| 02:39 | <MikeSmith> | if anybody here is familiar with the process of uploading stuff to Maven Central and can help me some time soon, I'd appreciate it |
| 02:40 | <MikeSmith> | in order to resolve https://github.com/validator/validator.github.io/issues/10 |
| 02:40 | <MikeSmith> | the OP there provides a link to http://maven.apache.org/guides/mini/guide-central-repository-upload.html but that guide seems near worthless |
| 02:41 | <MikeSmith> | at least worthless to somebody like me who knows next to nothing about Maven in general |
| 06:33 | <ionas> | If I have a semantic set of data called “record” and I want to display 1 record by specifing data of its fields - how would you map that most semantically? |
| 06:39 | <ionas> | Does this look all right? |
| 06:39 | <ionas> | http://pastie.org/9582924 |
| 07:08 | <zcorpan> | mathiasbynens: maybe our showModalDialog article should mention that only a <dialog> polyfill might not be enough if there's a lot to refactor - http://thinkingoutsidetheanglebrackets.blogspot.co.uk/2014/09/removal-of-showmodaldialog-and.html |
| 07:15 | <annevk> | I only use relative dates on the home page |
| 07:45 | <annevk> | Hixie_: bug reporting from the spec seems to take longer? |
| 07:47 | <ionas> | seems neither <nav> nor <section> want to live without at least a h6 |
| 07:51 | <ionas> | http://pastie.org/9583128 - trying to represent a data record most semantically |
| 08:19 | <MikeSmith> | annevk: for ServiceWorker, about the cause of the (proposed) decision from Google to not ship offline/cache is v1, is it maybe due in part to https://github.com/slightlyoff/ServiceWorker/issues/468 from mnot? |
| 08:20 | <annevk> | MikeSmith: I don't know |
| 08:20 | <MikeSmith> | ok |
| 08:21 | <MikeSmith> | I guess I'll just talk with Kenji |
| 08:21 | <MikeSmith> | ionas: I'd just use <div> |
| 08:21 | <MikeSmith> | ionas: with a class value if you want -- like, <div class=record> |
| 08:22 | <ionas> | MikeSmith: its not just an abstract division though, each section groups certain types |
| 08:22 | <ionas> | and the record has a title |
| 08:22 | <ionas> | and is self-sustained - eg. an article |
| 08:22 | <MikeSmith> | I don't think any of that makes it unsuitable for marking up with <div> |
| 08:23 | <MikeSmith> | the semantic elements like article are kind of a lost cause IMHO |
| 08:23 | <ionas> | MikeSmith: just because you CAN use div/span everywhere |
| 08:23 | <ionas> | does not mean you have to |
| 08:23 | <MikeSmith> | yeah I didn't say you had to |
| 08:23 | <ionas> | I think they are fine, they also add options to style things without cluttering tons of classes |
| 08:23 | <ionas> | like <div class=“record”> |
| 08:24 | <ionas> | the worst thing lately happening is all that css-class-soup coming through those pseudo frontend frameworks |
| 08:25 | <MikeSmith> | if that's the worst thing you've seen happening lately, you're lucky |
| 08:26 | <ionas> | such a fall-back into dark-table-ages - ultra huge DOM trees due to just wrapping divs with classes, that’s the other reason to go “semantic” (thinking about what data you represent) - so personally I’d love to see many more semantic tags like record, product, etc - generic “nouns” that can be adapted for different use cases and have some guideline how to use |
| 08:26 | <ionas> | hehe ;) - yes bootstrap/div-soup/class-soup really made some projects a big pain due to bad practise/design. |
| 08:37 | <annevk> | Hixie_: might just be W3C Bugzilla that is very slow |
| 08:41 | <MikeSmith> | annevk: yeah, that |
| 08:42 | <MikeSmith> | w3c systems are having some load problems right now |
| 08:42 | <MikeSmith> | systems team is working on getting it resolved |
| 08:53 | <annevk> | jgraham: https://plus.google.com/+IanBicking/posts/Cz95yhYK3aG seems apt for the further investment in DreamHost strategy |
| 08:53 | <annevk> | jgraham: although I suppose I might still get fed up with it after a while |
| 09:24 | <jgraham> | annevk: Right, the correct evaluation is long term cost of moving to another host vs long term cost of staying on Dreamhost. The first option is front-loaded in that there is some switching cost (although you don't have to make a big-bang change of course), but there amy well be a break even point and it might not even be too far in the future |
| 09:24 | <Ms2ger> | Amy? |
| 09:24 | Ms2ger | ducks |
| 09:28 | <gsnedders> | Amy's vital to any moving plan! |
| 09:33 | <jgraham> | Yes, she's going to do the heavy lifting |
| 09:34 | <gsnedders> | Assigning elements to sections is hard, part ten: <p class=category>cats</p><h2>Pillar</h2><p>Pillar is a three-legged cat |
| 09:35 | <annevk> | Amy? |
| 09:35 | <annevk> | jgraham: as the switching cost involves learning how to manage a server, it seems pretty high |
| 09:42 | <jgraham> | annevk: Well how hard that is depends a lot on what you're doing. Running a simple HTTP server is easy. Running a mail server is hard enough that I'm not prepared to try it. Running a HTTP server and a DB server is somewhere in the middle |
| 09:57 | <hsivonen> | jgraham: fyi, your take on the W3C HTML5 spec is being spun into "Mozilla doesn't care about a11y" propaganda on Twitter |
| 09:58 | <Ms2ger> | Of course it is |
| 09:59 | <hsivonen> | Ms2ger: I'm not suggesting that this is surprising, but failing to point to specific bogosities helps the propaganda |
| 09:59 | <Ms2ger> | And thank you for that |
| 10:16 | <zcorpan> | anyone know of a "browser" icon in svg that is cc0? |
| 10:16 | <zcorpan> | that looks good at 16x16px |
| 10:21 | <MikeSmith> | zcorpan: https://openclipart.org/ |
| 10:22 | <MikeSmith> | if you didn't try that already |
| 10:27 | <zcorpan> | didn't find anything suitable |
| 10:31 | <jgraham> | hsivonen: It seems like "Mozilla a11y are implementing based on the W3C version of the spec" would be useful input to the dev.platform thread. That is something we can use as leverage to get observable differences backported to the WHATWG version, which seems like a good outcome. I also don't have kind words for people who perfer to drum up a Twitter storm rather than engage in reasoned discussion. |
| 10:32 | <karlcow> | hsivonen: no. |
| 10:33 | <karlcow> | the message of Ms2ger is. Not the one of jgraham. |
| 10:36 | <MikeSmith> | zcorpan: http://commons.m.wikimedia.org/wiki/File:Deer_park_globe.svg maybe |
| 10:37 | <zcorpan> | MikeSmith: it looks a lot like firefox nightly |
| 10:37 | <zcorpan> | but maybe |
| 10:38 | <MikeSmith> | also I see now it's cc-by -_- |
| 10:38 | <MikeSmith> | not cc0 |
| 10:39 | <zcorpan> | looks like it actually is the nightly icon :-) |
| 10:40 | <MikeSmith> | zcorpan: Yeah but I doubt you'll find 1 normal person out of 1000 who would know it's the nightly icon |
| 10:41 | <zcorpan> | this is for https://html5.org/tools/web-apps-tracker not for normal persons :-P |
| 10:41 | <MikeSmith> | Ah |
| 10:44 | <annevk> | looks like an old version of the icon |
| 10:46 | <MikeSmith> | if it's a icon for recognition by people who actually work on browser code and testing, you need something that connotes pain and anxiety |
| 10:47 | <MikeSmith> | e.g., more like an exploding planet |
| 10:51 | <MikeSmith> | http://33.media.tumblr.com/b338a6a8b147ae1edba94b1087fcdf02/tumblr_momln1lZYE1qetz3to1_500.jpg |
| 10:51 | <annevk> | Can someone help me debug this: |
| 10:51 | <annevk> | RedirectMatch 301 ^/tools/web-apps-tracker\?from\=(\d+)\&to\=(\d+)$ https://html5.org/r/$2 |
| 10:51 | annevk | is still bad with regular expressions |
| 10:53 | <karlcow> | regex101 |
| 10:54 | <karlcow> | annevk: do you have an example of two URIs? |
| 10:55 | <annevk> | https://html5.org/tools/web-apps-tracker?from=8781&to=8782 -> https://html5.org/r/8782 |
| 10:55 | <karlcow> | merci |
| 10:57 | <MikeSmith> | annevk: you don't need to escape the = sign, right? though I guess it should still wish just the same if you do |
| 10:57 | <karlcow> | yes no need |
| 10:57 | <karlcow> | and also the & |
| 10:57 | <MikeSmith> | nor the & sign |
| 10:57 | <annevk> | yeah had that before |
| 11:01 | <karlcow> | ^/tools/web-apps-tracker?from=[0-9]+&to=([0-9]+)$ https://html5.org/r/$1 |
| 11:01 | <karlcow> | As I'm not sure if apache knows \d |
| 11:02 | <annevk> | it does |
| 11:03 | <annevk> | doesn't work |
| 11:05 | <darobin> | you need to escape the ? |
| 11:06 | <karlcow> | ah! |
| 11:06 | <darobin> | and indeed Apache regexes are pretty dumb |
| 11:06 | <darobin> | annevk: try with the ? escaped |
| 11:06 | <karlcow> | ? <- conditionnal |
| 11:06 | <darobin> | yeah |
| 11:06 | <annevk> | it's escaped |
| 11:07 | <darobin> | mmmmmmmm |
| 11:07 | <darobin> | oh |
| 11:07 | <darobin> | I'm not sure you can match against the query string |
| 11:07 | <annevk> | are query parameters passed in differently? |
| 11:07 | <annevk> | yeah, seems like it |
| 11:08 | <annevk> | bah |
| 11:08 | <darobin> | mmmm RedirectMatch — that's not mod_rewrite |
| 11:09 | <karlcow> | >Another case is when a CGI needs to be redirected and the CGI takes arguments by the GET method. A "Redirect" or "RedirectMatch" command will redirect the CGI but will not send the CGI's arguments along with it to the new URL. |
| 11:09 | <karlcow> | RewriteEngine on + RewriteRule is your friend |
| 11:09 | <darobin> | annevk: I think that you want to use mod_rewrite instead |
| 11:11 | <darobin> | annevk: RewriteRule can make backreferences to RewriteCond matches, and the latter can match on the QS |
| 11:12 | <darobin> | it's painful though |
| 11:12 | <darobin> | you could turn web-apps-tracker into a small CGI script that would do this far more easily :) |
| 11:13 | <annevk> | it's a CGI script, but I was hoping not to have to figure out the CGI API again |
| 11:14 | <darobin> | annevk: what language? |
| 11:14 | <annevk> | Python |
| 11:14 | <annevk> | need to figure out how to get the request URL |
| 11:14 | <darobin> | annevk: I presume it's already referencing the from/to values, |
| 11:14 | <darobin> | all you need to add is how to send a 301 |
| 11:14 | <annevk> | I think then I'll get into external redirects :-) |
| 11:14 | <karlcow> | Ooooh https://bugzilla.mozilla.org/show_bug.cgi?id=958887 |
| 11:18 | <darobin> | oh, it looks like it's just print 'Status: 301...' and then the Location header |
| 11:18 | <darobin> | nicely close to the metal :) |
| 11:39 | <hsivonen> | karlcow: there are people on twitter who are particularly taking issue with jgraham's wording |
| 11:56 | <zcorpan> | what wording? |
| 12:18 | <mathiasbynens> | zcorpan: this seems up your alley https://github.com/cure53/DOMPurify/issues/32#issuecomment-55027268 |
| 12:21 | <zcorpan> | mathiasbynens: https://github.com/cure53/DOMPurify/issues/32#issuecomment-55896661 seems to say they don't need a css parser anyway, can just iterate the cssom |
| 12:22 | <zcorpan> | but certainly making the cssom suck less is a long term goal |
| 12:23 | <zcorpan> | a problem with iterating the cssom is that you can't do it until the stylesheet is already applied |
| 12:23 | <zcorpan> | but i'm not sure how this thing works to begin with |
| 12:48 | <SteveF_> | hsivonen: "it doesn't do a11y any favors to hold the permathreads about longdesc as a symbol of pro-a11y activity. " who has been doing that? http://lists.w3.org/Archives/Public/public-html-admin/2014Aug/0038.html |
| 13:31 | <Ms2ger> | karlcow, hmm? People complaining about me? |
| 13:32 | <zcorpan> | does anyone here know how css positioning works with rtl and scrollable elements? http://lists.w3.org/Archives/Public/www-style/2014Sep/0294.html |
| 13:34 | <gsnedders> | Ms2ger: it's all your fault |
| 13:45 | <jgraham> | SteveF_: FWIW I think that, although longdesc hasn't been mentioned here, it is the constant elephant in the room for a11y discussions. The discussion reached a point where it was clear to me, at least, that if I go blind, anything that required authors to use longdesc would simply not be accessible to me, whether or not it is in the HTML spec. At that point a valid tactic on the "WHATWG side" (if you want that kind of unhelpful characterisation) wo |
| 13:45 | <jgraham> | ... think this attribute is entirely useless if not outright harmful, but it's not worth fighting about". Obviously that didn't happen, and perhaps it was a mistake. On the other side proponents of the attribute showed very little sign that they were engaging with the evidence that it wasn't really a helpful design. Instead they sunk a lot of political capital — more than they could really afford — on forcing it through. Several of the participa |
| 13:45 | <jgraham> | ... had toxic personalities and people weren't prepared to cut them out of the discussion, or disown their behaviour, particularly when they happened to agree with their viewpoint on this one issue. This discussion was one of the things that led to the group becoming a lot more bureaucratic with polls and extension specs and so on. For implementors, maintaining a high cadence is extremely important, not just because it isn't anyone's job to engage i |
| 13:45 | <jgraham> | ... debates, but also because it's necessary to keep improving specific products and the platform as a whole so as not to lose in the marketplace either against direct competitors or whole new technology stacks that don't come with all the legacy baggage. With that in mind a forum with protocols designed to neuter multi-year debates over a single attribute aren't suitable for any other kind of work, and that's why implementors lost interest. |
| 13:46 | <darobin> | well, that's a long thing to say on IRC |
| 13:47 | <jgraham> | So it seems |
| 13:48 | <SteveF_> | jgraham: like many others active in acc specification of html and related features, I have spent very little time in the past few years on longdesc, but it is usually waved around in discussions as it is an easy target |
| 13:52 | <SteveF_> | jgraham: my input on the URL spec thread may be of interest as its subject is relevant to the issues - http://lists.w3.org/Archives/Public/public-w3process/2014Sep/0015.html *warning* lacks hyperbole and fisticuffs |
| 13:53 | <jgraham> | SteveF_: Right, but this is something that happened a few years ago. The fact that everyone — in some particularly unhelpful way — got what they wanted, with a WHATWG spec not containing longdesc and a W3C extension spec containing longdesc, makes it unsurprising that people moved on to new things in the meantime. But the effects of that earlier phase are still shaping the way we operate today. |
| 13:56 | <Ms2ger> | jgraham, your wall of text above is missing things, fwiw |
| 13:56 | <jgraham> | Ms2ger: Not for me! |
| 13:56 | <tantek> | whoa - email in IRC - amazing! |
| 13:56 | <Ms2ger> | characterisation) wo |
| 13:56 | <Ms2ger> | <jgraham> ... think this |
| 13:57 | <jgraham> | 'would have been to say "we' |
| 13:57 | <SteveF_> | jgraham: i agree, I moved on to more interesting stuff to me, I interact with the whatwg on rare occasions I need to, rarely a pleasant experience, fact is the acc layer implementation stuff does not happen whatwg |
| 13:58 | <SteveF_> | at whatwg |
| 13:58 | <tantek> | this: "the effects of that earlier phase are still shaping the way we operate today" |
| 13:58 | <SteveF_> | nor does the specification of acc layer stuff |
| 13:59 | <Ms2ger> | SteveF_, yeah, so it seems |
| 13:59 | <Ms2ger> | SteveF_, I haven't followed that, though, so I'm not sure how that's relevant to the HTML 5 PR? |
| 13:59 | <tantek> | SteveF - where does specification of acc layer stuff happen that is *not* so political, and instead actually data-driven on helping the most people by creating acc tech that easy *and* reliable for authors to use? |
| 14:01 | <SteveF_> | tantek: at w3c in mailing lists, bugs etc, on browser bugs in discussion on irc , the usual |
| 14:01 | <annevk> | mathiasbynens: what's the status of javascript.spec.whatwg.org? |
| 14:02 | <annevk> | mathiasbynens: are you planning on pushing out an update to align with ES6? |
| 14:02 | <tantek> | SteveF - that hasn't been my experience with w3c mailing lists, bugs, etc. |
| 14:03 | <SteveF_> | rleevance to PR for me is that i went through the requirements in the aria section of the w3c HTML spec and tested them where there were issues I filed bugs either against browsers or against the spec |
| 14:03 | <tantek> | in that way, "longdesc" has served a purpose - it has provided an easy litmus test as to whether someone involved in working on acc layer stuff is being helpful or political. |
| 14:05 | <zcorpan> | jgraham: you should have said it on twitter instead |
| 14:05 | <SteveF_> | tantek: its all politics, i pick and choose who and which people/lists i interact with |
| 14:05 | <SteveF_> | tantek: to get stuff done |
| 14:05 | <tantek> | SteveF: "its all politics" <-- and that is why people simply walk away and go to WHATWG, or #indiewebcamp etc. |
| 14:06 | <tantek> | some of us don't believe in cultures of "its all politics" |
| 14:06 | <SteveF_> | tantek: a poltical statement in itself |
| 14:06 | <tantek> | SteveF - note, I didn't say *zero* politics, I am merely pointing out that any "all politics" cultures are basically deadends |
| 14:07 | <tantek> | and that implementers will shift to communities with more focus on building, and less focus on politics. |
| 14:08 | <SteveF_> | tantek: i work with moz acc engineers to get stuff done and to improve the acc implementation stuff defined and thats why you find people like Marco Zehe disagreeing with the dominant view |
| 14:08 | <Ms2ger> | tantek, seems like that might not be the case in a11y |
| 14:09 | <Ms2ger> | Anyway, if the people doing the a11y work want to do that at w3c, that seems fine with me |
| 14:09 | <Ms2ger> | We should definitely make sure that the whatwg spec matches implementations, though |
| 14:10 | <tantek> | Ms2ger - hence my original question of "where does specification of acc layer stuff happen that is *not* so political, and instead actually data-driven " - though I should is is not so *primarily* political. |
| 14:10 | <mathiasbynens> | annevk: no bandwidth to pick this up atm, but yeah – it now uses the latest draft as a ref |
| 14:10 | <tantek> | Ms2ger - agreed with "if the people doing the a11y work want to do that at w3c, that seems fine with me" - as long as it is features which actually *help* people and are practical for authors. |
| 14:10 | <mathiasbynens> | annevk: recently removed some things that have now progressed to the ES draft |
| 14:10 | <annevk> | mathiasbynens: can you maybe add a disclaimer somewhere? |
| 14:11 | <mathiasbynens> | but there are more |
| 14:11 | <jgraham> | It is very unclear to me that "create a specification -> change the specification in incompatible ways to address a11y issues -> try to backport the changes" is going to result in better a11y outcomes than "work together to create a specification with a11y designed in from the start" |
| 14:11 | <annevk> | mathiasbynens: yeah all the string stuff has moved iirc |
| 14:11 | <SteveF_> | "if the people doing the a11y work" i didn't say that, but i have much more in common with whatwg than others |
| 14:11 | <Ms2ger> | SteveF_, sorry, that's how I interpreted what you said about working with Mozilla engineers |
| 14:12 | <SteveF_> | Ms2ger: no what i meant is that i don't find w3c a pleasant place to work, just pleasenter than others |
| 14:12 | <Ms2ger> | I see |
| 14:13 | <Ms2ger> | Where by "others" you also mean whatwg? |
| 14:13 | <SteveF_> | and i don't have to spend cycles butting heads with hixie on stuff we can't reach agreement with |
| 14:13 | <darobin> | there are other others? |
| 14:13 | <darobin> | :) |
| 14:14 | <SteveF_> | also its where I can get feedback from acc engineers |
| 14:15 | <Ms2ger> | Well, the whatwg spec should eventually end up matching what browsers do, even if Hixie_ doesn't agree |
| 14:15 | <Ms2ger> | See: main / picture / style attribute / ... |
| 14:15 | <gsnedders> | SteveF_: I think my view is there's nothing wrong with people doing work in the W3C space, even though I think a lot of the policies around stuff like testsuites end up ultimately being detrimental |
| 14:17 | <SteveF_> | gsnedders: sure work where you find it most productive, but we could save some time if we work together instead of in a constant state of siege |
| 14:18 | <Ms2ger> | SteveF_, I certainly hope nobody disagrees with that sentiment here :) |
| 14:18 | <gsnedders> | Ms2ger: FORMAL OBJECTION. Arguments make for more interesting Twitter discussions! |
| 14:19 | Ms2ger | kicks gsnedders |
| 14:19 | <SteveF_> | Ms2ger: the elephant in the room for HTML is that no spec scratches the surface on acc layer implementation, i have been trying to work on that for a while only now is there some traction occuring. |
| 14:19 | <Ms2ger> | SteveF_, nobody except gsnedders here* ;) |
| 14:20 | <Ms2ger> | SteveF_, that's good to hear |
| 14:21 | <jgraham> | Yeah improving the a11y layer implementation seems like good work |
| 14:21 | <gsnedders> | (I'm not actually qualified to speak about anything recently, I've scarcely read whatwg or public-html in several years.) |
| 14:21 | <gsnedders> | (and I think making statements from how things were years ago probably isn't helpful) |
| 14:22 | <SteveF_> | FYI this is the re-start of html specific doc http://rawgit.com/w3c/aria/master/html-aam/html-aam.html |
| 14:22 | <SteveF_> | there is also one being developed for svg and a core/aria spec |
| 14:23 | <Ms2ger> | Looks interesting |
| 14:23 | <jgraham> | SteveF_: shouldn't that document start "this document *defines*", not "describes"? |
| 14:23 | <Ms2ger> | Not that I have expertise to say anything more |
| 14:24 | <SteveF_> | jgraham: probably, feedback welcome! as of now its just a copy of the informative doc i worked on in the past have to get down to editing it more |
| 14:26 | <SteveF_> | we are still working out the organization of stuff between the docs |
| 14:28 | SteveF_ | been too busy stoking the fires on twitter today to do anything productive |
| 14:30 | <SteveF_> | jgraham: btw would still like to get the acc tests into test repository, but unclear what hoops i need to jump to do so |
| 14:30 | <tantek> | never stoke the fires on Twitter |
| 14:30 | <SteveF_> | or whether they fit into the scheme |
| 14:30 | <gsnedders> | SteveF_: many of them need to be manual, no? afaik the manual test support is still a bit rough |
| 14:30 | <gsnedders> | SteveF_: they should totally be there |
| 14:31 | <SteveF_> | gsnedders: yes they need to be manual as they require using non automated tool to test |
| 14:31 | <gsnedders> | SteveF_: but they do need to be written at a level where a monkey can run them (or at least someone with no knowledge of a11y tools like say screenreaders) |
| 14:32 | <SteveF_> | gsnedders: the best i can do for that is to provide detailed instructions and examples |
| 14:33 | <SteveF_> | gsnedders: example of test http://stevefaulkner.github.io/html-mapping-tests/browser-tests/button.html |
| 14:35 | <SteveF_> | gsnedders: current instructions here http://stevefaulkner.github.io/html-mapping-tests/ think they need to be more detailed prescriptive http://stevefaulkner.github.io/html-mapping-tests/ |
| 14:35 | <gsnedders> | SteveF_: if you're around in a few hours, ping me then and I'll try and put together a list of what I think needs done |
| 14:35 | <jgraham> | SteveF_: If they are something that browsers can run they should be in there. It would be great if we can find a scheme to make them automatable |
| 14:35 | <jgraham> | (automated tests have thousands of times the value of non-automated tests) |
| 14:36 | <SteveF_> | gsendders: thanks! |
| 14:36 | <SteveF_> | jgraham: i agree and its a laborious task manually testing |
| 14:36 | <SteveF_> | jgraham: but am not aware of any automated method |
| 14:38 | <jgraham> | SteveF_: It seems like it should be quite possible — at least for that test — to expose the expectation information in some machine-parsable way. Then the actions required to run the test might be amenable to webdriver and getting the result out might be possible with platform-specific querying of the a11y API |
| 14:39 | <jgraham> | So I think with a bit of effort we can make it possible to run these tests in an automated way |
| 14:40 | <SteveF_> | jgraham: i know that you can get a dump of acc tree in chrome, but don't believe that you can do same in FF/IE, will ask moz acc |
| 14:41 | <jgraham> | SteveF_: Right, but I guess the platform has APIs for getting at it? I mean these are mappings to the actual platform feature, right? So even if you had to write a native program that queried the tree, it seems like it's possible |
| 14:42 | <annevk> | mathiasbynens: you could also ask on Twitter perhaps if there's someone willing to help out |
| 14:42 | <jgraham> | It isn't as cross-browser/cross-platform as I would like, but it's not impossible to imagine doing 90% of the work and saying to vendors "if you want to run these tests you ahve to do the last 10% yourselves" |
| 14:42 | <SteveF_> | >I mean these are mappings to the actual platform feature, right? yes, |
| 14:43 | <SteveF_> | features are mapped to platform apis |
| 14:43 | <SteveF_> | am talking with david bolter over on irc://moznet/accessibility |
| 14:48 | <gsnedders> | oh, jgraham has just said half of what I was going to say |
| 14:48 | <gsnedders> | Typical. |
| 14:49 | <annevk> | Was that the other half? |
| 14:49 | <gsnedders> | Nah, that was mostly about being more self-documenting |
| 14:49 | <gsnedders> | But that's for later when I'm not working |
| 14:50 | <wilhelm> | SteveF_: If that can be exposed to WebDriver (or even better: JavaScript) somehow, you're all good. |
| 14:51 | <gsnedders> | wilhelm: it can't be to JS, as it has to see over the whole rendered viewport, cross-origin |
| 14:52 | <SteveF_> | "davidb: we currently test our a11y core layer via automated tests but we don't open this up for web content (ripe security attack surface)." |
| 14:54 | <darobin> | that's fine |
| 14:55 | <wilhelm> | Then the difficult work is already done. Add a hook for WebDriver, and you've got your data. |
| 14:55 | <darobin> | I mean, you also don't expose everyone's browser using WebDriver to the open web :) |
| 14:55 | <jgraham> | Actually webdriver having access to the a11y tree seems independently good |
| 14:55 | <wilhelm> | Yes. |
| 14:55 | <darobin> | right |
| 14:56 | <darobin> | in fact I would be surprised if it didn't also enable things that aren't obviously a11y related |
| 14:56 | <jgraham> | It would allow people to write tests for their content being accessible with a much lower level of expertise than it currently requires |
| 14:57 | <wilhelm> | Yeah, you could do "iterate over all elements with an onclick handler, check if they have the appropriate roles set". |
| 14:58 | <darobin> | yeah, proper a11y linting would be a pretty damn good idea |
| 15:00 | <wilhelm> | The best way to do this, I think, is to work with a browser vendor on prototyping something. WebDriver is easily extended: https://dvcs.w3.org/hg/webdriver/raw-file/default/webdriver-spec.html#extending-the-protocol |
| 15:01 | <wilhelm> | Then we'll have implementation experience enough to put something in the spec. |
| 15:01 | <jgraham> | Hahaha |
| 15:01 | <wilhelm> | jgraham: Sssh! |
| 15:01 | <jgraham> | "webdriver is easily extended" |
| 15:01 | <jgraham> | So far no one actually implements that spec :| |
| 15:02 | <wilhelm> | Details! |
| 15:02 | <wilhelm> | (Says the chair of the WG.) |
| 15:02 | <wilhelm> | :P |
| 15:02 | <wilhelm> | jgraham: Many, uh, still extend it, though. |
| 15:03 | <jgraham> | Heh, well that's true |
| 15:03 | <caitp> | why do people believe that email addresses require top level domains in the domain part |
| 15:04 | <jgraham> | Certianly on my personal priority list "make an actual spec implementation" is higher than "figure out how to hook into the a11y tree and extend the protocol to support that" |
| 15:04 | <annevk> | darobin: hmm, forgot FormData depends on HTMLFormElement |
| 15:04 | <wilhelm> | And the spec is less "designing a new thing" and more "consolidating existing implementations". |
| 15:04 | <annevk> | darobin: we're doomed |
| 15:04 | <jgraham> | But hopefully someone else does both |
| 15:04 | <darobin> | annevk: it was just an example :) |
| 15:04 | <wilhelm> | jgraham: Different humans can work on different things. |
| 15:05 | <darobin> | annevk: but yeah, I know it's non-trivial, and am certainly not claiming that it is |
| 15:05 | <gsnedders> | caitp: because practically all those in the real world do |
| 15:05 | <jgraham> | wilhelm: I have asked around and apparently I can subcontract my work to as many Trafalgar Square pigeons as I like, but humans are out of the question |
| 15:06 | <gsnedders> | well I guess that means any crumbs are cleaned up quickly |
| 15:08 | <caitp> | i think the most commonly used email address in the world actually doesn't contain a TLD though :c |
| 15:08 | <caitp> | oh well |
| 15:08 | <gsnedders> | caitp: really? |
| 15:08 | <caitp> | yeah, like root@localhost |
| 15:09 | <wilhelm> | jgraham: The humans working on mapping HTML elements<->ARIA roles in browsers should also make their stuff testable. |
| 15:10 | <wilhelm> | Other humans can work on actually adhering to the WebDriver spec. |
| 15:10 | <gsnedders> | caitp: mhmm, maybe. but quite probably localhost is a special case in terms of those that see much use |
| 15:12 | <SteveF_> | willhem: note that ARIA roles are not generally mapped to HTML elements, only in cases where no platform role exists |
| 15:14 | <SteveF_> | willhelm: and then usually a string representation is exposed using the APIs standard methods |
| 15:16 | <jgraham> | SteveF_: I think the relevant question for WebDriver is "can you go from a reference to a DOM element to a a11y API subtree?" |
| 15:18 | <SteveF_> | jgraham: right, thats what the desktop tool we developed does |
| 15:19 | <SteveF_> | jgraham: http://www.paciellogroup.com/resources/aviewer/ (windows only) ;-( |
| 15:23 | <jgraham> | SteveF_: Right so Webdriver would presumably have an API like getAccessibilityTree(element) which would return some representation of that tree |
| 15:23 | <SteveF_> | jgraham: yes |
| 15:23 | <wilhelm> | Yes. |
| 16:28 | <Hixie_> | anyone understand https://www.w3.org/Bugs/Public/show_bug.cgi?id=26871 ? |
| 16:29 | <caitp> | > needs-more-info |
| 18:11 | <Hixie_> | you know what would be cool |
| 18:11 | <Hixie_> | a [Constructor] that returned a promise |
| 18:15 | <TabAtkins> | Hixie_: No, that violates what a constructor means. |
| 18:15 | <Hixie_> | i don't mean literally a [Constructor] |
| 18:15 | <Hixie_> | i mean something so there was a specific syntax to use like var f = async new Foo(); |
| 18:15 | <Hixie_> | rather than the current state of the art, factory methods |
| 18:16 | <annevk> | asyncnew |
| 18:18 | <Domenic> | Unclear why that is better than var f = foo(); |
| 18:18 | <Hixie_> | why is var f = new Foo(); better than var f = foo(); ? |
| 18:19 | <Domenic> | because you know that f will have Foo.prototype as its prototype. |
| 18:19 | <Hixie_> | i was going to go with "it's not", but ok |
| 18:19 | <Hixie_> | given your answer, an async new would be better than a factory method because "you know that the promise's value will have Foo.prototype as its prototype" |
| 18:20 | <Domenic> | hmm |
| 18:20 | <Hixie_> | but the real reason imho is because it more closely matches the programming mindset |
| 18:21 | <boogyman> | conventions change, what's "best practices" today, may not be so in the future. |
| 18:21 | <Hixie_> | there's a distinct conceptual difference between "Call a method" and "Call a constructor" which is still a real difference even if the result is delayed |
| 18:28 | <TabAtkins> | Agree that this is a bit of impedance mismatch from async getting grafted onto JS late. |
| 18:28 | <Hixie_> | foolip: ping https://www.w3.org/Bugs/Public/show_bug.cgi?id=25897 |
| 20:18 | <Domenic> | Using URLs as IDs, instead of as places to navigate to/download from, seems like a failure historically. (See: XML namespaces.) Do we have any good links on this topic I can point someone to? |
| 20:26 | <caitp> | yeah, or when a url is a resource in its own right, whose wacky idea was that |
| 20:27 | <TabAtkins> | caitp: I think I know what you're referring to, but can you clarify? |
| 20:28 | <caitp> | data-uris |
| 20:29 | <caitp> | but really I'm just making fun of the U(niform) in a pretty non-uniform system |
| 20:30 | <TabAtkins> | caitp: Oh, data-urls just point to a resource in the akashic record, which can be accessed intrinsically at all times. |
| 20:31 | <TabAtkins> | Or the platonic realm of knowledge, if your epistemology leans that way. |
| 20:31 | <TabAtkins> | There's just so much *stuff* in the Record that you can't give anything a short name, you've gotta be really specific about precisely which resource you're pointing at. |
| 21:03 | <annevk> | Domenic: there's https://wiki.whatwg.org/wiki/Namespace_confusion on namespaces |
| 21:05 | <Domenic> | waaaugh mixed content |
| 21:05 | <annevk> | Domenic: if you're not logged in there's some problem with cached content |
| 21:06 | <annevk> | Domenic: GPHemsley can fix it I think, or maybe provide you with access to the wiki so you can fix it if he doesn't have the time |
| 21:07 | <Domenic> | oh man we have official members? https://whatwg.org/charter |
| 21:07 | <Hixie_> | technically |
| 21:07 | <karlcow> | 06:06:51.920 Blocked loading mixed active content "http://wiki.whatwg.org/load.php?debug=false&lang=en&modules=ext.gadget.HotCat%2CReferenceTooltips%2CUTCLiveClock%2CWatchlistChangesBold%2Ccharinsert%2Cexlinks%2Cwidensearch%7Cmediawiki.legacy.commonPrint%2Cshared%7Cskins.vector&only=styles&skin=vector&*"[Learn More] Namespace_confusion |
| 21:07 | <karlcow> | 06:06:51.921 Blocked loading mixed active content "http://wiki.whatwg.org/load.php?debug=false&lang=en&modules=site&only=styles&skin=vector&*"[Learn More] Namespace_confusion |
| 21:07 | <karlcow> | 06:06:51.921 Blocked loading mixed active content "http://wiki.whatwg.org/load.php?debug=false&lang=en&modules=startup&only=scripts&skin=vector&*"[Learn More] Namespace_confusion |
| 21:07 | <karlcow> | 06:06:51.955 Blocked loading mixed active content "http://wiki.whatwg.org/load.php?debug=false&lang=en&modules=site&only=scripts&skin=vector&*"[Learn More] |
| 21:08 | <Hixie_> | Domenic: though that's about the only part of that page that's still accurate |
| 21:08 | <Hixie_> | karlcow: ping GPHemsley |
| 21:12 | <annevk> | oh wow, MediaWiki uses getPreventDefault() |
| 21:12 | <annevk> | there's a whole bunch of annoying warnings from MediaWiki |
| 21:12 | <annevk> | sad sad sad |
| 21:23 | <Hixie_> | getPreventDefault? |
| 21:28 | <Ms2ger> | Hixie_, an alias for defaultPrevented |
| 21:28 | <Hixie_> | nice |
| 21:40 | <caitp> | i assume it was added in gecko before defaultPrevented ever existed, but that's kind of a weird name for it :x |
| 22:44 | <TabAtkins> | Hixie_: Weird, http://software.hixie.ch/utilities/js/live-dom-viewer/?saved=3188 seems to transfer the overflow from inside the <iframe> to the outer page in Chrome. |
| 22:45 | <TabAtkins> | As does http://software.hixie.ch/utilities/js/live-dom-viewer/?saved=3185 |
| 22:45 | <Hixie_> | "transfer"? |
| 22:45 | <TabAtkins> | I assume this is a bug on Chrome's part. |
| 22:45 | <Hixie_> | oh wow |
| 22:45 | <TabAtkins> | The amount that the iframe document overflows the initial canvas, the outer document treats itself as overflowing as well. |
| 22:45 | <Hixie_> | i'm not seeing that |
| 22:45 | <Hixie_> | but i am seeing 0,0 not being at 0,0 |
| 22:46 | <Hixie_> | oh i see it's rtl |
| 22:46 | <TabAtkins> | You're seeing it in the top-left corner of the portion fo the iframe that's originally rendered (before scrolling), yes? |
| 22:46 | <TabAtkins> | Yeah. |
| 22:46 | <Hixie_> | what version are you in? i don't see what you describe |
| 22:46 | <TabAtkins> | Let's see... |
| 22:46 | <Hixie_> | i'm on 39.0.2150.5 dev |
| 22:46 | <Hixie_> | have a pending update |
| 22:46 | <Hixie_> | then again i also have a pending OS update |
| 22:47 | <TabAtkins> | 37 stable |
| 22:47 | <Hixie_> | maybe it's fixed? |
| 22:47 | <Hixie_> | try dev |
| 22:47 | <TabAtkins> | Maybe! |
| 22:47 | <TabAtkins> | I can't easily do that, since I'm on my pixel. ^_^ |
| 22:50 | <Hixie_> | set your device to dev mode :-) |
| 22:50 | <TabAtkins> | Nope, ChromeOS ships updates that break Crouton enough that I've had to stay on Stable permanently. |
| 22:50 | <TabAtkins> | (They usually fix it before it goes Stable.) |
| 22:51 | <TabAtkins> | I can't have my primary work laptop breaking on me every 6 weeks. :/ |
| 22:52 | <tantek> | two primary work laptops then? |
| 22:52 | <TabAtkins> | tantek: I have three, but only one is "primary". |
| 22:52 | <TabAtkins> | By definition. ^_^ |
| 22:52 | <Hixie_> | maybe use a different OS :-P |
| 22:52 | <TabAtkins> | Hixie_: Nah, I like ChromeOS. |
| 22:53 | <tantek> | TabAtkins - if they're all tethered to the same cloud identity, how is any of them primary? ;) |
| 22:53 | <TabAtkins> | Okay, preliminary twitter results suggest that 38 still has the issue. |
| 22:53 | <tantek> | Perhaps name them like borg? 1 of 3, 2 of 3, 3 of 3. |
| 22:54 | <TabAtkins> | tantek: Only one is actually usable as a work laptop - the other two are just for things I need Corp authentication for. ^_^ |
| 22:54 | <Hixie_> | TabAtkins: it's acting as i expect on 39, but i don't really know if i'm understanding it right. Want a screenshot? |
| 22:54 | <TabAtkins> | Sure. |
| 22:55 | <TabAtkins> | It's really just "is there a big horizontal scrollbar on the outer page?" |
| 22:56 | <Hixie_> | ah, yes, there's a scrollbar there |
| 22:56 | <Hixie_> | i was thinking you meant inside the iframe |
| 22:56 | <Hixie_> | that's crazy |
| 22:56 | <Hixie_> | i wonder what's up with that |
| 22:57 | <TabAtkins> | We're incorrectly calculating the size of the iframe when figuring out page scrollbars, assuming that its overflow is visible. |
| 22:58 | <TabAtkins> | And setting 'overflow' isn't doing anything, hm. |
| 22:58 | <TabAtkins> | Anyway, I'll reduce and report. |
| 23:07 | <TabAtkins> | Hixie_: Oh, it's your new "innerHTML View" pane that's causing it. |
| 23:08 | <Hixie_> | new? |
| 23:08 | <Hixie_> | that hasn't changed in, like, a decade |
| 23:08 | <Hixie_> | but yeah, looks like you're right |
| 23:08 | <Hixie_> | that makes me feel better :-) |
| 23:09 | <Hixie_> | still a bug, though |
| 23:09 | <Hixie_> | display:none should make the overflow go away |
| 23:09 | <Hixie_> | (though maybe i make it height:0 and not display:none?) |
| 23:09 | <TabAtkins> | Hixie_: Huh, I never even knew it existed, I guess. ^_^ |
| 23:09 | <Hixie_> | ah, it's just height:0 visibility:hidden |
| 23:09 | <TabAtkins> | Yes, you do height:0 for some reason |
| 23:09 | <TabAtkins> | That's a weird way to hide things, dude. |
| 23:10 | <Hixie_> | i think it's because in old browsers hiding the iframe with display:none would kill the browsing context |
| 23:10 | <Hixie_> | and i use the same styles for everything |
| 23:10 | <TabAtkins> | Ah. |
| 23:10 | <TabAtkins> | Well, fix please. ^^_ |
| 23:11 | <Hixie_> | i will add that to my todo list prioritised appropriately :-) |
| 23:12 | <TabAtkins> | All you gotta do is add a "pre.hidden { display: none; }" to the stylesheet! |
