcorpan: yeah, it's a should on purpose. I'm sure somebody will find some case where it would be appropriate. :-)
# [00:06] <hsivonen> do they think they are a bank now?
# [00:10] * Quits: Amorphous (i=jan@f048035151.adsl.alicedsl.de) (Read error: 110 (Connection timed out))
# [00:12] * Joins: Amorphous (i=jan@g227185134.adsl.alicedsl.de)
# [00:18] <hsivonen> This Week in HTML5 makes it on Robin Cover's radar: http://xml.coverpages.org/newsletter/news2008-08-22.html
# [00:19] <Hixie> sweet
# [00:19] <Hixie> go mark
# [00:20] <hsivonen> Google found it for me through a splog instead of the original source, though, so slogs are useful for something
# [00:21] <Hixie> hah
# [00:21] <Hixie> it would have found the real thing eventually
# [00:21] <Hixie> i think we need to do something to raise awareness of the way we're doing html5
# [00:21] <Hixie> - the openness
# [00:21] * Philip` thinks the XML Daily Newslink would be somewhat easier to read if it used paragraphs
# [00:21] <Hixie> - the history
# [00:22] <Hixie> - the non-consensus approach
# [00:22] <Hixie> i'm not sure how to get this awareness raised exactly
# [00:23] <Hixie> but several people have come to me recently with quite wrong preconceptions and then changed their attitude quite dramatically when i explained to them the history and so on
# [00:30] <webben> hsivonen: Have you tried following "Why does Delicious keep asking me to log in?" at http://delicious.com/help/faq ?
# [00:31] <webben> hsivonen: the inclusion of autocomplete is, hmm, curious to say the least
# [00:32] <webben> we don't have that on our main sign in boxes.
# [00:32] <jgraham> Hixie: What preconceptions did they have?
# [00:34] <Hixie> that it was a concensus-driven, closed working group that ignored the needs of users
# [00:38] <jgraham> Did they mainly associate HTML5 with WHATWG or W3C?
# [00:38] <Hixie> i don't think they really understood the difference
# [00:38] <Hixie> which makes sense
# [00:38] <Hixie> i'm not really worried about that per se
# [00:39] <jgraham> Interesting. I was wondering if they thought we were just another W3C working group or if they had independtly reached the wrong conclusions about WHATWG
# [00:40] <Hixie> good question
# [00:40] <Hixie> i don't know
# [00:42] <jgraham> (it should be noted that some people in the HTMLWG think we are a consensus driven group apt to ignore the needs of users...)
# [00:43] <gsnedders> My tongue hurts. I oughtn't bite it.
# [00:43] <jgraham> gsnedders: Literally?
# [00:43] <gsnedders> Literally.
# [00:44] <webben> yeah, eating yourself - not the best of ideas ;)
# [00:44] <Hixie> i bite my tongue literally all the time, hurts a bitch
# [00:45] <Hixie> and it inflates when you bite it, which just makes biting it more likely
# [00:45] <Hixie> it's terrible
# [00:45] <Hixie> jgraham: yeah, i'm not too worried about them. :-)
# [00:45] <jgraham> Hixie: They know :)
# [00:46] <gsnedders> Hixie: Exactly. I normally do it to the inside of my cheek, though, and not my tongue.
# [00:47] <jgraham> gsnedders: I take it this was accidentially biting it when you were aiming for food or something
# [00:47] <gsnedders> jgraham: Oh, probably just from thinking too hard
# [00:49] <Philip`> You can get various bitter-tasting liquids that you can put on your fingernails to stop yourself biting them; maybe you could put the same stuff on your tongue to stop you biting that
# [00:50] <Hixie> -_-
# [00:50] <gsnedders> Philip`: Tongue has tastebuds though, fingernails don't
# [00:51] <Philip`> Oh, darn
# [00:51] * gsnedders notes he can't wear any sort of toxic nail varnish, because he bites his nails too much (or at least, sort of sucks on the end of his fingers)
# [00:52] <gsnedders> (Yes, I did just make a reference to wearing nail varnish)
# [00:53] <gsnedders> And even if I put a bitter-tasting one on I still do it
# [00:53] * gsnedders is hopeless
# [00:54] <jgraham> gsnedders: You can wear toxic nail varnish in just the same way that everyone else can. It's just you're more likely to die the less-than-magnanimous death by nail-varnish poisoning
# [00:54] <gsnedders> jgraham: :)
# [00:55] <Philip`> Better than death by tea cosy
# [00:55] <Philip`> (Wikipedia reckons there's a 1 in 20 billion chance of that)
# [00:56] <gsnedders> Philip`: Link?
# [00:56] <gsnedders> http://en.wikipedia.org/wiki/List_of_causes_of_death_by_rate ?
# [00:56] <gsnedders> No, that doesn't contain tea cosy
# [00:56] <Philip`> http://en.wikipedia.org/wiki/Tea_cosy
# [00:57] <gsnedders> One in twenty billion isn't bad.
# [00:57] <gsnedders> I think the probability of dying is higher than that.
# [00:58] <csarven> HMm.. Suicide is more likely then drowning
# [00:58] * gsnedders almost committed suicide by drowning
# [00:58] <csarven> *common
# [00:58] <gsnedders> Which means I could've counted towards both of those stats!
# [01:00] <BenMillard> jgraham, the "Deliverable for Action 72 @headers" got through my junk filters
# [01:01] <jgraham> BenMillard: What do I win?
# [01:01] <BenMillard> jgraham, me as an audience for your messages :)
# [01:02] <jgraham> gsnedders: I guess given some sort of afterlife it would have given you the oppertunity to have a party anecdote about things you had in common with Virgina Woolfe</macabre>
# [01:02] <jgraham> BenMillard: A fine prize indeed, I feel
# [01:03] <BenMillard> wewt
# [01:03] <BenMillard> your table inspector's highlighting seems to work more reliably than I remember from last year
# [01:04] <jgraham> BenMillard: Did you start using Firefox 3 in the interim? That might have made a difference, or I might have changed the code
# [01:04] <BenMillard> jgraham, still Firefox 2. changes to the code sounds likely
# [01:05] <BenMillard> jgraham, you wrote "I think the absolute simplest message that we can give authors is 'mark up your table headers as <th>'."
# [01:05] <jgraham> gsnedders: Although I should say that the rest of us are much happier with a gsendders without that particular anecdote :)
# [01:05] <BenMillard> jgraham I agree with that quote...think I said that at the November 2007 meeting
# [01:06] <BenMillard> there's a common misconception I come across, where accessibility is assumed to require complexity
# [01:07] <jgraham> BenMillard: You might well have done. It bothers me that a lot of accessibility types seem to have the idea that as long as there is some way to do something then that's good enough, even if it is really hard to use
# [01:07] <jgraham> (the first sentence referred to "said that in Nov. 2007)
# [01:08] <BenMillard> yeah, I gathered :)
# [01:08] <BenMillard> often the things they want to do aren't at all useful
# [01:08] <BenMillard> like creating associations between things that are right next to each other
# [01:09] * Joins: shepazu (n=schepers@80.187.212.118)
# [01:09] <Hixie> careful, you might lose your accessibility expert credentials if you keep saying sensible things like that :-)
# [01:09] <BenMillard> hixie, lol :P
# [01:11] <jgraham> BenMillard: That's why I keep suggesting that user testing is a stronger argument than theory.
# [01:11] * Joins: tantek (n=tantek@adsl-63-195-114-133.dsl.snfc21.pacbell.net)
# [01:11] <BenMillard> jgraham, indeed
# [01:12] <BenMillard> I wonder if I have lost more than I've gained by ignoring Public-HTML for the past several months
# [01:12] <BenMillard> I've certainly gained a lot of time, with which I've done many fun and some productive things
# [01:14] <jgraham> BenMillard: It was pretty quiet for a bit. Then about a week or two ago it all kicked off again
# [01:15] <jgraham> Same topics, similar arguments
# [01:16] <BenMillard> since it all gets publically archived, I wonder if there's any gain in re-responding
# [01:16] * Joins: gsnedders_ (n=gsnedder@host86-139-217-149.range86-139.btcentralplus.com)
# [01:16] <Philip`> It gets archived but there's so much that nobody's ever going to actually read the archives
# [01:17] <Philip`> so clearly the solution is to keep reposting the better arguments, so that nobody will miss them
# [01:18] <jgraham> BenMillard: I guess the problem is xkcd syndrome http://xkcd.com/386/
# [01:19] <BenMillard> Philip`, alternatively, the people who decide HTML5's development probably saw and took on board the better points first time around, so repeating them doesn't change HTML5's direction very much?
# [01:20] <BenMillard> jgraham, perfect!
# [01:22] <Hixie> there are three main topics being discussed right now
# [01:22] <Hixie> img alt: i think i missed a couple of key things the first time (how the {} thing affects tool developers)
# [01:23] <Hixie> well, i guess the second time, not the first time
# [01:23] * Quits: sverrej (n=sverrej@89.10.27.245) (Read error: 110 (Connection timed out))
# [01:23] <Hixie> so i don't know what i'll do with that, probably go back to just making alt="" be not given when you have no alt text
# [01:23] <Hixie> table headers: i haven't yet seen a clear description of the problem, but if there is a problem, then we'll have to address it
# [01:24] <Hixie> rdfa: i haven't managed to get ben to actually describe the problem yet
# [01:25] <BenMillard> Hixie, could you disambiguate when you talk about "Ben"? a couple of times I've seen you say that and it's put me into a state of panic since I don't remember that happenning :)
# [01:26] <webben> if "no alt text" means no precise equivalent, I think that would probably force me into non-conformance since I don't think that yields the best results for end-users.
# [01:27] <Hixie> oh sorry
# [01:27] <Hixie> i mean ben adida in this case
# [01:27] * Quits: gsnedders (n=gsnedder@host217-44-35-200.range217-44.btcentralplus.com) (Read error: 110 (Connection timed out))
# [01:28] * Quits: eseidel (n=eseidel@c-24-130-13-197.hsd1.ca.comcast.net)
# [01:29] <Hixie> webben: so what text would you suggest for the following cases? a generated image for a fractal explorer; the images in google maps in map view and satellite view; the street view images in google maps; an image uploaded as part of a batch upload with no information; a webcam
# [01:29] <webben> Hixie: I'm happier with a text description of those than no alt text.
# [01:29] <Hixie> in all of these cases, the software has no idea what the image is so can't describe it at all
# [01:29] <webben> Yes, a text equivalent is better, but I think a text description is better than zilch.
# [01:30] <webben> Hixie: alt="fractal" alt="uploaded photo"
# [01:30] <jgraham> Hixie: FWIW my guess is that there is a strong enough use case for people accessing the images out of the main flow that something like "fractal", "map tile", "street view", "photo", "webcam" would be better than nothing
# [01:31] <Hixie> webben: yeah see i don't agree that that is better, i think it's actually worse
# [01:31] <jgraham> but IANAATU
# [01:31] <Hixie> ATU?
# [01:31] <Hixie> AT user?
# [01:31] <jgraham> yes :)
# [01:31] <Hixie> woot
# [01:31] <Hixie> well that's what the {phoot} idea was for
# [01:31] <Hixie> {photo} even
# [01:32] <webben> well, i'd prefer being able to demark an alt as non-equivalent with an attribute, but i'd still include it.
# [01:32] <jgraham> My feeling is that in order of worst to best we have nothing, alt={photo}, no-text-equivalent alt="photo"
# [01:33] <jgraham> But I agree that all three suck for different reasons
# [01:33] <Hixie> i just don't see an attribute working for this
# [01:33] <Hixie> it has so many issues
# [01:34] <Hixie> i think the net misuse of a new attribute would make the total level fo accessibility lower
# [01:34] <BenMillard> I've seen gallery software use alt="Photo" as a default
# [01:34] <BenMillard> forums have "User posted image" or similar
# [01:34] <jgraham> I guess the fourth possibility is no attribute and no special syntax, just require the end user to work out if it is supposed to be a description or a text equivalent
# [01:35] <webben> Hixie: does the attribute have any use-cases other than for hypothetical future software?
# [01:35] <BenMillard> jgraham, yes, which is hardly much of a challenge in the cases I've seen
# [01:35] <webben> Hixie: e.g. do you expect a near-future version of Lynx to treat non-equivalents differently?
# [01:36] <webben> or was the idea that {} would force Lynx into treating them differently?
# [01:36] <jgraham> webben: I think in theory AT could only voice "image" if there is @n-t-e or {} and just read the alt text with no delimiter otherwise
# [01:37] * webben can't see why an AT would want to do that.
# [01:37] <jgraham> But in practice misuse of {} or an attribute might make that not work
# [01:37] <jgraham> webben: Want to do what?
# [01:37] <webben> 'voice "image" if there is @n-t-e or {} '
# [01:38] <webben> oh wait i see what you mean
# [01:38] <Philip`> jgraham: Misuse of alt="photo" (regardless of what the spec says) means that in practice AT can't just read the alt text with no delimiter
# [01:40] <jgraham> Philip`: I would expect that to be the case as well. So unless there is something I haven't thought of (likely) it seems just as well to say "sometimes @alt is a replacement, sometimes it is a description, work it out for yourselves"
# [01:40] <jgraham> (with rules like the current ones about when it should be a replacement and when a description)
# [01:41] <Hixie> the idea is that on an ideally written page, an <img> with alt="" can be completely substituted for its alt="" text, with no indication that an image was there
# [01:41] <jgraham> I guess it makes Henri's image checker less useful
# [01:41] <Hixie> but that these <img>s we're talking about would need to indicate that there is an image present
# [01:41] <Hixie> since that's the whole point
# [01:42] <Hixie> so we theoretically need a distinction
# [01:42] <Hixie> lack of alt="", {}, n-t-e, something
# [01:42] <jgraham> Hixie: The problem with designing for an ideally written page is that you're pretty much the only person who writes them :)
# [01:43] * webben isn't persuaded that would be an ideal page.
# [01:43] <jgraham> (although when I read feeds in thunderbird with images disabled, it is immensely irritating when people describe a picture rather than replace it either with "" or an actual equivalent)
# [01:44] <jgraham> webben: For whom? I think part of the issue might be that the text browser case is a bit different from the screenreader case
# [01:44] <webben> People like to communicate with images. Therefore being able to retrieve and share and bookmark etc images is an important use-case.
# [01:44] <webben> i want to be able to do that with a text browser, I think I'd want to be able to do that if I were a screen reader user too.
# [01:45] <jgraham> Hmm.
# [01:45] <Hixie> you don't want to bookmark most images
# [01:45] <Hixie> you probably don't even notice most images
# [01:45] * jgraham is going to bed now even though he hasn't managed to put a stop to wrongness on the internet :)
# [01:45] <Hixie> e.g. "RSS" is often an image
# [01:45] * webben never said most images.
# [01:45] <Hixie> sure
# [01:46] <Hixie> i'm just saying that it's an important use case but only for a small subset of images
# [01:46] <webben> although being able to grab that icon _is_ useful to people authoring content
# [01:47] <BenMillard> jgraham, it's good to break that habit
# [01:47] <BenMillard> as in, put your personal wellbeing ahead of timely responses to internet arguments
# [02:01] <webben> Hixie: being able to differentiate that subset (@role?) might be useful. but i think that would need to work as an authorial opt-in, since most people probably wouldn't bother and there's a huge body of existing content not making such a differentiation.
# [02:14] <Hixie> the problem with any attribute here is it'd be massively misused
# [02:15] <Hixie> much more so than a special syntax, based on what i've seen of web authoring practices so far
# [02:15] <Dashiva> It seems they're all tripping over each other in making @scope seem like a plague...
# [02:18] <webben> Hixie: not having to include alt will also be massively "misused".
# [02:18] <Hixie> it already is
# [02:18] <Hixie> i don't think this would make things worse
# [02:18] <webben> What's the reasoning for that?
# [02:18] <Hixie> if anything it might be a net benefit since we might reduce the amount of totally bogus alt text, which is arguably worse than no alt text in some cases
# [02:19] <webben> is there evidence that totally bogus alt text is a worse overall problem for users than no alt text?
# [02:20] <webben> and might that change if the "no alt text" group got higher?
# [02:20] <webben> *larger
# [02:21] <Hixie> "I then went on safari. blahblah The elephants were cool, but the tigers were the best: bdasflsjd" is not as good an experience as "I then went on safari. [IMAGE] The elephants were cool, but the tigers were the best: [IMAGE]"
# [02:22] <Hixie> (not that either is especially great, of course)
# [02:22] <webben> i'm not talking about the individual experience
# [02:22] <webben> i'm talking about which actually constitutes the bigger problem for end-users on the web now.
# [02:22] <Hixie> (I posit that the second one above is also better than "I then went on safari. elephants The elephants were cool, but the tigers were the best: tigers")
# [02:23] <webben> I posit "I then went on safari. image elephants The elephants were cool, but the tigers were the best: image tigers " is even better.
# [02:23] <Hixie> (I personally alsos think the [IMAGE] case, where "[IMAGE]" is something the UA shows in a different font/voice/color/whatever, is better than "I then went on safari. photo The elephants were cool, but the tigers were the best: photo")
# [02:24] <Hixie> i don't know of any reasearch into what problems blind users have experienced on the web so far
# [02:25] <Hixie> my own personal attempts at using screen readers have indicated that the screen readers themselves are a far bigger problem than anything the web pages do (based primarily on my experience with JAWS)
# [02:25] <webben> well it's not just blind users who suffer from no alt and bogus alt.
# [02:26] * Joins: myakura (n=myakura@p3216-ipbf5106marunouchi.tokyo.ocn.ne.jp)
# [02:26] <Hixie> text browser users suffer from it a lot too, but there aren't many of us (i use text browsers several times a week and alt="" isn't a big deal for me, though personally I would much rather people omit alt="" than give bogus or repeating alt"")
# [02:26] <BenMillard> search engines are also a UA which makes use of alt for the benefit of many users, for example, image searches
# [02:27] <Hixie> i work for a search engine
# [02:28] <Hixie> i assure you alt="" is the least of our worries
# [02:28] <Hixie> though if we were forced to comment on it, we'd much rather have no alt="" at all than have bogus alt=""
# [02:28] <Hixie> repetition isn't such a big deal for us though
# [02:29] <Hixie> since we can just filter out repeated text
# [02:29] <webben> also voice recognition users
# [02:29] <Hixie> voice recognition users?
# [02:29] <webben> *speech recognition, sorry
# [02:29] <Hixie> you mean for input? or do you mean speech synthesis?
# [02:29] <webben> input
# [02:30] <Hixie> i don't understand how alt="" affects input, can you elaborate?
# [02:30] <webben> images and icons that are controls mainly
# [02:30] <webben> e.g. an image of home, you say "home", the speech recognition looks for a link or button with "home" as a detectable text equivalent
# [02:31] * Quits: svl (n=me@ip565744a7.direct-adsl.nl) ("And back he spurred like a madman, shrieking a curse to the sky.")
# [02:31] <webben> I think in practice they have to make a lot of use of title too; I dunno how they try to handle missing alt and title other than maybe link numbering.
# [02:32] <webben> Hixie: here's some actual HTML guidelines from a vendor: http://ct.scansoft.com/customerfiles/kbasefiles/3067/wp_DNS_HTML.pdf in case that's of interest
# [02:33] <Hixie> oh well for images that are controls the spec is clear that alt="" is required
# [02:33] <webben> I was going to ask. That's good at least.
# [02:33] <webben> that won't change if you make alt non-conforming?
# [02:34] <webben> what if you have a div with tabindex and onclick and an img inside it?
# [02:34] <Hixie> the only thing in http://www.whatwg.org/specs/web-apps/current-work/multipage/embedded0.html#alt that is likely to change is the tiny subsubsection labeled "Images whose contents are not known"
# [02:35] <webben> yeah, problem is it's not easy for a validator to distinguish those from other images.
# [02:35] * Hixie looks at the file above
# [02:35] <Hixie> oh well images are far from the biggest problem there
# [02:36] <Hixie> i mean, you have to use <h1> only for headers but how is the validator supposed to know?
# [02:37] <webben> Hixie: well, yes, it's a general problem with conformance testing. but in terms of trying to ensure control images have alt, it's potentially important
# [02:37] <webben> depends how you define control images in such a way that a checker could distinguish the two
# [02:38] <Hixie> i don't think we can
# [02:38] <Hixie> but we have this problem anyway distinguishing decorative elements from other elements
# [02:38] <webben> Hixie: can you define a subset of controls that can be checked?
# [02:38] <Hixie> well if you read the link i gave above you'll see one of the first things it does say though is that an image in a link that has no other text must have non-empty alt=""
# [02:39] <Hixie> which is probably all wecan do really
# [02:39] <Hixie> machine-checkability-wise, anyway
# [02:40] <Hixie> dragon's document seems in line with what the spec says today
# [02:40] <Hixie> doesn't really affect the problematic images we were talking about
# [02:40] <webben> also button and input type="image" and imagemap?
# [02:41] <Hixie> <input type=image> and <area> are dealt with separately and have separate requirements that basically always require non-empty alt=""
# [02:41] <webben> Hixie: I don't think it does affect the problematic images, no.
# [02:41] <webben> okay
# [02:41] <Hixie> (well <input> isn't in the spec yet but will be)
# [02:41] <webben> button?
# [02:41] <Hixie> what about it?
# [02:41] <webben> shouldn't the spec mention button along with link at the above url?
# [02:41] <Hixie> oooh, good call
# [02:43] * Parts: BenMillard (n=cerbera@cpc1-flee1-0-0-cust285.glfd.cable.ntl.com)
# [02:46] * Quits: sverrej_ (n=sverrej@89.10.27.245) (Read error: 110 (Connection timed out))
# [02:47] * Joins: sverrej_ (n=sverrej@89.10.27.245)
# [03:02] * Quits: shepazu (n=schepers@80.187.212.118) (Read error: 110 (Connection timed out))
# [03:05] <Hixie> webben: spec updated
# [03:06] * Joins: jeremyb_ (n=jeremyb@unaffiliated/jeremyb)
# [03:43] * Quits: tndH (i=Rob@adsl-77-86-6-71.karoo.KCOM.COM) ("ChatZilla 0.9.83-rdmsoft [XULRunner 1.9/2008061013]")
# [03:49] * Joins: shepazu (n=schepers@80.187.212.118)
# [04:04] * Joins: eseidel (n=eseidel@c-24-6-171-94.hsd1.ca.comcast.net)
# [04:18] * Quits: shepazu (n=schepers@80.187.212.118) (Read error: 110 (Connection timed out))
# [04:30] * Joins: MikeSmith (n=MikeSmit@EM60-254-240-98.pool.e-mobile.ne.jp)
# [04:33] * Quits: MikeSmith (n=MikeSmit@EM60-254-240-98.pool.e-mobile.ne.jp) (Remote closed the connection)
# [05:17] * Quits: csarven (n=csarven@modemcable144.140-202-24.mc.videotron.ca) (Read error: 110 (Connection timed out))
# [05:18] * Joins: csarven (n=csarven@modemcable144.140-202-24.mc.videotron.ca)
# [05:32] * Quits: tantek (n=tantek@adsl-63-195-114-133.dsl.snfc21.pacbell.net)
# [05:49] * Joins: tantek (n=tantek@c-67-161-5-143.hsd1.ca.comcast.net)
# [05:58] * Quits: KevinMarks (n=KevinMar@c-98-207-134-151.hsd1.ca.comcast.net) ("The computer fell asleep")
# [06:07] * Joins: tantek_ (n=tantek@c-67-161-5-143.hsd1.ca.comcast.net)
# [06:08] * Quits: tantek_ (n=tantek@c-67-161-5-143.hsd1.ca.comcast.net) (Client Quit)
# [06:12] * Quits: jeremyb_ (n=jeremyb@unaffiliated/jeremyb) (Read error: 54 (Connection reset by peer))
# [06:13] * Joins: jeremyb_ (n=jeremyb@unaffiliated/jeremyb)
# [06:15] * Quits: csarven (n=csarven@modemcable144.140-202-24.mc.videotron.ca) (Read error: 110 (Connection timed out))
# [06:16] * Joins: jeremyb__ (n=jeremyb@162.84.188.152)
# [06:16] * Quits: jeremyb_ (n=jeremyb@unaffiliated/jeremyb) (Read error: 104 (Connection reset by peer))
# [06:22] <takkaria> urgh, 521 new mails
# [06:23] * Quits: tantek (n=tantek@c-67-161-5-143.hsd1.ca.comcast.net) (Read error: 110 (Connection timed out))
# [06:26] * Joins: tantek (n=tantek@c-71-202-120-131.hsd1.ca.comcast.net)
# [06:30] * Quits: jeremyb__ (n=jeremyb@unaffiliated/jeremyb) (Read error: 54 (Connection reset by peer))
# [06:30] * Joins: jeremyb_ (n=jeremyb@unaffiliated/jeremyb)
# [07:05] * Quits: jeremyb_ (n=jeremyb@unaffiliated/jeremyb) (Connection reset by peer)
# [07:06] * Joins: jeremyb_ (n=jeremyb@unaffiliated/jeremyb)
# [07:10] * Quits: tantek (n=tantek@c-71-202-120-131.hsd1.ca.comcast.net)
# [07:36] * Quits: roc (n=roc@202.0.36.64)
# [07:47] * Quits: eseidel (n=eseidel@c-24-6-171-94.hsd1.ca.comcast.net)
# [08:08] * Quits: virtuelv (n=virtuelv@163.80-202-65.nextgentel.com) ("Leaving")
# [08:10] * Quits: mcarter (n=mcarter@adsl-71-135-109-164.dsl.pltn13.pacbell.net) (kornbluth.freenode.net irc.freenode.net)
# [08:15] * Joins: mcarter (n=mcarter@adsl-71-135-109-164.dsl.pltn13.pacbell.net)
# [08:22] * Joins: tantek (n=tantek@adsl-63-195-114-133.dsl.snfc21.pacbell.net)
# [09:06] * Quits: jeremyb_ (n=jeremyb@unaffiliated/jeremyb) (Read error: 104 (Connection reset by peer))
# [09:07] * Joins: jeremyb_ (n=jeremyb@unaffiliated/jeremyb)
# [09:08] * Joins: MikeSmith (n=MikeSmit@EM60-254-223-42.pool.e-mobile.ne.jp)
# [09:12] * Joins: shepazu (n=schepers@80.187.212.118)
# [09:18] <Hixie> what does the "@" mean in "-rw-r--r--@"?
# [09:20] <hsivonen> Hixie: ACL I think
# [09:21] <Hixie> any idea how i view the acl info?
# [09:22] <Hixie> ls -e didn't work
# [09:23] <jacobolus> Hixie: http://arstechnica.com/reviews/os/macosx-10-4.ars/8
# [09:23] <jacobolus> really, -e didn't work?
# [09:23] <jacobolus> hrm
# [09:23] <Hixie> that suggests @ does not mean acl
# [09:23] <Hixie> since no @ appears in their listings
# [09:23] <jacobolus> indeed
# [09:23] <hsivonen> Hixie: I can't remember and Googling gives me the article jacobolus mentioned
# [09:23] <Hixie> maybe it means there's fork data or something
# [09:24] <Hixie> or attributes
# [09:24] <hsivonen> I might misremember the meaning of @
# [09:24] * Quits: jeremyb_ (n=jeremyb@unaffiliated/jeremyb) (Read error: 104 (Connection reset by peer))
# [09:25] * Joins: jeremyb_ (n=jeremyb@unaffiliated/jeremyb)
# [09:25] <jacobolus> Hixie: If the file or directory has extended attributes, the permissions field printed by the -l option is followed by a '@' character.
# [09:26] <hsivonen> xattr --list file
# [09:26] <jacobolus> so actually the page you want is http://arstechnica.com/reviews/os/macosx-10-4.ars/7 :)
# [09:26] <Hixie> yeah i was just playing with mdls which also seems to list attributes
# [09:27] <Hixie> ok anyway
# [09:27] <Hixie> it is an attributes thing
# [09:28] <jacobolus> no, mdls lists info from the spotlight store
# [09:28] <jacobolus> which is different from xattrs
# [09:28] * Joins: harig (n=harig_in@pat-tdc.opera.com)
# [09:29] <hsivonen> are ACLs a special case of attributes?
# [09:29] <jacobolus> and it is *very very* frustrating that setting custom xattrs can’t put custom metadata in spotlight :(
# [09:29] <jacobolus> hsivonen: yes, ACLs are stored in xattrs
# [09:29] <hsivonen> ok. so my memory isn't totally broken
# [09:30] <Hixie> man mac is confusing :-P
# [09:31] <jacobolus> hsivonen: I'm not sure you can get to them though through normal xattr APIs
# [09:32] * Hixie sighs loudly
# [09:33] <Hixie> i wish i had some data to go on with this <img> thing
# [09:33] <Hixie> i really have no idea what to do
# [09:33] <Hixie> which is a pretty ridiculous position to be in given the rarity with which this will affect users
# [09:33] <Hixie> (it's not like blind users surf to flickr often)
# [09:34] <jacobolus> why do you have no idea what to do?
# [09:34] <jacobolus> I thought it was figured out
# [09:34] <jacobolus> with the {}'s etc.
# [09:34] <Hixie> people have raised a valid concern with the {}
# [09:35] <Hixie> which is that it increases the effort of people trying to write accessible content
# [09:35] <hsivonen> Hixie: do you have data on how often people with text input difficulties execise their right to speech by publishing photos?
# [09:35] <jacobolus> increases the effort?
# [09:35] <jacobolus> as in, it's too hard to put {…} in?
# [09:35] <Hixie> because they now have to work around cases where the alt="" might end up containing {} through no fault of their own, e.g. if the alt is user input
# [09:36] <tantek> what's wrong with {} ?
# [09:36] <Hixie> hsivonen: 100% of my sample of 1 has done it
# [09:36] <jacobolus> that seems like a vanishingly unlikely edge case
# [09:36] <jacobolus> what naïve user is going to make the whole alt name start w/ { and end w/ }?
# [09:37] <Hixie> jacobolus: yeah but the people who care about accessibility (the group we want to piss off the least) will want to do the right thing in all cases anyway, so it will disproportionately hurt them
# [09:37] * Joins: hdh (n=hdh@118.71.123.99)
# [09:38] <hsivonen> tantek: if you give the user a text field for writing alt text, your program needs to do something when the value entered into the field starts with { and ends with }, since just passing that through would leak user input into writing the wrong meaning into the file format
# [09:38] <jacobolus> Hixie: so “do the right thing” here will be interpreted as what? strip out the {} if users add them?
# [09:38] <Hixie> jacobolus: or add a space or something
# [09:38] <jacobolus> bleh
# [09:38] <jacobolus> Hixie: it strikes me that it's still easier to “do the right thing” under the {} system than currently
# [09:38] <tantek> why would { } be the wrong meaning?
# [09:39] <jacobolus> e.g. what if a user types nothing in the box?
# [09:39] <tantek> what if the img was a picture of some curly braces?
# [09:39] <Hixie> tantek: {...} has special meaning according to html5 right now (will likely change)
# [09:39] <Hixie> jacobolus: like i said, i've no idea what to do
# [09:39] <tantek> then escape it
# [09:40] <tantek> & has special meaning hence &
# [09:40] <tantek> not sure what the big deal is
# [09:40] <hsivonen> tantek: right, so escaping means more data munging cases to handle than putting the flag out of band into a separate attribute
# [09:40] <Hixie> tantek: there's no defined escaping mechanism for this right now, but even if there was, the big deal is that we are making life harder for the only group who are doing the right thing, which is staggeringly bad language design
# [09:40] <jacobolus> so what's the alternate proposal?
# [09:40] <jacobolus> make up a new attribute?
# [09:41] <Hixie> i have 23 alternative proposals on my list right now
# [09:41] <Hixie> they all suck in one way or another
# [09:41] <jacobolus> eep
# [09:41] <jacobolus> how many with any realistic chances?
# [09:41] <Hixie> 0
# [09:42] <tantek> what is the URL that documents the need for this { } feature?
# [09:42] <jacobolus> tantek: whatwg.com/html5
# [09:42] <Hixie> whatwg.org
# [09:42] <jacobolus> erm, org
# [09:42] <Hixie> but the spec doesn't document the need
# [09:43] <tantek> the use case, the background, the group/people that asked for it etc.
# [09:43] <Hixie> tantek: http://lists.w3.org/Archives/Public/public-html/2008Aug/0602.html documents some of it
# [09:43] <Hixie> search for {...}
# [09:43] <Hixie> er
# [09:43] <Hixie> actually
# [09:43] <tantek> similar problem exists today if someone exj�8s < or " or > into user input meant for an alt attribute right?
# [09:43] <Hixie> search for "the tentative decision"
# [09:43] <Hixie> no, we have a defined escaping for that and it is wildly implemented already
# [09:44] <hsivonen> I really need to get to the bottom of 4 concurrent validations of up to 1 MB pages exhausting 1 GB of RAM
# [09:44] <hsivonen> there has to be some silly data structure growth somewhere
# [09:44] <hsivonen> and some crazy markup out there on the Web
# [09:44] <Hixie> jacobolus: http://damowmow.com/temp/alt lists the proposals so far
# [09:45] <Hixie> hsivonen: how is the 1,000,000 validations plan going?
# [09:45] <hsivonen> Hixie: I was able to download 93% of the pages successfully. validating them now
# [09:45] <hsivonen> Hixie: it seems that either the parser or the validator has a memory leak
# [09:45] <Hixie> ah
# [09:45] <hsivonen> (I'm reusing the parser and validator objects)
# [09:46] <jacobolus> Hixie: just go for B.1
# [09:46] <hsivonen> the dmoz validation run hung for unknown reasons
# [09:46] <Hixie> jacobolus: disallowing flickr is not a workable solution
# [09:47] <jacobolus> (joke)
# [09:49] <Hixie> oh many people think B.1 is a good solution
# [09:49] <jacobolus> many people are nutgs
# [09:49] <hsivonen> Hixie: oh, and real Web content has made a couple of assertions to fire
# [09:49] <jacobolus> or nuts even
# [09:49] <hsivonen> which is interesting
# [09:49] <Hixie> real web content is insane
# [09:50] * Quits: jeremyb_ (n=jeremyb@unaffiliated/jeremyb) (Read error: 104 (Connection reset by peer))
# [09:50] * Joins: jeremyb_ (n=jeremyb@unaffiliated/jeremyb)
# [09:50] <hsivonen> also, Japanese Web content exposed a bug in my charset decoder driver code that I hadn't seen with UTF-8 content
# [09:53] * Quits: sverrej_ (n=sverrej@89.10.27.245) (Read error: 110 (Connection timed out))
# [09:54] <jacobolus> Hixie: what would the practical effect be if a website allowed user input and the user put in “{picture of grandma on her 80th birthday}” or whatever?
# [09:54] <jacobolus> i.e. what would a screen reader do with that content
# [09:55] <Hixie> well that would actually be correct alternative text for that image
# [09:55] <Hixie> it would cause the UA to attract attention to the fact that there is an image with the caption "picture of..."
# [09:55] <jacobolus> that is, if the {} proposal goes through
# [09:56] <jacobolus> how is {foo} treated differently from foo
# [09:56] <Hixie> right that 's what i was assuming
# [09:56] <Hixie> {foo} causes the UA to attract attention to the fact that there is an image
# [09:56] <Hixie> foo causes the UA to say "foo" without mentioning the image
# [09:56] <jacobolus> I see
# [09:56] <jacobolus> that seems like pretty minimal harm then from the vanishingly unlikely false positives
# [09:57] * Joins: virtuelv (n=virtuelv@pat-tdc.opera.com)
# [09:57] <Lachy> Hixie, that doesn't match what existing screen readers doo with the "foo" case. They do say "image: foo"
# [09:57] <virtuelv> JohnResig: yt?
# [09:57] <Hixie> existing screen readers fail long before they get around to rendering any html
# [09:57] <Hixie> jacobolus: i agree
# [09:57] <Hixie> jacobolus: won't stop the people who care from being even more careful though
# [09:57] <jacobolus> I mean, {} aren’t even meaningful characters in english
# [09:58] <tantek> FWIW - as someone who prefers not to see new syntax, capturing such an annotation in another attribute seems preferable to overloading *both* an existing attribute and the use of characters which could mean something themselves.
# [09:58] <Lachy> sure, but in the "foo" case, why should they stop notifying the user that there's an image there?
# [09:58] <jacobolus> maybe someone will accidentally put them in a description of a mathematical formula or something
# [09:58] * Joins: annevk2 (n=annevk@77.163.243.203)
# [09:59] <Hixie> tantek: another attribute doesn't work because the abuse such an attribute would get would almost certainly cause more damage than the little benefit it would cause
# [09:59] * Quits: annevk (n=annevk@77.163.243.203) (Read error: 104 (Connection reset by peer))
# [09:59] * Joins: roc (n=roc@121-72-167-161.dsl.telstraclear.net)
# [09:59] <Hixie> tantek: we've already seen that authors are completely incapable of reliably making use of attributes like alt="" or longdesc=""
# [09:59] <Hixie> tantek: and this would be an even more subtle attribute
# [10:00] * Quits: jeremyb_ (n=jeremyb@unaffiliated/jeremyb) (Read error: 54 (Connection reset by peer))
# [10:00] <Hixie> Lachy: the whole point of alt="" is that it is replacement text that removes mention of the image
# [10:00] <jacobolus> virtuelv: isn't it 4 AM in EDT?
# [10:00] * Quits: shepazu (n=schepers@80.187.212.118) (Read error: 110 (Connection timed out))
# [10:00] <virtuelv> jacobolus: probably
# [10:00] <virtuelv> I'll have to catch up when I get to Turin this afternoon then
# [10:00] <Hixie> jacobolus: the original complaint re alt={} was latex, fwiw, which could quite legitimately end up with {}
# [10:00] * Joins: jeremyb_ (n=jeremyb@unaffiliated/jeremyb)
# [10:01] <virtuelv> I'm a bit puzzled about one thing he's doing with Sizzle
# [10:01] * Joins: shepazu (n=schepers@80.187.212.118)
# [10:01] <jacobolus> how does it legitimately end w/ {} ??
# [10:02] <Hixie> i mean {somethingsomething}
# [10:02] <jacobolus> Hixie: you just mean that's what latex → html converters do currently?
# [10:02] <Hixie> yes
# [10:02] <jacobolus> probably one particular latex → html converter, which could be trivially changed
# [10:04] <tantek> Hixie, your point that the cognitive/incentive barrier to usage of such an attribute makes it useless is a good one, and justifies not only not introducing such an effectively useless attribute, but any such equivalent mechanism, *especially* one that introduces additional syntax.
# [10:05] <Hixie> tantek: agreed
# [10:06] <Hixie> tantek: so that suggests making alt="" optional for this case, but then we have no way to orient speech synthesis users when they navigate image-by-image
# [10:06] <tantek> the lack of a mechanism does not rationally justify introducing an ineffective mechanism
# [10:06] <Hixie> tantek: agreed
# [10:07] <Hixie> tantek: however the aforementioned case is important enough to certain authors that at least one has said they will always include some alternative text, even if it is not suitable for linear navigation (i.e. reading the page straight through)
# [10:08] <tantek> do they do so now on existing pages?
# [10:08] <Hixie> those do yeah
# [10:08] <Hixie> it makes those pages harder to read if you have images disabled though, e.g. in a text browser
# [10:08] <Hixie> which is bad
# [10:08] <Hixie> and so not really what we want to require
# [10:09] <hsivonen> those authors have *way* fewer photos in their photostream than tantek
# [10:09] <Hixie> yup
# [10:10] <tantek> below what author/content/sample size threshold do you reach the point of diminishing returns of language design improvements/features?
# [10:11] * Quits: myakura (n=myakura@p3216-ipbf5106marunouchi.tokyo.ocn.ne.jp) ("Leaving...")
# [10:12] <Hixie> tantek: oh we're way below that already
# [10:13] <Hixie> tantek: but i still have to say _something_
# [10:13] <hsivonen> Hixie: it seems to me, though, that part of the problem is that you want the result to make sense in Lynx instead of sacrificing Lynx and focusing solely on the AT case
# [10:15] <Hixie> as a regular lynx user, i don't think it's unreasonable to have things make sense in lynx
# [10:15] <Hixie> it's also a matter of it making sense in graphical browsers with images disabled
# [10:16] <hsivonen> Hixie: that means you are letting non-accessibility issues get in the way of accessibility issues
# [10:16] <Hixie> this isn't an accessibility issue
# [10:16] <Hixie> it's an html issue
# [10:17] <hsivonen> right :-)
# [10:18] * Quits: jeremyb_ (n=jeremyb@unaffiliated/jeremyb) (Read error: 104 (Connection reset by peer))
# [10:18] <hsivonen> however, the accessibility alt-need conditions aren't curable. the need to turn off images in a graphical browser is curable (by flat-rate 3G/EDGE mobile subscriptions)
# [10:18] * Joins: jeremyb_ (n=jeremyb@unaffiliated/jeremyb)
# [10:19] <Hixie> yeah well it might be curable in theory, but many parts of africa aren't getting 3G in the next ten years
# [10:19] <Hixie> and hell, many parts of america don't have decent EDGE
# [10:19] <tantek> another case for alt text being readable is the very common occurrence of web based mail interfaces (e.g. Gmail) presenting HTML email with image loading disabled for security/privacy reasons.
# [10:20] <Hixie> hsivonen: and frankly, it's only a problem for AT users when they navigate image-by-image
# [10:20] <hsivonen> Hixie: the experts say that's the usual mode of operation
# [10:21] * Joins: virtuelv_ (n=virtuelv@213.236.208.247)
# [10:21] <Hixie> hsivonen: sure, so we have to handle it
# [10:21] <hsivonen> Hixie: after all, VO, for one, doesn't even have sane UI for reading content in a continuous manner
# [10:21] <tantek> hsivonen, the usual mode of operation of web based email interfaces is image loading disabled.
# [10:21] <Hixie> anyway
# [10:22] <hsivonen> (I'd use VO for non-accessibility things if it had a "read on from here" command)
# [10:22] <Hixie> no. idead. what. to. do.
# [10:22] * Quits: virtuelv (n=virtuelv@pat-tdc.opera.com) (Read error: 110 (Connection timed out))
# [10:22] <Hixie> hsivonen: just select a bunch of text and use the speech service
# [10:22] <tantek> thus the Lynx / graphical browsers with images disabled case is far more common than those two specific use cases might imply
# [10:23] <Hixie> doesn't really matter how common all these cases are, they're all common enough that we have to deal with them
# [10:25] * Quits: harig (n=harig_in@pat-tdc.opera.com) (Read error: 110 (Connection timed out))
# [10:26] * Joins: Maurice (i=copyman@cc498817-b.emmen1.dr.home.nl)
# [10:28] <hsivonen> it seems to me that the headers/id discussion and the RDFa discussion revolve around a similar issue (except headers/id has more UA legacy than RDFa)
# [10:29] <Hixie> they certainly have in common that i don't understand the problems they are trying to solve
# [10:29] <hsivonen> that is, if we consider something like "simple things should be easy and complex things should be possible", there's one faction that is willing to make simple cases less easy in order to support complex cases with the same model and another faction that mainly cares about making simple case easy and is willing to cut support for some complex cases
# [10:30] <annevk2> hsivonen, turning images off is also common in e.g. Russia
# [10:30] * Quits: Lachy (n=Lachlan@85.196.122.246) ("This computer has gone to sleep")
# [10:30] * annevk2 is now known as annevk
# [10:31] <annevk> as bandwidth is ridiculously expensive there still
# [10:32] * Joins: heycam (i=cam@88.128.89.135)
# [10:32] <hsivonen> Hixie: I *think* I understand the problems RDFa is supposed to solve. In the case of ccREL, I just think that some of the problems are non-problem and some problem won't be successfully solved by the proposed solution
# [10:32] <hsivonen> Hixie: also, I don't think that the generic problem RDFa solves is a problem whose solutions yield good markup
# [10:33] <Hixie> hsivonen: i'm hoping ben adida can actually explain the problem
# [10:34] <Hixie> i can come up with problems that i think maybe they are solving
# [10:34] <hsivonen> I think RDFa proponents aren't doing their cause any favors when their conference slides take well-known microformats and then show how it can by done "right" with RDFa
# [10:34] <hsivonen> and the RDFa version is always more verbose
# [10:34] <hsivonen> i.e. crufty
# [10:34] <annevk> yeah, that's funny
# [10:35] * Joins: aaronlev (n=chatzill@e179056253.adsl.alicedsl.de)
# [10:35] * Quits: jeremyb_ (n=jeremyb@unaffiliated/jeremyb) (Read error: 104 (Connection reset by peer))
# [10:36] <Hixie> can anyone find the link to joshue's usability videos?
# [10:36] * Joins: jeremyb_ (n=jeremyb@unaffiliated/jeremyb)