16:05:38 <API> #startmeeting
16:05:38 <tota11y> Meeting started Thu Apr  4 16:05:38 2013 CET.  The chair is API. Information about MeetBot at http://wiki.debian.org/MeetBot.
16:05:38 <tota11y> Useful Commands: #action #agreed #help #info #idea #link #topic.
16:05:43 <clown> joanie is truly the fastest gun in the ease.
16:05:44 <API> GSoC
16:05:45 <API> ups
16:05:46 <clown> "east"
16:05:50 <API> #topic GSoC
16:06:00 <API> joanie?
16:06:05 <joanie> sure
16:06:11 <joanie> i'll info as I go
16:06:41 <joanie> #info As discussed last meeting, we have officially added caret and focus tracking for gnome-shell to the list of GNOME's GSoC items.
16:07:09 <joanie> #info We got a response from Marina stating that Matthias thinks we should have a supporting gnome-shell team mentor.
16:07:23 <joanie> #info No one on the CC list (from gnome-shell) has replied yet.
16:07:59 <joanie> #info But the email seems to suggest that our idea is seen as good enough to be a real idea. :)
16:08:27 <joanie> I would propose that API take an action item to ping the gnome-shell folks to get us a supporting mentor.
16:08:30 <joanie> (done)
16:08:34 * clown needs to re-read that.
16:08:45 <API> one thing
16:08:58 <API> I don't have any problem to ping gnome-shell folks
16:09:21 <API> but I guess that you proposed me because Im "near of the gnome-shell team" somehow
16:09:31 <joanie> that and you are fearless leader
16:09:34 <API> but ... Marina is part of the gnome-shell team
16:09:52 <joanie> and no one responded to her email
16:09:52 <API> shouldn't be her a better candidate?
16:10:04 <clown> I think Jasper would be good.  Was he on the list.
16:10:06 <clown> ?
16:10:11 <API> I assume that gnome-shell team can "forget" me, but would be strange to "forget" Marina
16:10:12 <joanie> Jasper was I think
16:10:13 <API> at least twice
16:10:22 <clown> He wrote an event handling bridge for JS coders.  I used it in my code.
16:10:28 <joanie> API: I don't follow you.
16:10:43 <joanie> Jasper seems uber responsive too
16:10:59 <joanie> so if clown thinks he's a good candidate, let's ping jasper
16:11:07 <API> just wondering if gnome-shell folks would give more attention to any ping from Marina, as she is part of their team
16:11:09 <joanie> he might even be in #a11y
16:11:13 <clown> Maybe I should reply to that email with suggesting Jasper?
16:11:21 <joanie> one sec
16:11:43 <API> yes, jasper is right now on a11y
16:11:54 <API> he made the review of some of the gnome-shell a11y patches
16:11:57 <joanie> I invited him to join us here.
16:12:01 <Jasper> joanie, OK, but I'm not exactly sure what you need.
16:12:05 <joanie> :)
16:12:27 <joanie> So we have added gnome-shell caret and focus tracking for gnome-shell magnifier as a GSoC idea.
16:12:38 <clown> joanie, is the GSoC email from Marina on list somewhwere?
16:12:39 <joanie> and it will be co-mentored by clown API and myself
16:12:47 <joanie> clown: no, was private
16:12:57 <Jasper> I thought it was on d-d-l?
16:13:01 <clown> so, if Jasper was cc'ed he would have it.
16:13:13 <joanie> oh maybe it was cc'ed
16:13:17 <joanie> (to ddl)
16:13:19 <joanie> anyhoo
16:13:38 <joanie> during this meeting, we were thinking that Jasper would be an excellent co-mentor from the gnome-shell team
16:13:42 <clown> I volunteered you, Jasper :-)
16:13:49 <joanie> so rather than have an action item to ping you
16:13:53 <joanie> we invited you to this meeting
16:13:54 <joanie> :)
16:14:07 <Jasper> I'm not sure I"m the best candidate. rtcm knows more of that sort of stuff.
16:14:26 <Jasper> I'm welcome to learn it and mentor if rtcm is unavailable, though.
16:14:35 <clown> Okay, I suggested you since you wrote the signal.js, and I made extensive use of it in my first attempts.
16:14:45 <Jasper> signal.js?
16:15:06 <joanie> and you also help fix a11y issues, review a11y patches, and are uber-responsive (thanks!!)
16:15:09 <clown> I think that was the name of the package.  It allowed added signal handlers to any JS object.
16:15:24 <clown> It's part of GJS.
16:16:14 <joanie> so.... what is the conclusion?
16:16:22 <joanie> ping rtcm?
16:16:39 <Jasper> clown, I didn't write it.
16:16:43 <clown> https://git.gnome.org/browse/gjs/tree/modules/signals.js
16:16:43 <joanie> heh
16:16:48 <Jasper> https://git.gnome.org/browse/gjs/log/modules/signals.js
16:16:51 <clown> oh, never mind then...
16:16:59 <Jasper> I wrote Lang.Class.
16:17:24 <joanie> well, what I said still stands (re a11y issues, patches, and responsiveness)
16:17:25 <Jasper> But I don't think that makes me a good candidate for a mentor of a random gnome-shell feature.
16:17:36 <joanie> ok
16:17:40 <Jasper> joanie, if rtcm is unavailable, I'll volunteer to mentor.
16:17:48 <Jasper> But I really think rtcm would be a better fit for this feature.
16:17:50 <joanie> woo hoo!
16:17:52 <joanie> ok
16:18:03 <joanie> (we have a meetbot, so ignore me)
16:18:29 <joanie> #info Jasper suggested we ping rtcm. But if rtcm is unavailable, Jasper will volunteer to co-mentor.
16:18:41 <joanie> #action Pineiro will ping rtcm.
16:18:56 <joanie> okay, any other thoughts on this front?
16:19:04 <joanie> (and Jasper, thanks for joining us!)
16:19:35 <clown> yes, thanks Jasper.
16:19:36 <API> ok, so anything else in this point?
16:20:13 <joanie> i think that means "no"
16:20:14 <joanie> :)
16:20:18 <Jasper> OK.
16:20:46 <API> so moving on
16:21:08 <API> although Jasper was also a good guy for next point
16:21:19 <joanie> oops
16:21:25 <API> #topic Wayland
16:21:34 <API> joanie, no problem, as this is a starting point
16:21:39 <API> anyway, going on
16:22:01 <API> #info Although Wayland was seen as the future for X, no real plans or deadline were set
16:22:06 <API> #info until now
16:22:20 <joanie> dun-dun-dun
16:22:33 <API> #info recently Matthias Clasen sent a email trying to start setting some deadlines:
16:22:47 <API> https://live.gnome.org/Wayland/
16:23:07 <API> https://live.gnome.org/Wayland/#Proposed_roadmap
16:23:43 <API> #info as you can recall, we created a bug about "start to check Wayland" some time ago
16:24:01 <API> #info and we briefly discussed it on the 2011 hackfest, but was basically skipped
16:24:20 <API> #info first reason, lack of knowledge, second one, lack of deadlines
16:24:36 <API> #info now it is time to start again this
16:24:48 <API> #info I helped to fill accessibility section:
16:24:58 <API> https://live.gnome.org/Wayland/Gaps#Accessibility
16:25:12 <API> but probably would need more info
16:25:35 <API> #action piñeiro will send a email to accessibility-dev, just in case that section needs to be filled
16:25:53 <API> #info as you recall, I also sent a email a long time ago to X devel
16:25:55 <API> https://mail.gnome.org/archives/gnome-accessibility-list/2012-May/msg00007.html
16:26:09 <API> #info for some needs of accessibility (one was that snooping thing)
16:26:22 <API> #info so my plan is sending a similar one to Wayland mailing list
16:26:48 <API> #action Piñeiro will fill a "what accessibility needs to wayland, how to use it" email, ask for review, send to the list
16:26:56 <API> (done with the #infos)
16:26:59 <API> in that sense
16:27:25 <API> that mail that I sent to X-dev was more like a "what we want" but I didn't add "why we want"
16:27:38 <API> recently I had a brief chat with Daniel Stone
16:28:04 <API> that has a lot of expeirence on X, and is doing a lot of the stuff on Wayland
16:28:31 <API> and I came to the conclusion that as part of that discussion,
16:28:45 <API> so they would have more info about solutions to propose
16:28:54 <API> it would be good to say why we need that stuff
16:29:06 <API> having said so, I have been talking a lot about this
16:29:10 <API> so?
16:29:14 <API> comments, doubts, questions?
16:29:43 <joanie> #info Joanie wonders what we will do about at-spi2.
16:29:49 <API> mgorse, last time I was reviewing X code on at-spi2, I felt that was somewhat messy
16:30:00 <API> so I guess that this would be also a good moment to do some cleaning
16:30:19 <mgorse> You mean the code that handles mouse events?
16:30:30 <API> as I assume that no matters if Wayland arrives, at-spi2 would need to support X for a long while
16:30:37 <API> mgorse, well, in general any X code
16:30:48 <API> because it is true that are mouse event handling
16:30:52 <API> but there are others
16:31:00 <API> in several cases C&P from at-spi1
16:32:08 <mgorse> Would these things be handled by Wayland? When I looked at its Wikipedia page, I think it said that Wayland was used for rendering and things like input need to be handled with APIs outside of Wayland
16:32:43 <API> well in general
16:32:49 <API> also from that Daniel Stone conversation
16:33:02 <API> Wayland is moving a lot of what X did to the client
16:33:25 <API> the poster boy here is the window decorations (before was renderer by X, now is needed to be rendered by the toolkits)
16:33:47 <API> but event input from devices are still managed by Wayland somehow
16:34:04 <API> yes, probably they delegate more on other parts
16:34:11 <API> but Wayland is involved there
16:34:26 <API> also Wayland provides support for extension
16:34:54 <API> on GUADEC, when this was suggested, a quick answer from <insert-name-here> Bradford, from intel
16:35:12 <clown> just out of interest, where do focus and caret events come from?  inside X or inside GTK (or whatever toolkit)?
16:35:28 <API> was that a possibility would be a extension of the protocol on the compositor (on that case could be gnome-shell)
16:35:39 <API> anyway, I'm just babbling, as I lack a lot of details
16:35:45 <joanie> clown: toolkit
16:35:54 <clown> whew!
16:36:02 <joanie> :)
16:36:10 <API> clown, yes, although on the GTK implementation
16:36:16 <API> they used X grabs and a lot of stuff
16:36:23 <clown> oh no, there's a "but".
16:36:30 <API> that came from X
16:36:37 <API> so joanie was right
16:36:42 <clown> what's an "X grabs"?
16:36:44 <API> from the a11y pov that comes from gtk
16:36:46 <API> and will c ome
16:36:53 <joanie> (was bound to be right eventually)
16:36:56 <API> s/gtk/the toolkit
16:37:09 <API> from the toolkit pov, probably they will be using X or Wayland stuff somehow
16:37:34 <mgorse> Also, I think that what Orca needs is (1) the ability to passively listen for keys (ie, if a user presses control, then it should silence, but it doesn't need to intercept the key), and (2) the ability to intercept keys, being flexible about the combinations it defines  (ie, kp_insert plus another key). I don't think that it necessarily needs to be able to intercept all keys, unless I'm missing something, although intercepting all keys is one way
16:38:01 <mgorse> There are problems with the way key grabs are done currently. Ie, if Orca is busy, then keys can bleed through to the application, and I don't see a good way to fix that with the current design
16:38:37 <joanie> mgorse: in a perfect world, Orca wouldn't need to know all input events.
16:38:40 <API> mgorse, about (2), that is needed for orca keybindings or for any other reason?
16:38:43 <joanie> but this is not a perfect world
16:39:11 <joanie> and when we have to heuristically guess about object events, knowing what the user just did is helpful.
16:39:23 <joanie> also, keypresses should interrupt speech
16:39:32 <joanie> not just control, but when typing fast
16:39:47 <mgorse> joanie: In those cases, Orca needs to know about the events, but it doesn't need to block them
16:39:56 <joanie> right
16:41:29 <joanie> so....
16:41:45 <joanie> do we have a plan?
16:41:59 <joanie> mgorse: do you know if you'll have time to work on this?
16:43:30 * joanie hums
16:43:32 <mgorse> for  Wayland? I'd need to figure out what API is going to be used there.
16:44:28 <mgorse> and who to talk to in general if we're going to add API, but, yeah, I think I could help with it
16:44:39 <joanie> yay re helping!
16:45:59 <API> taking into account that we already talked a little about it
16:46:08 <API> what do you think about the action items I proposed before?
16:46:15 <joanie> I think they make sense
16:46:25 <joanie> I just want to be sure we have some sort of plan
16:46:42 <joanie> a11y has been improving quite nicely recently
16:46:52 <mgorse> Yeah, makes sense. Which list are you sending the email to, about what we need? g-a-devel?
16:46:53 <joanie> and I don't want to lose that to Wayland
16:47:09 <API> mgorse, well as I said, but will summarize:
16:47:25 <API> 1.) send a ping to ask people to fill that gaps section just in case
16:47:56 <API> 2.) start to write a "what (and why) accessibility needs, please gave us some guidance" mail
16:48:04 <API> 3.) ask for review to the team
16:48:12 <API> 4) send that email to wayland-dev
16:48:32 <mgorse> ok
16:48:36 <API> as I said, we can't coordinate, split all the tasks now
16:48:43 <API> but at least we need this to start rolling
16:48:48 * joanie nods
16:48:55 <API> btw: random rant, I don't like wayland-dev
16:48:58 <API> in the sense
16:49:12 <API> that they use  that list for discussions
16:49:16 <API> but also for patch reviews
16:49:30 <API> have a lot of traffic, and most of the stuff is patch reviewing
16:49:34 <API> rant-off
16:49:37 <joanie> :)
16:49:40 <API> ok, so finishing the topic?
16:49:42 <joanie> .procmailrc ftw
16:49:55 * clown joseph-the-nag notes the lack of infos...
16:50:06 <joanie> the infos were before
16:50:14 <joanie> but....
16:50:33 <joanie> #action Joanie will take the summary from Pineiro and add it to the minutes.
16:50:37 <joanie> meeting is going long
16:50:42 <clown> the last info I see is "Joanie wonders what we will do about at-spi2."
16:50:46 <joanie> though it was an important topic.
16:51:00 <clown> okay.  more work for joanie
16:51:01 <joanie> clown: what else should be infoed?
16:51:10 <joanie> not a big deal
16:51:14 <clown> gah, more work for clown...
16:51:17 <joanie> hah
16:51:23 <joanie> let's move on
16:51:38 <clown> yes, let's.  forget I said anything.
16:52:09 <API> so moving?
16:52:20 <API> I will skip marketing as jjmarin is not here
16:52:28 <API> #topic W3C updates
16:52:30 <API> clown?
16:52:35 <clown> hi there.
16:53:23 <clown> #info Testing browsers for compliance/conformance for ARIA 1.0 has sped up.
16:53:45 <clown> #info Hoping to wrap that up in the next month or so.
16:54:11 <clown> #info Tentative "release" of ARIA 1.0 would be Jun.
16:54:17 <clown> questions?
16:54:36 <joanie> wow
16:54:52 <joanie> what motivated the speed-up?
16:55:37 <clown> well, there was always motivation.  what sped things up were "bodies" — testers who would run the tests.
16:55:53 <joanie> aha
16:55:54 <joanie> cool
16:56:18 <clown> there have been a couple of students conscripted from the University of Illinois.
16:56:50 <joanie> student labor yay! :)
16:56:52 <clown> I hear rumours that someone from the Google Chrome team started to test Chrome on Windows with the tests.
16:56:58 <clown> but just rumours...
16:57:07 * clown wishes the test results page was public.
16:57:24 <joanie> we can ask Mark :)
16:58:06 <clown> sure.  A number of us have been pressuring Michael Cooper, who the W3C person who manages the test harnes.
16:58:14 <clown> "harness".
16:59:32 <joanie> anyhoo... nearly meeting end time
16:59:48 * joanie pokes API
16:59:52 <clown> I have a related technical question, but maybe it's better for misc time.
17:00:09 <API> clown, ok
17:00:14 <API> #topic Misc time
17:00:19 <clown> lol
17:00:20 <API> done ;)
17:00:48 <clown> I've been digging through the documentaion for IA2 and ATK (and AT-SPI).
17:00:56 <clown> "AT-SPI"
17:01:22 <clown> looking for restrictions on the controller-for/controlled-by relation.
17:01:42 <joanie> that has always been a mystery relation to me
17:01:46 <clown> I don't see any.  Are there restrictions on their use?
17:02:05 <clown> It looks like any object can be declared as a controller for any other object.
17:02:08 <joanie> so are the controlled relations controlled?
17:02:10 <joanie> :)
17:02:15 <clown> hee, hee.
17:02:28 <API> fwiw, I never used that relation
17:02:35 <joanie> me neither
17:02:38 <API> when I arrived to this world, they were already there
17:02:52 <API> on atk the documentation is really slim
17:02:53 <API> ATK_RELATION_CONTROLLED_BY
17:02:53 <API> Indicates an object controlled by one or more target objects.
17:02:57 <joanie> heh
17:03:00 <API> controlled how?
17:03:00 <clown> I even went as far back as the java accessilbility interface.  same thing, e.g., "Indicates an object is a controller for one or more target objects"
17:03:01 <joanie> fail
17:03:05 <joanie> exactly
17:03:05 <API> when?
17:03:09 <API> when use the relation?
17:03:12 <API> etc
17:03:16 <joanie> why?
17:03:20 <clown> wow.  even the documenation is exactly the same...
17:03:29 <API> clown, fwiw
17:03:41 <API> atk/at-spi was heavily based on the java stuff
17:03:42 <clown> tab list is a common example.
17:04:00 <mgorse> gtk doesn't use those relations anywhere
17:04:01 <clown> the tab controls the visbility of the associated tab pane.
17:04:06 <joanie> #info Joseph asked about the controller-for/controlled-by accessible relations.
17:04:10 <API> something that I guess that makes sense as they both came from Sun
17:04:19 <joanie> #info The team is perplexed.
17:04:20 <joanie> :)
17:04:32 <clown> you actually info'ed that???
17:04:38 <joanie> clown: yeah
17:04:45 <joanie> we never have items under misc time minutes
17:04:49 <joanie> it's sad
17:05:00 <joanie> it's also related to the next thing I'll bring up.
17:05:30 <joanie> #info Speaking of nebulous API, the accessible text interface is crazy huge and confusing.
17:05:40 <joanie> #info We have discussed simplifying it before.
17:05:49 <joanie> #info Joanie thinks perhaps we should actually do that.
17:05:59 * joanie looks for a blog post from Surkov
17:06:06 <API> and never advance too much on that as "it works well enough and other stuff doesn't work at all"
17:06:18 <joanie> ?
17:06:23 <API> afair, initial plan was taking a look to ia2
17:06:30 <API> as a example of something more simple
17:06:33 <API> <joanie> ?
17:06:41 <API> I mean that the reason taht simplification never happened
17:06:44 <API> or never worked on that
17:06:55 <joanie> http://asurkov.blogspot.ru/2013/03/an-easy-way-to-understanding-atk-text.html
17:06:57 <API> was because we needed to fix first the stuff that didn't work at all
17:06:59 <mgorse> Is i2's text interface not pretty much the same as atk's?
17:07:08 <joanie> i think it is
17:07:12 <joanie> that doesn't make it right or good
17:07:13 <joanie> ;)
17:07:22 <mgorse> yeah
17:07:48 <joanie> so if we are going to do a major API change re text, we should do it early in the cycle
17:07:54 <joanie> and start deprecating stuff
17:07:55 <joanie> imho
17:08:14 <joanie> and while we're at it, maybe controlled-by/controller-for should be nuked
17:08:17 <joanie> (deprecated)
17:08:18 <mgorse> UIA's text pattern is pretty different from AtkText, fwiw. On the other hand, I have no idea if any screen readers are using it
17:09:06 <joanie> anyhoo, something to think about -- and probably soonish.
17:09:13 <joanie> right API?
17:09:35 <API> we should do the changes early in the cycle
17:09:39 <API> but in my opinion
17:09:46 <API> with a clear idea of what we want
17:09:49 <clown> I'll just note that FF passes gloriously on IA2 and AT-SPI for the aria-controls attribute tests.
17:09:50 <mgorse> me wonders when a gtk 4 will be released
17:09:53 <API> so lets go step by step
17:10:05 <API> step one would be getting a API everybody is happy with
17:10:10 <joanie> yup
17:10:16 <API> step two will be decide how to integrate it on any cycle
17:11:02 <joanie> yup
17:12:52 <joanie> all is silent
17:12:54 <joanie> meeting over?
17:13:54 <joanie> tota11y: chair
17:13:54 <tota11y> joanie: Error: "chair" is not a valid command.
17:13:57 <joanie> hmmm
17:14:05 <joanie> tota11y: addchair
17:14:05 <tota11y> joanie: (addchair <channel> <network> <nick>) -- Add a nick as a chair to the meeting.
17:14:06 <clown> tota11y:  sofa
17:14:06 <tota11y> clown: Error: "sofa" is not a valid command.
17:14:19 <joanie> tota11y: addchair joanie
17:14:19 <tota11y> joanie: (addchair <channel> <network> <nick>) -- Add a nick as a chair to the meeting.
17:14:24 <joanie> ?
17:14:31 <clown> you forgot the channel
17:14:38 <joanie> it should be implied
17:14:40 <joanie> but yeah
17:14:46 <API> sorry, someone pinged me
17:14:53 <API> so yes
17:14:55 <API> meeting over
17:14:58 <API> #endmeeting