15:07:04 <API> #startmeeting
15:07:04 <Services> Meeting started Thu Nov 13 15:07:04 2014 UTC.  The chair is API. Information about MeetBot at http://wiki.debian.org/MeetBot.
15:07:04 <Services> Useful Commands: #action #agreed #help #info #idea #link #topic.
15:07:16 <API> #topic Progress towards 3.16
15:08:07 <API> #info In my personal case (Pi�eiro), the only new stuff is reviewing some bugs opened by joanie. Most of them related to new roles needed due current w3c conversations. On top of that, a wayland crash.
15:08:12 <API> done
15:08:23 <API> btw
15:08:35 <API> joanie, as soon as this state/roles stuff finish
15:08:38 <API> on atk
15:08:54 <API> we need to not forget to open a atspi bug
15:08:59 <API> to port the new stuff there
15:09:00 <joanie> yup
15:09:02 <joanie> I know
15:09:11 <API> ok, was just a reminder
15:09:12 <clown> right.
15:09:17 <API> done (now is true)
15:09:20 <joanie> #info Joanie has begun working on figuring out what all new API we'll need for some web app stuff.
15:09:36 <joanie> #info Joanie will make these changes a priority and then resume working on Orca for 3.16.
15:09:54 <joanie> #info Joanie hasn't yet had time to review all the Wayland-related issues to see if they still exist.
15:09:58 <joanie> done
15:10:38 <mgorse> That reminds me that evolution uses atk_role_register in a few places. I figure I should file a bug on atk, and we can figure out if we should add new roles
15:10:54 <API> on atk?
15:11:02 <joanie> for the missing roles I assume
15:11:03 <API> mgorse, do you want to bring back atk_role_register?
15:11:05 <API> ah
15:11:08 <API> ok
15:11:16 <API> well, I think that the first step wold be
15:11:20 <API> open a bug on evolution
15:11:25 <joanie> the bargain is you can't make roles up, but we have to accommodate role requests.
15:11:26 <clown> yes, from the docs:  "atk_role_register is deprecated and should not be used in newly-written code."
15:11:40 <API> there are several new roles
15:11:48 <API> hopefully now evolution has suitable roles
15:11:55 <joanie> clown: that reminds me: Region, Section, bla bla bla
15:12:02 * joanie notes that down
15:12:04 <API> if not, then I guess that yes, some new roles would be needed
15:12:23 <mgorse> yeah, good point--we should figure out if we have something that fits already
15:12:39 <joanie> related: LibreOffice makes up crap too
15:12:43 * joanie notes that down
15:13:20 <joanie> related++: I need to tackle subroles soon
15:13:44 <joanie> because we don't want role explosion if we can help it
15:13:52 <API> technically
15:13:58 <API> I consider that current situation
15:14:03 <API> is a crater of a role explosion
15:14:06 <joanie> ?
15:14:24 <API> I mean that we already had the role explosion
15:14:45 <API> at least in my opinion
15:14:49 <API> we have so much roles
15:14:58 <joanie> roles, roles everywhere!
15:15:01 <API> that subroles is not only suitable to avoid adding too much in the future
15:15:13 <API> but to tidy/organize what we have
15:15:17 * joanie nods
15:15:22 <API> having said so
15:15:25 <API> something else in this topic?
15:15:27 <API> moving?
15:15:38 * joanie bumps subroles near the top of her list
15:15:39 <joanie> moving
15:15:47 <jjmarin> #info Juanjo is taking a look to the keyboard navigation in GNOME. He learnt a new shortcut, Ctrl+Tab, that moves between groups of controls. Also it seems that the arrows are used in confused ways for navigation. I will write to accessibility list to get some input before reporting bugs to applications.
15:15:54 <jjmarin> #info another related issue is that the accelerators are not shown in the menus is some new redesigned applications like Gedit
15:16:11 <jjmarin> done
15:16:12 <joanie> yay jjmarin! Thanks for doing that
15:16:17 <clown> good stuff.
15:16:29 <jjmarin> np! keyboard is good :-)
15:17:02 <jjmarin> moving :-)
15:17:07 <API> jjmarin : that ctrl+tab is a gtk thing right?
15:17:07 <joanie> :)
15:17:18 <API> I mean that afaik I don't see it implemented on gnome-shell
15:17:28 <API> hmm, I have just realized that Im on a really old gnome-shell
15:17:31 * clown tries ctl-tab...
15:17:44 <jjmarin> no idea, it is https://help.gnome.org/users/gnome-help/stable/keyboard-nav.html.en
15:17:59 <API> as you are already looking at that
15:18:16 <API> could you check if it is a gtk+ thing, or if it is also available on gnome-shell?
15:18:23 <jjmarin> and you can use for example in the configuration panels of the gnome control center
15:18:35 <jjmarin> language for example
15:18:42 <API> check on gnome-shell == check on a recent gnome-shell
15:18:52 <clown> it works in FF to move from tab to tab.  ctl-shift-tab goes backwards.
15:19:11 * clown too many uses of the word "tab".
15:19:51 <API> well, just to move
15:20:10 <API> #action jjmarin with confirm if ctrl+tab shortcut works too on a recent gnome-shell
15:20:23 <API> #topic w3c updates
15:20:24 <API> clown, ?
15:20:34 <clown> a couple of things.
15:21:23 <clown> #info We plan to publish a new working draft of the spec, the core mappings, and the accessible name computation in Dec or early Jan.
15:22:02 <clown> #info There is a lot of activity on three items:  aria-current, role text, and aria-modal.
15:23:00 <clown> #info aria-current is to a way to say "you are here" in some navigation system; for example a table of contents displayed off to the left — the current page might be in bold text visually.
15:23:17 <clown> #info aria-current on that toc entry will relay that info to the a11y API.
15:23:46 <clown> I'm going to skip role text and leave that to joanie, because I think she understands it better.
15:23:58 <joanie> not by choice ;)
15:24:34 <clown> #info aria-modal is going to be an option for the window role and it sub roles to declare when an author creates a window or dialog that restrcits keyboard and mouse events to that window.
15:25:14 <clown> #info I have looked at the various a11y APIs and all of them already have some kind of "modal" state that the aria-modal could be mapped to.  I need to confirm that is okay with everyone.
15:25:18 <clown> done, questions?
15:25:30 <joanie> clown: STATE_MODAL is correct
15:25:41 <clown> that's one confirmation.
15:25:54 <clown> 3 (4) to go.
15:27:50 <API> sorry, was just looking to the definition of ATK_STATE_MODAL
15:27:58 <API> well, without reading aria-modal
15:28:20 <API> I guess that yes, _STATE_MODAL is the more suitable at atk/atspi world
15:29:23 * joanie hums
15:29:29 <clown> thanks API.
15:30:45 <clown> #info Spec for aria-modal:  http://rawgit.com/w3c/aria/master/aria/aria.html#aria-modal
15:31:42 <API> I know that sometime the line is too thin
15:31:59 <API> but it was discussed why an aria property and not an aria state?
15:32:22 <clown> actuall no.  I thought it was state until you just pointed it out.
15:32:27 <API> I guess that is because usually dialogs doesn't change from one to the other, but I was curious
15:32:36 <clown> I guess the thought is it's unlikely to change.
15:33:06 <clown> we are on similar wave lengths, API.
15:33:49 <API> I guess that is all terminology
15:33:55 <API> now Im reading property vs state
15:34:13 <API> and for example, aria-activedescendant, is a property expected to change constantly
15:34:21 <API> anyway, this was just simple curiosity
15:34:26 <API> so Im done with questions
15:34:32 <API> anyone else? next one?
15:34:39 <joanie> arguably in ATK we call things "states" that are really "properties"
15:34:44 <joanie> like multiline
15:34:47 * joanie shrugs
15:35:48 <clown> it's odd, because aria is markup, and in markup there is only attributes and values.  The idea behind calling some things states vs. properties is to give a clue as to which are likely to change, and spawn events (usually states).
15:38:08 <clown> *crickets*
15:38:11 <API> so if I follow what you say
15:38:17 * API slow writing
15:38:22 <clown> no problem
15:38:30 <API> a role is an attribute?
15:38:52 <clown> in markup lingo, yes.
15:39:14 <clown> <div role="dialog" aria-modal="true"> ...
15:39:41 <clown> the "div" in that markup is the element name.  the rest are the elements attributes.
15:40:13 <API> ok
15:40:17 <API> so as you finished
15:40:22 <API> joanie, as clown implied
15:40:34 <API> do you want to give an updated realated to text
15:40:39 <joanie> heh not really
15:40:41 <joanie> but ok
15:40:45 <API> or do you prefer to do that when is more mature?
15:40:55 <joanie> I don't think it's going to mature
15:40:56 <joanie> ;)
15:41:10 <joanie> clown: if you get a link from the spec, I'll start infoing
15:41:20 * clown looking
15:41:37 <joanie> #info The purpose of the new (ARIA 1.1) role="text" is to tell ATs to treat the content as if it were plain or static text.
15:41:43 <clown> #info:  http://rawgit.com/w3c/aria/master/aria/aria.html#text
15:41:47 <joanie> #info Note that this does not mean there is actual text there.
15:42:04 <joanie> #info As the examples in the document Joseph linked to above show, it can apply to an image.
15:42:59 <joanie> #info The motivation, as Joanie understands it, is to cause screen readers to do two things: 1) not say the role because the role doesn't really matter as far as the user is concerned and 2) to prevent screen readers from pausing.
15:43:37 <joanie> #info The pausing happens because screen readers presenting multiple objects at once (e.g. because they're all on the same line) tend to pause -- or really, present each accessible object seperately.
15:44:06 <joanie> #info Joanie understands the rationale and need. Joanie does not understand why it is a good thing to expose non-textual objects as role="text".
15:44:07 <jjmarin> sounds like a hack :-)
15:44:13 <joanie> jjmarin: bingo!!
15:44:18 <joanie> (almost done)
15:44:39 <joanie> #info Having given this much thought and discussion with Piñeiro, our proposal is a new AtkRole.
15:44:54 <joanie> #info That new role is currently -- and likely to remain -- ATK_ROLE_STATIC.
15:44:59 <joanie> I'll get the bug in a second
15:45:17 <joanie> #info Joanie has provided a patch for ATK, complete with lots of documentation for implementors.
15:45:32 <joanie> #info Joanie just wants Alex Surkov (Mozilla a11y developer) to chime in.
15:45:36 * joanie fetches bug
15:46:03 <joanie> https://bugzilla.gnome.org/show_bug.cgi?id=739981
15:46:14 <Services> 04Bug 739981: normal, Normal, ---, jdiggs, ASSIGNED, Need a role for static/generic objects
15:46:15 <joanie> hellllo Services
15:46:17 <joanie> thanks
15:46:19 <joanie> silly bot
15:46:36 <joanie> clown: so once that's blessed (or not), I'll update my action for W3C
15:46:44 <clown> joanie, okay.
15:46:45 <joanie> and do similar patches for AT-SPI2
15:46:48 <joanie> and add it to Orca
15:46:50 <joanie> bla bla bla
15:47:05 <joanie> so I'm done with role="text" in all its ahem glory
15:47:07 <joanie> questions?
15:47:35 <clown> question: would this new role be used also for, say, tables maked with role presentation?  (yes, that old nut).
15:47:40 <joanie> no
15:47:47 * joanie looks at the documentation
15:48:18 <joanie> well, if the table cell were a mere word, one might could make that claim
15:48:25 <clown> oh, is because the new role is "inline", but the table cells (text) content is "block"?
15:48:31 <joanie> but if the table cell is not, and has lots of text....
15:48:48 <joanie> conceptually yes
15:49:00 <joanie> I didn't put inline there because it applies to more than just web content
15:49:20 <clown> okay.  So, the new role applies only to a single word?
15:49:27 <joanie> not only
15:49:40 <joanie> my point is that a table marked as presentation probably isn't a data table
15:49:51 <clown> better not be.
15:49:54 <joanie> data table cells tend to have very little content per cell
15:49:55 <clown> that would be an author error.
15:50:15 <joanie> but a layout table on the other hand I would think stands a good chance of having big chunks of text per cell
15:50:25 <joanie> and the name of that cell probably shouldn't be all that text
15:50:39 <joanie> so as the docs state
15:50:47 <clown> likely the cell shouldn't have a name.
15:50:54 <joanie> crap
15:51:00 <joanie> meant to copy the text and moved to a new link
15:51:08 <joanie> but it states "not a lot of text"
15:51:14 <joanie> and "expose it as a name"
15:51:22 <joanie> and "if that seems wrong, consider a different role"
15:51:38 <clown> okay, the "expose it as a name", is, I think, the distinguishing feature.
15:51:51 <joanie> but let's see what surkov says
15:51:55 <clown> sure.
15:51:56 <clown> thanks.
15:51:57 <joanie> bottom line is that the role should be that
15:52:05 <joanie> and we'll make the docs say what they need to
15:52:23 <joanie> so that Orca's not screwed by this new role ;)
15:52:26 <joanie> and that role="text" doesn't map to ATK_ROLE_TEXT
15:52:38 <joanie> which the patch also addresses
15:52:42 <clown> yes, that is confusing, on the surface.
15:52:52 <joanie> which?
15:53:01 <joanie> anyhoo, maybe we're deep diving?
15:53:08 <clown> that role="text" does not map to ATK_ROLE _TEXT.
15:53:13 <joanie> aha
15:53:14 <clown> yes, I have no more questions.
15:53:15 <joanie> agreed
15:53:27 <joanie> API moving on?
15:53:38 <API> yes, I was waiting for the green light
15:53:42 <API> #topic marketing
15:53:43 <API> jjmarin, ?
15:53:48 <jjmarin> #info No activity in the a11y marketing front lately. I suggest to write some posts in this cycle about the shinny new stuff to showing how gnome a11y is getting better all the time :-)
15:54:01 <jjmarin> (Lennon replies: It can't get more worse)
15:54:07 <joanie> hah
15:54:19 <clown> good 'ol Lennon.
15:54:29 <jjmarin> yeep
15:54:49 <API> the good old blogs
15:54:51 <API> hmm
15:54:54 <API> "eventually"
15:55:04 <API> in any case, the more visible stuff is done by joanie
15:55:12 <API> at least for the users
15:55:13 <joanie> don't make me blog
15:55:21 <joanie> :P
15:55:22 <API> I don't need to make you blog
15:55:27 <API> you already promised that
15:55:33 <API> you made you make to blgo
15:55:35 <API> *blog
15:55:36 <joanie> it's all on "the list"(tm)
15:55:41 * joanie sighs
15:55:47 <API> anyway, no pressures
15:55:57 <API> and so as just 5 minutes to go
15:56:09 <API> #topic misc time
15:56:10 <jjmarin> think about a sort of trailer :-)
15:56:16 <API> and as we were talking about "the list"
15:56:31 <API> #topic API eventually review the last patches for atspi tests
15:56:32 <joanie> a trailer for my car because the TODO list is too big to fit?
15:56:34 <API> argh
15:56:36 <API> #topic misc time
15:56:42 <API> #info API eventually review the last patches for atspi tests
15:56:46 <API> and additionally
15:57:06 <API> #info meanwhile, API published his custom scripts, that can be used as a base for atspi examples:
15:57:16 <API> https://github.com/infapi00/atspi-examples
15:57:17 <jjmarin> hehe (joanie's trailer)
15:57:26 <API> right now they have some C examples
15:57:29 <clown> blocking her rear view mirror.
15:57:43 <API> something that was scarce before I started to wrote them
15:57:53 <API> atspi has some examples, that were outdadted and didn't work
15:58:16 <API> anyway, my idea would be include javascript examples, python-pyatspi2 examples and python-gobject-introspected examples
15:58:20 <API> but as they are just examples
15:58:23 <API> just a bunch
15:58:25 <clown> i recall a focus/caret tracker atspi example.  :-p
15:58:26 <API> clean them a little
15:58:34 <mgorse> API: Do you mean that you reviewed the patch in comment 16, or the earlier patch?
15:58:44 <API> what?
15:58:55 <API> mgorse, ah, the atspi tests
15:59:03 <API> no, I mean that I will
15:59:11 <API> right now the patches are waiting to be reviewed
15:59:18 <API> eventually I would like to review them
15:59:24 <API> if you do that first, they you win ;)
15:59:29 <mgorse> oh, I misread. Sorry
15:59:33 <API> clown, yes that true
15:59:42 <API> as my memory is fading out
15:59:51 <mgorse> I finally started looking at them last night. Mostly looked okay to me, aside from a few small typos
15:59:58 <API> was that example uploaded at pyatspi2, right?
16:00:23 <API> mgorse, ok
16:00:47 <clown> API, I think so.  But, it depends on the magnifier's dbus interface, and that only worked in python2.
16:01:08 <API> it depends on the magnifier dbus interface?
16:01:16 <API> I though that was using only atspi stuff
16:01:23 * joanie has another meeting now
16:01:42 <clown> it uses pyatspi2 to track the events, and the magnifier's dbus interface to tell the magnifier what to "focus" on.
16:02:01 <API> ah ok
16:02:02 <API> hmm,
16:02:12 <API> well it will be really likely that that DBUS interface will disapper
16:02:23 <API> taking into account current bugzilla bug
16:02:25 <clown> yes, I know.  I commented on that bug.
16:02:35 <API> ok
16:02:37 <API> in any case
16:02:42 <API> as I mentioned on the mailing list
16:02:56 <API> I think that makes sense to have all the examples on a different repository
16:03:11 <API> so you could have all of then on the same place
16:03:17 <API> independetly of the language
16:03:20 <API> if you disagree
16:03:27 <clown> sounds good to me.
16:03:27 <API> go to the mailing list
16:03:31 <API> because we are out of time
16:03:39 <API> #endmeeting