23:24:13 RRSAgent has joined #aria 23:24:13 logging to https://www.w3.org/2019/09/16-aria-irc 23:24:18 Zakim has joined #aria 23:24:28 rrsagent, meeting spans midnight 23:24:38 rrsagent, make log world 23:24:51 Meeting: ARIA WG F2F TPAC - Day 2 23:33:29 kzms2 has joined #aria 23:43:22 MichaelC has joined #aria 23:51:11 jihye has joined #aria 00:02:05 we are in #pwg 00:03:28 zcorpan has joined #aria 00:04:39 aboxhall_ has joined #aria 00:05:38 https://global.gotomeeting.com/join/994278485 00:05:45 we are in #pwg 00:06:24 AmeliaBR has joined #aria 00:06:28 spectranaut has joined #aria 00:09:28 Jemma_ has joined #aria 00:28:42 we are in #pwg 00:41:51 Irfan has joined #aria 00:42:37 zcorpan has joined #aria 00:43:02 zcorpan has joined #aria 00:43:05 scribe: ZoeBijl 00:43:39 TOPIC: Repeated content 00:43:40 https://github.com/w3c/aria/issues/1044 00:44:02 Jemma has joined #aria 00:44:07 https://github.com/w3c/aria/issues/1044 00:45:42 mck__ has joined #aria 00:45:46 JD: James Craig how strong do you feel about linearised? 00:45:55 JC: *scribe fell behind* 00:46:02 Who is the primary client fo this CSS spec 00:46:07 mck__ has joined #aria 00:46:09 It felt like this was a similar thing 00:46:16 In medium/long form articles 00:46:19 …or books 00:46:27 The pullquotes which were he primary case 00:46:33 You might want to navigate to them 00:46:41 present+ 00:46:47 But read them in the linearised version 00:46:56 This is kinda like a CSS media type 00:47:14 If you tell your SR to “read all” you don’t want it to announce repeated content 00:47:34 JD: So repeated content are things that you’d want to be read in certain context 00:47:38 JC: yes 00:47:59 Like with VO if you do a two finger swipe down it’ll read the entire page 00:48:13 You might no want tthe repeated content to be read 00:48:29 My issue with the current proposal is that it might be too narrow a usecase 00:48:52 Perhaps we can apply this a broader use case 00:49:05 *James Nurthen makes a joke about aria-sometimes* 00:49:30 JD: So you’re saying… aria-linearised set to default? 00:49:47 The content author sets when the content is read. 00:51:42 But should it be the SR that says “hey this is repeated and my user said they don’t want that” 00:52:12 mck___ has joined #aria 00:52:13 ZB: Would something like an aside be another example of “not entirely relevant content” 00:52:18 *group is unsure* 00:52:28 JN: Redundant links 00:52:40 Anything that’s not repeated you want to read all the time right? 00:52:45 *group agrees* 00:52:58 Can we come up with anything that’s not repeated content 00:53:06 MK: I like the word redundant bettter 00:53:13 JD: You do? I think I do too. 00:53:41 JC: We should look at vocabulary 00:53:53 I think repeated content is a lot clearer and easier to understand than redundant 00:54:39 ZB: I agree, I think repeated is easier and more appropriate than redundant 00:55:16 Maybe we can make redundant a synonym ;) 00:55:29 MK: I’m not entirely sold on the use case period 00:55:44 If we’re trying to craft the end user experience 00:55:50 the screen reader experience 00:56:01 in ways that assume the SR user’s intent 00:56:18 you need to somehow, the screen reader, needs to somehow communicate their intent 00:56:29 JD: SR would have to implement an option for this 00:57:03 MK: JC you said something about the two finger swipe down 00:57:14 JC: if we know the authors inttent 00:57:16 present+ Joanmarie_Diggs 00:57:19 and it’s declaritive 00:57:19 present+ 00:57:28 q+ 00:57:34 MK: Having it be a SR option instead of a content option 00:57:40 Would make it a per user thing 00:57:55 I don’t think SRs need more options 00:58:40 ZB: could this be incorporated into a SR’s verbosity settings? 00:58:49 JC: We already have something for repeated labels 00:58:59 MK: That could be 00:59:10 But the difference between reading a text book 00:59:16 continues reading 00:59:23 versus not wanting to skip them 01:00:47 ZB: do you find it annoying to get repeated content read to you in articles 01:00:54 MK: That can be a bit confusing 01:01:23 +q 01:01:40 Just trying to figure out who is responsible for that 01:02:08 ack jamesn 01:02:25 MK: it feels like something that has good intent and could be useful 01:02:41 But before you put it in a spec it ought to be tested with real people in real situations 01:02:55 What would implementations look like? 01:03:16 JN: Isn’t that the way we work? 01:03:53 Freedom Scientific already said they’re interested in it 01:04:11 JC: Just want to emphasise that I don’t like redundant 01:04:18 JD: Can you comment on the issue? 01:04:22 ack Jemma 01:04:35 JK: we already have landmark roles 01:04:40 if we look at the landmark concept 01:04:42 jcraig has joined #aria 01:04:49 you can skip the navigation 01:05:02 JD: using pull quote as an example 01:05:15 jcraig has joined #aria 01:05:25 JD: you can mentally ignore it 01:05:37 Because it’s styled differenly 01:05:44 So visually it’s easier to ignore it 01:05:51 That’s a lot harder if you use a SR 01:05:55 Issue? 01:06:01 https://github.com/w3c/aria/issues/1044 01:06:29 *something about repeated headers* 01:06:37 Well he and others are debating that 01:06:42 JN: That’s not our debate 01:06:47 MK: It’s kinda interesting 01:06:54 Like 90% of the headers is repeated 01:06:59 But the page numbers aren’t 01:07:12 But I often like to hear page numbers when I’m reading a book 01:07:21 It makes you aware of the transitions 01:07:27 JD: dPub has a role for that 01:07:31 Perhaps we can use that 01:07:52 But for truly repeated content we could use aria-repeatedcontent 01:08:03 MK: I’m not too annoyed by repeated content 01:08:18 JN: Yeah but if you get a long pull quote it might be different 01:08:20 MK: Ia gree 01:08:28 s/Ia gree/I agree/ 01:09:39 JK: Are you saying that a pull quote can’t be a separate issue? 01:10:31 JD: Most users probably don’t want to be interrupted by all this content 01:10:32 Commented in the issue: After the F2F discussion, I withdraw the suggestion for the broader approach. I think “repeated content” or something similar is easier for authors to understand than the concept of linearized content for screen reader reading modes. 01:10:45 Unless you’re proofreading or something like that. 01:10:57 MK: It could be a nice feature to have 01:11:16 q? 01:14:52 mck___ has left #aria 01:16:27 mck has joined #aria 01:22:15 Matt_King_ has joined #aria 01:25:07 kurosawa has joined #aria 01:39:44 The minutes from yesterday don't seem to have set as scribe for "HTML Accessibility Issues" https://www.w3.org/2019/09/15-aria-minutes.html 01:39:44 Yesterday this was resolved by someone inserting a scribe historically via `i/Table Ontology/scribe: ZoeBijl` 01:39:44 So something like `i/HTML Accessibility Issues/scribe: tink/`should resolve that. 01:42:06 i/HTML Accessibility Issues/scribe: tink/ 01:42:15 RRSAgent, make minutes 01:42:15 I have made the request to generate https://www.w3.org/2019/09/16-aria-minutes.html ZoeBijl 01:43:37 RRSAgent, make yesterday’s minutes 01:43:37 I'm logging. I don't understand 'make yesterday’s minutes', ZoeBijl. Try /msg RRSAgent help 01:44:05 heh 01:46:20 ty 01:50:57 kurosawa has joined #aria 01:56:47 IanPouncey has joined #aria 01:58:38 https://github.com/w3c/csswg-drafts/issues/3708 01:59:27 spectranaut has joined #aria 02:03:02 zcorpan has joined #aria 02:08:30 Matt_King has joined #aria 02:08:43 we are now in #css 02:09:45 Matt_King_ has joined #aria 02:09:55 mhakkinen has joined #aria 02:18:26 MichaelC has joined #aria 02:20:12 Jemma_ has joined #ARIA 02:38:42 Matt_King_ has joined #aria 02:46:42 Matt_King has joined #aria 02:48:35 zcorpan has joined #aria 03:07:56 MichaelC has joined #aria 03:12:06 zcorpan has joined #aria 03:13:51 zcorpan has joined #aria 03:15:51 spectranaut_ has joined #aria 03:19:53 zcorpan has joined #aria 03:19:55 zcorpan has joined #aria 03:26:00 Zakim has left #aria 04:06:16 AutomatedTester has joined #aria 04:06:20 zcorpan has joined #aria 04:07:07 Irfan has joined #aria 04:08:21 present+ 04:08:25 present+ 04:08:34 RRSAgent, make minutes 04:08:34 I have made the request to generate https://www.w3.org/2019/09/16-aria-minutes.html ZoeBijl 04:09:50 dot-miniscule has joined #aria 04:10:04 https://bocoup.github.io/presentation-aria-and-webdriver/#/ 04:10:17 https://bocoup.github.io/presentation-aria-and-webdriver/#/ 04:10:20 mhakkinen has joined #aria 04:10:29 Boaz has joined #ARIA 04:10:29 bwald_ has joined #aria 04:10:30 scribe: ZoeBijl 04:10:32 present+ Boaz Sender, Bocoup 04:10:38 link to slides: https://bocoup.github.io/presentation-aria-and-webdriver/#/ 04:10:56 Present+ David Burns, Mozilla 04:11:03 Matt_King has joined #aria 04:13:21 Note: this meeting will not be minuted as the script for the talk is in the speaker notes. Potential discussion after the presentation will be minuted. 04:13:41 kurosawa has joined #aria 04:17:13 MichaelC has joined #aria 04:24:03 Example of pushButton documentation: https://bocoup.github.io/aria-practices/aria-practices.html#automation-pushbutton 04:25:16 bwald_ has joined #aria 04:29:05 *JC pointed out that close attention should be paid to how the role of an element is determined* 04:29:32 CharlesL has joined #aria 04:30:12 Related to the “inferring the role” part of the proposed documentation linked to earlier ⤴️ 04:31:58 *start of discussion/feedback* 04:32:01 Val: again there are multiple ways to check something like a button’s label 04:32:17 Do you think the guidelines are stable enough to use it for this? 04:32:37 Is the time now to do this? 04:32:42 Avneesh has joined #aria 04:32:50 So we’re testing ideas that are in the APG 04:32:58 q? 04:33:00 Those are the topics and questions for now 04:33:10 SP: This is webdriver extensions 04:33:24 In that sense it makes sense to document them in the webdriver spec 04:33:36 bwald_ has joined #aria 04:33:57 AutomatedTester: the webdriver spec tries to give us primitives to allow people to automate 04:34:05 with WD we think we have three audiences 04:34:08 web qa person 04:34:10 spec authpors 04:34:18 and people that want to write automation for something 04:34:22 Like a webcrawler 04:34:32 We try to cater to these three audiences 04:34:35 Zakim has joined #aria 04:34:41 and we try to make the primitives as low as possib;le 04:34:48 I’m not against this being in the WD spec 04:34:55 But perhaps it’s a new primitive 04:34:59 It’s not like a core… 04:35:07 Not how e historically thought about this 04:35:10 q+ 04:35:12 It’s not a subset of people 04:35:25 that’s why I wasn’t sure if it fits in the WD spec 04:35:29 q+ 04:35:35 That’s where my initial gut feel came from 04:35:48 average user, it’s kind of like, fitting those three groups of people 04:36:01 The spectrum is incredible broad 04:36:04 How does that fit in 04:36:08 With push button as an example 04:36:20 No one has come to us to ask “how do I push a button” 04:36:59 ZB: I think it would be good to automate this, take away people having to think about accessibility 04:37:03 q+ 04:37:03 Even if no one asked for that 04:37:07 q- 04:37:09 AutomatedTester: absolutely 04:37:29 And then you would go and try and do a click or keyboard interaction 04:37:40 And at that point it uses the accessibility tree rather than DOM commands 04:37:48 So what Val and Simon were saying 04:37:56 You get these stale ???? 04:38:00 If you did find it 04:38:05 and you try to interact with it 04:38:16 you throw an error saying “hey this isn’t in the a11y tree” 04:38:33 JJ: I was thinking in the case of the examples 04:38:42 the dev would have alrteady have given the role to the element 04:38:55 I think it would be more interesting to tell them ?? 04:38:58 zcorpan has joined #aria 04:39:16 Would rather have it say click this role via the accessibility tree 04:39:28 Val: I think that’s the intention 04:39:30 q- 04:39:36 ack jc 04:39:43 JC: Pitched this idea years ago 04:39:53 Probably in some other repo 04:40:05 Some of the primitives that this could pile onto is element.computedRole 04:40:16 Not necessarily go through the script but ask the engine 04:40:22 You can get the role and label from there 04:40:46 You can find elements by computed roles 04:41:10 AutomatedTester: can we get this via JavaScript? 04:41:12 JC: No 04:41:36 element.computedRole is experimental and buggy 04:41:48 It’s going through the DOM tree, but that doesn’t necesarily mean that the browsers are doing the right thing, you can’t check that through the DOM. 04:41:59 JN: I believe you can get it in puppeteer now 04:42:33 JC: Getting access to the entire accessibility tree is going to take a long time 04:42:42 Getting access in webdriver would be trivial tho 04:42:59 SP: Why wouldn’t we give access to all developers 04:43:17 JC: That’s a good question, right now there’s a significant performance hit 04:43:48 SP: Is it heavy to request the role of an element? 04:43:56 sarah_higley has joined #aria 04:44:49 JC: Not necessarily but there are complications 04:45:00 ZB: Would it be heavy to get all elements of a certain role? 04:45:33 JC: Yes. I suggested a :role selector for CSS years ago. But when we tried to implement it with the CSS WG we found that it was too heavy. 04:46:43 Select by label (similar to what’s on slide #8) is something we use a lot to find things 04:46:46 q? 04:46:50 This works through the accessibility tree 04:46:56 So I hink this is a great idea 04:47:10 I think this should be closer to the APG (?) 04:47:44 AutomatedTester: whereever it lives doesn’t stop it from being implemented 04:48:02 ZB: As long as it’s not he APG, because that’s a note not a spec 04:48:23 AutomatedTester: webdriver puts lots of effort into extensibility into its spec 04:48:40 q? 04:48:54 I think it fits better in WebDriver 04:49:20 Boaz: ???? 04:49:36 I don’t think this needs to be in the WD spec 04:49:56 AutomatedTester: the other thing I tried to advance is 04:50:01 this is very input driven testing 04:50:08 ? 04:50:13 which is one of the easier sides of testing accessibility 04:50:17 boaz: I think the idea here is to change web developer's mindsets 04:50:18 ack Boaz 04:50:43 s/????/I think the idea here is to change web developer's mindsets/ 04:51:51 CharlesL has joined #aria 04:51:55 CharlesL has left #aria 04:52:07 CharlesL has joined #aria 04:52:18 present+ 04:52:25 zcorpan has joined #aria 04:52:26 present+ 04:52:45 TOPIC: aria-details 04:53:13 https://github.com/w3c/aria/issues/1001 04:53:19 https://github.com/w3c/aria/issues/1001#issuecomment-521076833 04:53:31 https://github.com/w3c/aria/issues/1009 04:54:31 scribe: spectranaut 04:54:44 rrsagent, make minutes 04:54:45 I have made the request to generate https://www.w3.org/2019/09/16-aria-minutes.html Jemma_ 04:55:06 From the spec: 04:55:15 "In some user agents, multiple reference relationships for descriptive information are not supported by the accessibility API. In such cases, if both aria-describedby 04:55:16 and aria-details are provided on an element, aria-details takes precedence." 04:55:27 https://github.com/w3c/aria/issues/1009 04:55:35 present+ 04:55:40 present+ 04:55:45 present+ Joanmarie_Diggs 04:55:45 present+ 04:55:54 present+ 04:55:57 present+ 04:56:50 mk: (describes jongunds issue) the way we spec'd aria-details, it is not meant to describe an accessible description 04:57:05 Joshue108 has joined #aria 04:57:20 the challenge is that we don't know what an accessible description is, and what the difference of intent is between aria-details 04:57:36 zcorpan_ has joined #aria 04:58:09 the content in aria-details should be navigatable to, but accessible description does not have to exist in the dom because it could be a hidden element 04:58:48 q+ 04:58:58 the main thing we would like to accomplish is: what is the content referenced by aria-details in particular? what is the AT expectations? 04:59:30 if you use both, you maybe overwritting the accessible description? 04:59:38 so are the details the description? 05:00:00 ack cl 05:00:12 present+ 05:00:14 ack CharlesL 05:00:16 cl: we looking to use aria-details int he publisher context for enhanced image descriptions 05:00:28 q+ 05:00:42 say we have an image that is complex, like a table (heaven forbid) we would like to use aria-details to put a table in aria-details 05:01:06 or if we had an image of a math equation 05:01:26 can we use aria-details in this way? 05:01:45 when a screen reader hits an element with aria-details, it says, "has details" 05:01:49 q+ 05:01:55 q+ 05:01:58 we would like it to be a linkable, clickable, navigatable section 05:02:12 have a mechanism to the go back where they were in reading 05:02:58 mk: question about images. is there a reason the image wouldn't be in a figure and all of that details content would be in a fig-caption, directly associate with the image? 05:03:40 cl: maybe a publisher wants all of those description sin an appendix, instead of encapsulated in a fig caption with the image 05:04:02 mh: I share charles's interest. We generate a lot of content with complex images. 05:04:03 ack mh 05:04:16 we need to provide a link to image and structure decision 05:04:36 s/decision/description/ 05:05:29 without aria-details, we could use a figure with text descriptio. structure info about a image and flatten it into aria-describeby is bad 05:05:39 q+ JamesC 05:06:06 I'm not thrilled with aria-details. does the screen reader need to tell you how to get to the structure content. what if that content has another link. 05:06:10 ack me 05:06:31 jn: one good thing about aria-details is that we are about to get more implementaiton sof it because of aria annotations work 05:06:44 hopefully screen readers will add a way to navigate to the annotation 05:07:01 bwald_ has joined #aria 05:07:28 https://github.com/daisy/epub-accessibility-tests/tree/master/content/epub30-test-0340/ 05:07:29 ack av 05:07:36 q+ To state we need to do something about the mappings. 05:07:51 av epub use cases link link above 05:08:17 aria-details and aria-describeby is huge. aria-describeby truncates tables and does not provide good descriptions 05:08:47 people do not want to put descriptions by image by at the end of the page 05:09:10 aria-details is not announced by screen reader, but we need a way to move between the image to where the description is and back again 05:09:29 ack jc 05:09:33 zcorpan has joined #aria 05:09:35 bwald_ has left #aria 05:09:37 ack jamesC 05:09:50 q? 05:09:56 q+ 05:11:07 jc: annotations. aria-details reminds us of longdesc. but there are benefits -- same page. 05:11:34 ack me 05:11:35 joanie, you wanted to state we need to do something about the mappings. 05:11:39 BUT aria-details actually has use cases than longdesc 05:11:57 https://w3c.github.io/core-aam/#details-id-105 05:12:00 s/than/unlike 05:12:02 s/than/unlike/ 05:13:09 jd: if there are details and describe-by is due to UIA -- if they have one description. 05:13:34 if microsoft can add something to their api to differentiate between aria-details and aria-describeby 05:13:48 what happens if both appear and we can only expose one of them? 05:14:12 UIA is microsoft's user interface automation 05:14:38 q? 05:14:48 ack Ma 05:15:45 q+ 05:15:53 mk: support for aria-details. no one is suggesting to get rid of it. Microsoft wasn't read to support it but it went into the spec anyway. if we can get rid of the conflict that only exist in UIA then we could make aria-details fully functional on all platforms 05:16:25 if in a year from now we have aria-details support. from an authoring perspective, it is not the accessible description. 05:16:42 Identifies the element that provides a detailed, extended description for the object. See related aria-describedby. 05:17:17 The aria-details attribute references a single element that provides more detailed information than would normally be provided by aria-describedby. 05:18:07 mk: "accessible name and description calculation" -- aria-details has nothing to do with that 05:18:20 jd: confirms that is in the spec 05:18:24 Unlike elements referenced by aria-describedby, the element referenced by aria-details is not used in either the Accessible Name Computation or the Accessible Description Computation as defined in the Accessible Name and Description specification [ACCNAME-1.1]. 05:18:30 jc: annotations are not descriptions 05:18:50 mk: we need to find another way to communicate this in the spec not using description 05:19:15 q+ To ask about the "alternatively... link to web page" bit 05:19:20 q- 05:19:21 what will the authoring practice say? if annotations go forward, we can include information about that 05:20:13 q? 05:20:13 ack me 05:20:15 joanie, you wanted to ask about the "alternatively... link to web page" bit 05:20:24 Alternatively, aria-details may refer to a link to a web page having the extended description, as shown in the following example. 05:20:36 jd: something else the spec says that we should get rid of (above) 05:20:58 q+ 05:21:07 unofficial decision: get rid of reference to "Extended description" 05:21:08 q+ 05:21:17 ack av 05:21:30 av: some people want to put accessible descriptions at the end of the book 05:21:53 they say the authors have control of content on the pages 05:21:58 different webpage 05:22:10 each chapter is an html page 05:22:16 last page has descriptions 05:22:34 present+ Benjamin_Young 05:23:24 av: we want aria-details to be able to link to a different webpage 05:23:43 aria-details links to a link that will link to a different web page 05:24:38 mk: we will have aria-details that links to a link. the language of the spec (above) implies the user should be able to bypass that directly -- if aria-detail points to an anchor, then the use will not navigate to the location of the anchor tag but instead to the target 05:25:32