15:06:10 <API> #startmeeting
15:06:10 <Services> Meeting started Thu Nov  6 15:06:10 2014 UTC.  The chair is API. Information about MeetBot at http://wiki.debian.org/MeetBot.
15:06:10 <Services> Useful Commands: #action #agreed #help #info #idea #link #topic.
15:06:14 <API> #topic Progress towards 3.16
15:07:07 <API> #info recently we had the 3.15.1 release
15:07:22 <API> #info as far as I know, not a lot of new stuff, at least in relation with API additions
15:07:23 <API> done
15:07:41 <clown> API additions to a11y interfaces?
15:07:41 * joanie apologizes
15:07:44 <joanie> there will be more
15:08:00 <API> clown, yes I mean that
15:08:04 <clown> okay, thanks.
15:08:09 <API> anyway joanie started to shake bugzilla
15:08:19 <API> pinging old bugs and crate new ones
15:08:20 <joanie> #info Joanie started filing bugs for what is missing.
15:08:21 <API> create too
15:08:23 * clown feels sorry for bugzilla
15:08:57 <joanie> #info She has not yet done the actions-related stuff insofar as filing a bug with all of them.
15:09:16 <joanie> #info The ongoing question is interfaces versus the use of AtkAction.
15:09:41 <API> fwiw, not sure if worthy to info
15:09:51 <API> joanie and I had a small debate about that "versus"
15:09:52 <joanie> ok, I'll stop.
15:10:02 <API> oh sorry, I thought you stopped
15:10:05 <API> go on please
15:10:08 <joanie> also a discussion with Benjamin
15:10:10 * API waiting to joanie done
15:10:11 <joanie> nah
15:10:14 <joanie> done
15:10:18 <joanie> I always deep dive anyway
15:10:20 <joanie> and shouldn't
15:10:22 <joanie> please go on
15:10:32 <API> well, my summary non-worthy to info
15:10:32 * joanie fetches some coffee from the kitchen
15:10:44 <API> is that after your informat chat with Benjamin
15:10:49 <API> you informally debate with me
15:11:09 <API> and although the conclusion is that going on with as much API as possible, possiblity deprecating actions in the future
15:11:12 <API> the situation is tricky
15:11:27 <API> you used as example the key bindings of current AtkAction
15:11:30 <API> that is an feature you like
15:11:37 <API> well,
15:11:44 <API> s/you like/you don't want to lose
15:12:04 <clown> are we talking about :  https://developer.gnome.org/atk/stable/AtkAction.html
15:12:05 <clown> ?
15:12:17 <API> yep
15:12:23 <API> so to give some context
15:12:29 <clown> go on.
15:12:33 <API> there are some stuff that
15:12:41 <API> you could argue that fit on AtkAction
15:12:47 <API> but are exposed as API
15:12:55 <API> and the feeling is that
15:13:25 <API> "evident" stuff has their own API
15:13:32 <API> and AtkAction is there for the rest
15:13:38 <API> something similar to the old
15:13:52 <API> having a list of roles, and a extended roles procedure to add roles on the fly
15:14:01 <API> so the discussion was about trying to unify
15:14:06 <API> or having everything as actions
15:14:11 <API> or everything as API
15:14:27 <API> but as I mentioned, the conclusion is that is tricky
15:14:34 <API> and that there are several corner cases
15:14:36 <API> and done
15:14:43 <API> joanie, did I forget something?
15:14:47 <clown> it sounds like the "debate" between a11y apis like ATK vs UIA.
15:15:10 <joanie> clown: Cynthia and I chatted about it
15:15:16 <clown> makes sesnse.
15:15:18 <joanie> And there's also user intentions
15:15:34 <clown> yes, I saw a bunch of emails about that out of the W3C.
15:15:35 <joanie> so my temption at the moment is to wait and see what happens post-TPAC
15:15:43 <clown> I can't say I understand the issues there.
15:15:44 <API> user intentions?
15:15:58 <joanie> i'll talk about it a W3C
15:16:01 <API> ok
15:16:03 <joanie> at the topic
15:16:06 <clown> User intention events — what does the user intend by a certain, for example, mouse click.
15:16:15 <API> so, if no ones wants to add anything to this topic I will move on
15:16:32 <clown> they don't intend to click the mouse.  they intend something else (I'm not clear what though).
15:16:43 <clown> move on.
15:16:54 <API> yes, joanie and I discussed something similar on our private chat
15:16:55 <API> in any case
15:16:58 <API> #topic W3C updates
15:17:11 <joanie> heh
15:17:14 <clown> seems we started to do that already :-)
15:17:14 <joanie> so soon?
15:17:31 <API> joanie, yes thats what happens when nobody shares updates
15:17:34 <API> we move quickly
15:17:38 <clown> joanie, shall I go first?
15:17:42 <joanie> clown: sure
15:17:43 <clown> I don't have much.
15:18:19 <clown> #info Joseph put out a relatively complete, but still rough, editors' draft of the name and description computation.
15:18:32 <clown> #info It is here:  http://rawgit.com/w3c/aria/master/accname-aam/accname-aam.html
15:18:54 <clown> #info the guts of the algorithm is in this section:  http://rawgit.com/w3c/aria/master/accname-aam/accname-aam.html#mapping_additional_nd_te
15:19:17 <clown> #info there is also section on mapping of name and description:  http://rawgit.com/w3c/aria/master/accname-aam/accname-aam.html#accessible-name-and-description-mapping
15:19:34 <clown> #info and, a section on name/desription change events:  http://rawgit.com/w3c/aria/master/accname-aam/accname-aam.html#events
15:20:02 <clown> #info Because of the intense TPAC meetings, most W3C meetings were cancelled this week.
15:20:10 <clown> done. questions?
15:20:39 <API> so is a kind of hangover?
15:21:21 <API> in any case, you can ignore my question
15:21:22 <clown> ask joanie :-)
15:21:32 <API> no real questions from my side
15:23:02 <joanie> * Joanie spent the last week at the W3C's TPAC. Lots of meetings, hallway conversations, etc.
15:23:04 <joanie> * Amongst the most notable topics are user intentions and the possibility of a role explostion.
15:23:07 <joanie> * User intentions, which we were talking about previous topic, are (as Joseph was stating) a reflection of what the user wants (is intending) to accomplish. For instance, not "click" but "play music"; not "shift+right" but "select the next character".
15:23:11 <joanie> (not done)
15:23:48 <joanie> * A lot of the interest around user intentions is coming out of the Editing efforts / folks.
15:24:23 <joanie> * Editing code being more than "editing"; it involves any interaction with text, including non-editable text.
15:24:27 <joanie> anyhoo....
15:24:55 <joanie> * There is similar intentiony stuff being done in the IndieUI working group (which Joanie has just joined)
15:25:31 <joanie> * There is also a document related to all of this being done by someone else (accessibility person) at Microsoft. (Cynhtia Shelly)
15:25:38 <joanie> anyhoo again....
15:26:14 <joanie> * During TPAC, one of the main guys (if not the main guy) behind the editing user intentions stuff met with us (Accessibility folks).
15:26:20 <joanie> oh hell
15:26:24 <joanie> I'm not infoing
15:26:35 <joanie> side effect of irc standup meeting
15:26:38 <joanie> one sec
15:27:55 <joanie> #info Joanie spent the last week at the W3C's TPAC. Lots of meetings, hallway conversations, etc.
15:27:58 <joanie> #info Amongst the most notable topics are user intentions and the possibility of a role explostion.
15:28:01 <joanie> #info User intentions, which we were talking about previous topic, are (as Joseph was stating) a reflection of what the user wants (is intending) to accomplish. For instance, not "click" but "play music"; not "shift+right" but "select the next character".
15:28:05 <joanie> #info A lot of the interest around user intentions is coming out of the Editing efforts / folks.
15:28:08 <joanie> #info Editing code being more than "editing"; it involves any interaction with text, including non-editable text.
15:28:11 <joanie> #info There is similar intentiony stuff being done in the IndieUI working group (which Joanie has just joined)
15:28:14 <joanie> #info There is also a document related to all of this being done by someone else (accessibility person) at Microsoft. (Cynthia Shelly)
15:28:17 <joanie> my bad
15:28:19 <joanie> to continue
15:28:37 <joanie> #info During TPAC, the (editing) guy leading the effort on user intentions talked to us (accessibility) folks on several occasions.
15:29:05 <joanie> #info I gave an explanation to him (and others) about why this stuff should be bigger than editing.
15:29:34 <joanie> #info Example: A user in Twitter moves to the next tweet in the timeline by pressing J
15:30:06 <joanie> #info What ATs get as a result is even spewage based on what happened to the DOM (and I guess render tree as well).
15:30:59 <joanie> #info What ATs do NOT get is something like "object:focus-next-item". As a result, they play many sad, heuristic, guessing games trying to decide what to present, what to ignore, etc.
15:31:29 <joanie> #info Happily, there seems to be some general agreement that, yes, user intentions are awesome for accessibility :)
15:32:18 <joanie> #info And Ben (Microsoft editing guy) asked me to file some issues around this against his Editing-specific docs in progress so that we can take them into account *somewhere* (to be determined).
15:32:45 <joanie> #info That somewhere may be IndieUI or not. <insert shrug here>
15:33:26 <joanie> #info But as you can (hopefully) see from my deep dive above, intentions ARE actions -- if we do actions better, expanding them beyond device-specific stuff.
15:33:54 <joanie> #info And some are already doing that sort of expansion. Example: Mozilla did an AtkAction "showlongdesc".
15:34:26 <joanie> #info The other, item of interest -- which I won't deep dive on (really) is the possibility of a role explosion^W expansion.
15:34:59 <joanie> #info We met with the digital publishing group and their roles are going to be written up by them for our (accessibility) consideration.
15:35:21 <joanie> #info But the feeling seems to be that valid ones will be included and mapped like other roles.
15:35:53 <joanie> #info Joanie thinks we (ARIA) should consider subroles. But after suggesting that and discussing it in depth at TPAC, the impression she gets is that won't happen.
15:36:31 <joanie> #info Having said that, ATK/AT-SPI2 can still have subroles (like Apple's AX API already does). So an ARIA-based role explosion doesn't have to hurt us. :)
15:36:56 <joanie> #info Lastly, related to this, we discussed (again) localized role names and the possibility of localized state names.
15:37:50 <joanie> #info Two examples of the latter: An on/off "switch" is a toggle button (pressed/not pressed) or two-state checkbox (checked/not checked).
15:38:11 <joanie> #info But what sighted users see and what end users probably want to hear is "on/off".
15:38:26 <joanie> #info If and how to implement this is still under discussion.
15:38:49 <joanie> #info The other example btw is testing related. But I guess that doesn't need to be mentioned here.
15:38:59 <joanie> I guess that's more than enough
15:39:05 * joanie smiles sadly.
15:39:09 <joanie> anyhoo, questions?
15:39:11 * API raises hand for questions
15:39:15 <clown> lots to digest.
15:39:25 <joanie> sorry
15:39:32 <clown> not your fault.
15:39:34 <joanie> that wasn't my plan
15:39:36 <joanie> it just came out
15:39:42 <API> clown, I assume that "lots to digest" can be translated to "you first"
15:39:44 <joanie> slowly too because my fingers are tired
15:39:46 <API> so...
15:39:54 <clown> it's a good summary.
15:40:11 <API> so the conclusion of this "role explosion" and "won't happen" in relation to subroles
15:40:15 <API> means
15:40:22 <API> that they want to add a lot of roles,
15:40:25 <API> but at first
15:40:32 <API> it will not be categorized at all?
15:40:34 <joanie> they *may* want to. We don't have the list yet.
15:40:41 <joanie> it will be categorized
15:40:50 <joanie> that is part of the work dpub agreed to do
15:41:04 <joanie> not just a simple list, but details about where it fits into everything else
15:41:21 <clown> I expect the dpub has a number of roles in mind that are not in aria now.  Roles for books, like "chapter"
15:41:27 * joanie nods
15:41:38 <API> so the categories will be "implicit" by documentation but not explicit as a different api for roles and subroles, right?
15:42:00 <joanie> not sure I get your question
15:42:22 <joanie> my understanding is that dpub will give us (Protocols and Formats and/or ARIA) a proposed list of roles with details.
15:42:34 <joanie> Which we (PF and/or ARIA) will consider and discuss
15:42:59 <joanie> But those we accept (e.g. chapter) will likely (though not definitely) be full-fledged roles like other ARIA roles.
15:43:21 <API> yeah but what I mean is that it will be still aria-role="dpub-role1", aria-role="dpub-role2"
15:43:26 <joanie> Once that is the case, then we (led by clown) map them to our individual APIs
15:43:29 <API> and some documentation explaining that those roles
15:43:36 <API> would be used here and here
15:44:00 <API> but right now the plan is not doing a aria-role="dpub" aria-subrole="new-role1"
15:44:00 <joanie> so a chapter in ATK might be ATK_ROLE_SECTION and ATK_SUBROLE_CHAPTER
15:44:08 <joanie> API: EXACTLY
15:44:09 <clown> I'm the lead?
15:44:19 <API> well, technically we could do the same on ATK
15:44:34 <joanie> clown: of UAIG or whatever AAM
15:44:34 <API> but I feel that it would be messy
15:44:48 <API> anyway, not sure why, I thought that aria was in favor of (eventually) adding subroles
15:44:59 <joanie> API what I was suggesting is aria-role (which is really just role) = <something we already have in ARIA>
15:45:08 <API> probably was because somehow, the concept of abstract roles is simiar to that
15:45:08 <joanie> and then a new aria-subrole="chapter"
15:45:22 <joanie> API: what would be messy?
15:45:40 <joanie> argh html a11y concall coming up like soon
15:45:51 <joanie> so I need to refrain from debating
15:45:53 <API> joanie, taking the current list of roles, and just add documentation to categorize it, instead of adding real categories
15:46:08 <API> ok, I have finished with this question
15:46:13 <API> thanks for the explanation
15:46:15 <joanie> API meaning there would not be any new role "chapter"
15:46:17 <joanie> but instead
15:46:18 <clown> joanie, something like: <div role="region" aria-subrole="chapter" >?
15:46:25 <API> but I have one more
15:46:28 <joanie> some doc would say chapter is an example section?
15:46:34 * API waiting, clown is leading now
15:46:39 <joanie> clown: that is what I was proposing, yes
15:46:46 <joanie> got a big ol' thumbs down ;)
15:46:48 * joanie shuts up
15:46:52 <clown> gotcha.  looks ok at first blush.
15:47:05 <clown> to me at least.
15:47:16 <joanie> because you are wise and brilliant and all that :P
15:47:17 <clown> joanie, one thing I just tried with twitter and J.
15:47:32 <clown> I am seeing a DOM focus event on the tweet that, well, gets focus.
15:47:40 <joanie> yup
15:47:58 <clown> so, I'm baffled why FF is not putting out an atspi focus event as well.
15:48:01 <clown> done.
15:48:04 <joanie> it does
15:48:08 <joanie> and some children changed
15:48:13 <joanie> and some text inserted
15:48:23 <clown> really?  yeesh.
15:48:32 <joanie> look in the indie ui minutes
15:48:39 <clown> okay.
15:48:40 <joanie> or I'll give you the link/doc later
15:48:49 <joanie> but I actually captured them as an example for ben
15:48:49 <clown> that's okay with me.
15:49:02 <joanie> oh, there are also two name-changed events
15:49:20 <joanie> it's event salad.
15:49:22 * clown wonders what the name of a tweet is (or should be).
15:49:38 * clown perhaps the tweet-er.
15:49:42 <joanie> I forgot to see what all was emitting the name too
15:49:51 <joanie> but I don't think it's really anything named
15:50:06 <joanie> anyhoo, other stuff?
15:50:23 <clown> sure.
15:50:24 <API> just a quick one
15:50:28 <API> <joanie> #info And some are already doing that sort of expansion. Example: Mozilla did an AtkAction "showlongdesc".
15:50:33 <API> not sure how to fell about that
15:50:47 <joanie> API mixed :)
15:50:57 <joanie> or ATK_STATE_INDETERMINATE :P
15:50:59 <API> do you think that is the way to go to implement showlongdesc?
15:51:10 <joanie> under the circumstances, yes
15:51:16 <API> ok, in any case
15:51:17 <joanie> and we officially blessed it ;)
15:51:32 <joanie> but it's... complicated
15:51:34 <API> I feel that this is another case of using AtkAction to implement "other stuff without API"
15:51:40 <joanie> in part
15:51:54 <API> Im like a saint, I bless everyone, doesn't matter if they are evil
15:52:04 <API> in any case
15:52:05 <joanie> :)
15:52:08 <API> that is answer enough
15:52:10 <clown> how does ATK implement a link?  As an AtkAction?
15:52:13 <API> at least for this meeting
15:52:18 <joanie> clown: no
15:52:21 <API> AtkHyperlink
15:52:25 <API> AtkHypertext
15:52:32 <API> and the evel AtkHyperlinkImpl
15:52:34 <API> *evil
15:52:35 <clown> 'cause a longdesc is a link.
15:52:47 <joanie> clown: but in this case the action doesn't do that
15:52:54 <joanie> well
15:52:58 <joanie> it sorta does
15:53:02 <clown> "do that" == ?
15:53:25 <joanie> if I click on a link I get linky behavior
15:53:28 <joanie> :)
15:54:04 <clown> if you choose "Show londesc" (or whatever it's called) from the context menu, you get, I think, linky behaviour.
15:54:14 <joanie> yeah....
15:54:41 <joanie> but how links and longdescs are handled might differ in a given user agent
15:54:53 <joanie> and how does a screen reader know to tell the user there's a longdesc
15:55:26 <joanie> and how does the user know if a link is a link to new content, or a description of the current content?
15:55:27 <clown> well, what info is in the accessible for the img with the longdesc?
15:55:42 <joanie> all the image stuff
15:55:53 <joanie> and it implements the AtkImage interface
15:55:59 <clown> I would think from the name/description fields in the AtkAction.  Maybe.  Just speculation.
15:56:06 <joanie> ?
15:56:20 <joanie> oh, but that's the point (sorta)
15:56:23 <clown> https://developer.gnome.org/atk/stable/AtkAction.html#atk-action-get-name
15:56:26 <joanie> links are NOT done via AtkAction
15:56:37 <joanie> if they were done in the same way as links
15:56:43 <joanie> they would not have an action name
15:56:53 <joanie> oh, related:
15:56:54 <clown> oh, never mind.  the docs there say the name returns things like "click".
15:57:04 <joanie> when a user who is blind is told "foo link"
15:57:15 <joanie> they assume that moving to foo and pressing Enter will activate the link
15:57:19 <joanie> that doesn't work with longdesc
15:57:49 <joanie> something special is required of the user and the user agent and or the AT
15:57:59 <clown> what if longdesc was handled like a link?
15:58:05 <joanie> handled?
15:58:21 * joanie grabs headset for next call
15:58:23 <API> well, the meeting time is almost finishing so, could we end the debate here, and continue at another point at the ml, IRC channel or next meeting?
15:58:30 <joanie> yes please
15:58:32 <clown> put another way:  what if longdesc was not mapped to AtkAction, but to whatever links are mapped to?
15:58:32 * joanie waves
15:58:39 <clown> bye.
15:58:39 <joanie> clown: it would break
15:58:41 <joanie> ;)
15:58:44 <API> #topic misc time (even marketing!)
15:58:45 <joanie> or not work
15:58:46 <clown> say hi to rich for me.
15:58:56 <joanie> clown: I think rich is on vacay
15:59:07 <API> so anyone (jjmarin?) wants to add something quick before ending the meeting?
16:00:05 <API> nobody says anything
16:00:05 <API> so
16:00:26 <API> #info bug 730505 was reviewed by mgorse and API
16:00:37 <Services> 04Bug http://bugzilla.gnome.org/show_bug.cgi?id=730505 normal, Normal, ---, at-spi-maint, UNCONFIRMED, Draft for atspi tests
16:00:37 <API> #info patryk already added new patches
16:00:42 <API> #info they would be reviewed when possible
16:00:56 <API> and having said so, I will close the meeting unless someone complains ....
16:01:38 <API> nobody is complaining
16:01:42 <API> so closing the meeting
16:01:46 <API> thanks everybody for coming
16:01:51 <API> see you on #a11y
16:01:53 <API> #endmeeting