14:05:26 <API> #startmeeting
14:05:26 <Services> Meeting started Thu May 22 14:05:26 2014 UTC.  The chair is API. Information about MeetBot at http://wiki.debian.org/MeetBot.
14:05:26 <Services> Useful Commands: #action #agreed #help #info #idea #link #topic.
14:05:29 * clown starts paying attention
14:05:45 <API> hmm, joanie, do you know if it is normal the lack of topic at the room?
14:05:55 <joanie> ?
14:06:07 <API> I mean even before starting the meeting
14:06:13 <API> in any case, lets go on with the meeting
14:06:31 <API> #topic Progress towards 3.14
14:06:34 <clown> yes, when I joined, there was no "This is the GNOME A11y team meeting room", or some such.
14:06:46 <joanie> I can put it back after
14:06:49 <clown> but, no matter.
14:06:52 <API> joanie, ok thanks
14:07:04 <API> #info the main topic this week is a 3.12 regression
14:07:29 <clown> how is that possible?  :-)
14:07:43 <API> #info specifically bug 730118, and similar one, bug 728319
14:07:43 <Services> 04Bug http://bugzilla.gnome.org/show_bug.cgi?id=730118 normal, Normal, ---, gtk-bugs, NEW, GtkTreeView very slow since 3.12
14:07:44 <Services> 04Bug http://bugzilla.gnome.org/show_bug.cgi?id=728319 normal, Normal, ---, at-spi-maint, UNCONFIRMED, Hangs when browsing using artist view
14:08:28 <API> #info probably those two bugs are the same, but for now, it makes sense to keep it independently reported
14:08:32 <mgorse> I get the impression there was IRC discussion that I missed. I'll look at the bug, anyhow
14:08:51 <joanie> mgorse: I think it was in a gtk channel
14:08:58 <API> #info gtktreeview event flood was a "classical" bug for gtk 2.xx
14:09:39 <API> #info Benjamin Otte mitigated it on 3.XX by stopping to emit children-changed events (and others) under the assumption that under the state MANAGE_DESCENDANTS, that was not needed
14:10:07 <API> #info it seems that 3.10 has some leaks related with the individual children of gtktreeview. As a collateral effect of solving that, for 3.12, those events are back
14:10:14 <API> in any case
14:10:34 <API> #info users are experiencing this, even if no AT is listening
14:10:43 <API> #info So I see two aspects to solve
14:11:13 <API> #info 1. check why the bus is getting flooded when no AT is listening, as in theory no DBUS event should be triggered on that case
14:11:36 <API> #info 2. mitigate the problem when AT is listening
14:11:48 <API> in that sense, there are two regressions here
14:12:26 <API> mgorse, btw, I'm right on #1? when all this "a11y on default" or "a11y always on" was set
14:12:34 <API> I remember that one of the things that you improved, was
14:12:48 <API> not emitting events through the bus if no AT is listening
14:12:57 <API> is that right?
14:13:35 <mgorse> I'll need to look. Children-changed, state-changed, etc. are special-cased because they update the cache
14:13:53 <mgorse> not sure off-hand if they end up getting emitted even when an AT isn't listening, but really that would be a bug if they are
14:14:10 <clown> regarding no AT listening...
14:14:33 <mgorse> I think I've mitigated it for 3.13--they now don't get emitted for transient objects--but I haven't really tested that the patch improves things. I should be able to do that in the next few days
14:14:38 <clown> when gnome-shell starts up, it immediately attaches listeners for focus and tracking events (or did, last time I looked).
14:14:52 <clown> not tracking events, but caret events.
14:14:56 <mgorse> but there might still be a bug if they're being emitted when no AT is listening
14:14:58 <API> mgorse, well on the bug someone tested and said no improvement
14:15:02 <mgorse> oh, wait. I might know what's going on
14:15:23 <mgorse> I think I looked into that a while ago, and gnome-shell was still registering somehow
14:15:23 <API> clown, fwiw, some people mentioned that this is even a problem on starting
14:15:25 * API looking
14:16:07 <API> mgorse, well, probably is because gnome-shell is one of the first applications that registers
14:16:12 <API> not sure if that makes it a special case
14:16:25 * clown recalls his original design with a singleton listener object that registered only when the user requested focus/caret tracking.
14:16:32 <API> mgorse, in any case, do you agree with my #1 and #2 classification of the issue?
14:16:43 <mgorse> I thought gcampax committed code that made it not register if the magnifier wasn't being used, but it seemed that I was still seeing it register anyhow. Not sure why
14:17:06 <mgorse> yeah
14:17:10 <API> hmmm true
14:17:11 <clown> mgrorse, dunno.  I'm not paying close attention anymore.  I could look around if you like.
14:17:20 <API> gnome-shell could be that AT listening ....
14:17:27 <API> I didn't realize that
14:17:34 <API> clown, sorry I misunderstood your first comment
14:17:40 <joanie> huh
14:17:47 <API> I thought you were talking about registering as an application
14:17:58 <API> but it is true, it is also registering himself as an AT
14:18:07 <joanie> This'll be a fun cycle then ;)
14:18:12 <clown> "it" == gnome-shell?
14:18:18 <API> clown, yep
14:18:31 <API> so there is an option that is also a bug on gnome-shell
14:18:42 <API> being #1 not a problem on at-spi but on gnome-shell
14:18:53 <API> hmm
14:18:55 <API> but is odd
14:19:01 <API> I think that someone mentioned classic mode
14:19:04 * API checking
14:19:17 <joanie> but classic mode is now an extension
14:19:18 <API> https://bugzilla.gnome.org/show_bug.cgi?id=730118#c14
14:19:18 <Services> 04Bug 730118: normal, Normal, ---, gtk-bugs, NEW, GtkTreeView very slow since 3.12
14:19:25 <mgorse> You could argue that maybe the caching stuff should be redesigned, but I think that's reallhy another topic--currently I think it's by design
14:19:32 <API> "
14:19:32 <API> Not that I know of. I'm using classic mode if that matters."
14:20:29 <API> mgorse, well but that cache is there because an application is listening, right?
14:20:36 <API> I mean, if no  AT  is there
14:20:59 <API> would be the events still emitted to update the cache?
14:21:09 <mgorse> Yeah, I think it's by design. Also, you said that someone tested the at-spi patch. Did they leave a comment somewhere? If so, then I haven't seen it
14:21:32 <API> mgorse:
14:21:34 <API> https://bugzilla.gnome.org/show_bug.cgi?id=730118#c15
14:21:34 <Services> 04Bug 730118: normal, Normal, ---, gtk-bugs, NEW, GtkTreeView very slow since 3.12
14:21:35 <mgorse> I don't think so, although it's probably worth double-checking
14:21:53 <mgorse> oh, yeah--haven't read that bug yet
14:21:59 <mgorse> Thanks
14:22:04 <API> you are welcome
14:22:13 <API> in any case, as this is just an informative meeting
14:22:27 <API> mgorse, so you will have time to take a look to this issue?
14:23:09 <mgorse> #action mgorse will check that no events are fired when an AT is not listening, including events related to the cache
14:24:14 <API> mgorse, if needed, it would be easy to create a hack to avoid gnome-shell registering, in case that helps
14:24:33 <API> just if needed to confirm if gnome-shell is registering always or not
14:24:33 <mgorse> ok, that might be useful
14:24:36 <API> ok
14:25:10 <clown> yes, ideally g-s registers only when the user preference is for some type of tracking.
14:25:34 <API> clown, well, as I say, just to confirm, I would use the hacky "comment out the registering lines"
14:25:43 <clown> and, deregisters if the user ever turns it off.
14:25:44 <API> if confirmed that the problem is there, we can
14:25:50 <API> go for the real solution later
14:25:57 <clown> right API.
14:26:03 <API> so having said so
14:26:16 <API> more questions? doubts? more on the current topic?
14:27:51 <API> is the sound ... of the crickets
14:27:58 <API> #topic W3C updates
14:28:00 <API> joanie, clown ?
14:28:26 * joanie defers to clown
14:28:47 <clown> I have one official info, and then I know about other stuff, but joanie knows better.
14:28:56 <joanie> heh
14:28:56 <clown> I'll give the info for now...
14:29:30 <clown> #info The ARIA editors met yeseterday, and agreed to try to publish the next working draft of the 1.1 spec of all the documents by early June.
14:30:20 <clown> I've was away on Mon, but I heard discussions about annotations and "you-are-here" continued.  Over to you joanie.
14:30:31 <clown> :-)
14:30:34 <joanie> about annotations?
14:30:40 <joanie> or mapping of presentational children
14:30:56 <clown> all, or whatever you feel is good to summarize for this meeting.
14:31:00 <clown> I leave it up to you.
14:31:06 * joanie smiles
14:31:16 <joanie> #info The question of how to map "presentational" children came up.
14:31:40 <joanie> #info Presentational children are expected to be treated like generic objects, so the answer is "like a div"
14:31:54 <joanie> #info Alex Surkov seems to think I need to prove it. ;) ;)
14:32:09 <joanie> how's that? :P
14:32:23 <clown> joanie, are you working on a theorem for Alex?  :-)
14:32:39 <joanie> #info Both WebKitGtk and Gecko fail to do the correct thing, but they do the wrong thing in their own unique fashion.
14:32:46 <joanie> #info Joanie filed bugs for each.
14:33:02 <clown> This might be relevant:  http://www.w3.org/WAI/PF/aria-practices/Overview.html#presentation_inheritance
14:33:20 <joanie> #info Related: Because the ATK docs are not really explicit on roles, Alex Surkov had mapped it to ATK_ROLE_TEXT
14:33:41 * clown shuts up for now.
14:33:43 <joanie> #info Which, as we all know is how other implementors expose text widgets like GtkTextView.
14:34:03 <joanie> #info Joanie hopes to not have to add yet another hack (tm) to Orca for Gecko.
14:34:23 <joanie> #info If Mario and his colleagues don't fix the WebKitGtk one, Joanie will do so herself.
14:34:37 <joanie> aside from links to the bugs I think that's all I have on that topic
14:34:53 <joanie> https://bugzilla.mozilla.org/show_bug.cgi?id=1013584
14:34:53 <Services> 04Bug 1013584: normal, --, ---, nobody, NEW , [ATK] Children of tables and lists with role='presentation' should each be exposed as if they were div elements
14:35:20 <joanie> https://bugs.webkit.org/show_bug.cgi?id=133133
14:35:20 <Services> 04Bug 133133: Normal, P2, ---, webkit-unassigned@lists.webkit.org, NEW, AX: [ATK] Children of tables and lists with role='presentation' should each be exposed as if they were div elements
14:35:30 <joanie> clown: anything else?
14:35:45 <clown> just a question:  where is Alex asking for proof?
14:35:56 <joanie> I exaggerated a tad
14:35:59 <joanie> hence the winky
14:36:02 <joanie> but in the bug I filed
14:36:06 <clown> two of them!
14:36:09 <joanie> I say "please map it like this"
14:36:19 <joanie> He says "but...."
14:36:41 <joanie> it's all in the mozilla bug
14:37:03 <clown> okay, I'll look closely later.
14:37:08 <joanie> progress, he's asking Jamie now
14:37:13 <joanie> who hopefully will agree
14:37:21 <joanie> because if it's good for windows we can have it
14:37:25 <joanie> </rant>
14:37:37 <clown> lol
14:38:23 <jjmarin> who's Jamie ?
14:38:31 <joanie> One of the NVDA devs
14:38:38 <jjmarin> ah, ok
14:38:40 <joanie> and the one who does the Gecko stuff I believe
14:38:52 <joanie> I don't think Mick (other main dev) hacks on gecko support
14:39:39 <joanie> assuming nothing else, moving on?
14:39:40 <clown> reading the mozilla bug, the fact that the list items disappears seems to have been forgotten.
14:39:56 <joanie> once they are exposed as divs, that problem will be solved
14:40:00 <clown> sure, move on.
14:40:06 <clown> right, joanie.
14:40:08 <API> if you finished I have just a small thing to add
14:40:21 <API> as part of me reading this days about html canvas accessibility
14:40:23 <API> I found this:
14:40:39 <API> #info HTML Canvas 2D Context Returns to Last Call
14:40:43 <API> http://www.w3.org/blog/news/archives/3850
14:40:59 <API> #info some of the changes are accessibility related
14:41:12 <API> in short, canvas accessibility is based on provide a fallback html
14:41:14 <joanie> I'm assuming PFWG is on top of that??
14:41:23 <joanie> those who go to that concall
14:41:38 <API> that draft includes some of the stuff that could be useful for that fallback html
14:41:49 <API> like  the ability to visually indicate when a region is in focus
14:41:50 <clown> from the document:  http://www.w3.org/TR/2014/WD-2dcontext-20140520/#changes-causing-the-progression-back-to-last-call
14:41:53 <API> <joanie> I'm assuming PFWG is on top of that??
14:41:55 <API> well no idea
14:42:00 <API> that is part of the reasons I mention it
14:42:33 <joanie> API I've been going to ARIA and UAIG meetings but stopped the PFWG general meeting
14:42:40 <clown> PF is sort of on top of it.  There is an associated task force called the HTML Accessiblity Task Force.
14:42:41 <joanie> three a week was too much
14:42:59 <clown> The HTML a11y TG is made up of members from PF and HTML working groups.  I'm not one of thme.
14:43:02 <clown> *them.
14:43:03 <API> joanie, yes I know, but as I catched that
14:43:08 <API> just mentioned just in case
14:43:12 <jjmarin> sure, it sounds like a job :-)
14:43:19 <clown> But, Rich is very much involved.
14:44:16 <API> anyway, that was somewhat a side-line comment
14:44:27 <API> so if you don't have more additional information or questions
14:44:31 <API> I will move on
14:44:35 <API> is that fine?
14:44:43 <joanie> fine with me
14:44:46 <clown> It's likely in good hands, but if you or anyone want to point out a11y problems, being in "last call" give you that opportunity.
14:45:36 <API> well, I think that Im arriving late for that boat ;)
14:45:36 <clown> okay, move on is fine with me.
14:45:37 <API> but hanks
14:45:39 <API> *thanks
14:46:06 <API> #topic Marketing
14:46:08 <API> jjmarin, ?
14:47:30 <jjmarin> #info Juanjo has started to add the new entries in the Spanish Wikipedia. It very early stage though
14:47:39 <jjmarin> done :-)
14:48:00 <joanie> jjmarin: Did you give that talk?
14:48:17 <jjmarin> yes, it was a little audience :-)
14:48:24 <joanie> how'd it go?
14:48:25 <jjmarin> about 15 people or so
14:48:35 <joanie> that's good turn out for a11y
14:48:40 <joanie> :)
14:48:51 <jjmarin> :-)
14:49:06 <jjmarin> it was more a general talk about GNOME
14:49:13 <jjmarin> though I mentioned a11y :-)
14:49:25 <joanie> cool
14:50:08 <API> ok, thanks, so anything else? questions and/or doubts?
14:51:28 <API> I will take that as a "no" ;)
14:51:33 <API> #topic Miscellaneous Time
14:51:42 <API> something not included on agenta to mention?
14:51:44 <API> *agenda
14:52:00 <joanie> #info We really need to write a bunch of ATK documentation.
14:52:05 <joanie> I know we know that
14:52:11 <joanie> and I don't know when we'll get to it
14:52:35 <joanie> maybe we should have a documentation a11y hackfest
14:52:40 <joanie> three days of writing
14:53:03 <joanie> that's all
14:53:04 <joanie> :)
14:53:37 <joanie> s/documentation/developer documentation/
14:53:42 <jhernandez> joanie: you should make your suggestion look like more attractive
14:53:50 <joanie> heh
14:53:58 <jhernandez> I'd say: three days of drinking beers and writing doc
14:53:59 <jhernandez> :P
14:54:25 <joanie> Ok, jhernandez you're in charge of organizing it
14:54:29 <jhernandez> xDDDD
14:54:35 <joanie> And you live in Beerland now
14:54:43 <joanie> so we'll come there
14:54:48 <clown> do you have to drink beer to read the final docs produced by this hackfest in order to understand them?
14:54:58 <jhernandez> xDD
14:55:02 <jhernandez> clown: probably, yes
14:55:03 <API> joanie, fwiw, it would be good to also include at-spi2-atk documentation
14:55:04 <jhernandez> xDDD
14:55:08 <jjmarin> jhernandez cooks his own beer :-)
14:55:09 <API> just reviewing a recent bug
14:55:17 <API> I realized that has exactly 0 documentation
14:55:19 <joanie> API sure
14:55:39 <joanie> API at least the docs aren't vague then :P
14:55:46 <mgorse> lol
14:55:54 <jjmarin> then the documentation status can be easily improved :-)
14:56:35 <joanie> Snark aside, the problem is that implementors do what imho is clearly wrong. Then I go to the docs to prove it and understand why an implementor might have done what he/she did.
14:57:05 <joanie> Or failed to do something
14:57:21 <API> joanie, we still have pending that "focus, what we want, what you should do" documentation for gtk folks, remember ;)
14:57:37 <joanie> ball's in matthias's court I think on that one
14:57:46 <joanie> but maybe not
14:57:54 <joanie> we should have a hackfest for this
14:58:19 <joanie> maybe around boston summit?
14:58:29 <API> a focus hackfest?
14:58:32 <joanie> no sir
14:58:41 <jjmarin> hehe
14:58:44 <joanie> a documentation-writing (and specing out) hackfest
14:59:10 <clown> a hackfest focussed on documentation?
14:59:15 <joanie> if we do it before or after boston summit, it wouldn't conflict with Canadian Thanksgiving
14:59:16 * clown sorry...
14:59:22 <API> well, it seems that Im not really ambitious and was thinking on the focus stuff thing
14:59:30 <jhernandez> btw, just remembered that I have a few questions regarding gnome a11y and browsers (mostly related to screen enhacements such as high-contrast or font-size)
14:59:50 <API> jhernandez, just in end-of-meeting time!
14:59:51 * clown hears GPII sneaking into this room.
14:59:55 <jhernandez> API: heh
15:00:05 <API> you had 10 minutes of misc time, and you wasted it! congrats!
15:00:06 <jhernandez> and clown, yes
15:00:09 <API> cough cough
15:00:10 <API> I mean
15:00:11 <jhernandez> :)
15:00:12 <joanie> my fault
15:00:16 <joanie> but we should do this
15:00:19 <API> few questions
15:00:20 <joanie> seriously
15:00:27 <API> sounds something that needs more that just some minutes
15:00:29 <jhernandez> nope, I was going to ask if I should mail you in private
15:00:31 <jhernandez> yes
15:00:44 <API> could those questions being formualted as emails to the dev ml?
15:00:55 <clown> a11y-devel?
15:01:02 <jhernandez> not sure enough
15:01:23 <API> not sure enough?
15:01:56 <jhernandez> yeah, it's not a11y-dev, it's more related to how 3rd party apps (such as a browser) make use of our a11y features
15:02:10 <jhernandez> ie: the high contrast
15:02:32 <jhernandez> as an example of it, firefox changes when HighContrast is on, but not its content
15:02:52 <jhernandez> IIRC same for epiphany
15:03:09 <clown> because the content is defined by author and/or user style sheets.
15:03:15 <jhernandez> clown: yes!
15:03:29 <jhernandez> it's a firefox problem ofc
15:03:34 <clown> you want GPII to assert a "user" style sheet for the browser to use.
15:03:40 <API> well, gnome-accessibility-devel has plenty of "non-exactly" gnome information now and then
15:03:59 <clown> there is a related FF bugzilla about that.
15:03:59 <API> so I still don't know why it would be a problem to send the questions there
15:04:01 * clown looks.
15:04:24 <jhernandez> ie: https://bugzilla.mozilla.org/show_bug.cgi?id=239914
15:04:24 <Services> 04Bug 239914: normal, P3, Future, nobody, NEW , [Linux] When high contrast theme used, automatically ignore author's colors
15:04:41 <jhernandez> API: ok
15:05:01 <clown> jhernandez:  there's another one, raised a while back by Bryan (sp?)
15:05:08 * clown still looking for it
15:05:17 <API> so as everybody agreed
15:05:21 <API> and is happy and joy joy
15:05:24 <API> and we are over time
15:05:30 <API> Im going to close the meeting
15:05:31 <jhernandez> API: thanks
15:05:35 <API> someone against?
15:05:42 <jhernandez> and sorry for bringing this topic at the very last minute
15:05:48 <jhernandez> :P
15:05:50 <API> np
15:05:59 <API> I really think that did the same a lot of tome
15:06:06 <API> so closing meeting
15:06:09 <API> #endmeeting