14:36:50 <API> #startmeeting
14:36:50 <tota11y> Meeting started Thu Jun 23 14:36:50 2011 UTC.  The chair is API. Information about MeetBot at http://wiki.debian.org/MeetBot.
14:36:50 <tota11y> Useful Commands: #action #agreed #help #info #idea #link #topic.
14:37:06 <API> so as we don't have subtopics here
14:37:07 <API> AFAIK
14:37:22 <API> #topic 1.1 Testing. Better way to keep on top of changes? (e.g.gnome-contacts)
14:37:36 <API> joanie, I think that you added this point
14:37:40 * joanie nods
14:38:20 <joanie> #info The other day Matthias happened to mention in #a11y that the gnome-contacts contact list had been redone with new custom widgets and that we should test it.
14:38:53 <joanie> #info Joanie thinks we need a way better system than relying upon the kindness of Matthias to give us such head-up pings.
14:39:01 <joanie> So... I dunno what the right answer is
14:39:02 * API I don't use gnome-contacts at all ...
14:39:07 <joanie> but I think we should find one
14:39:18 <joanie> Well, I'm still fighting with getting it to build
14:39:21 <joanie> and I will test it
14:39:25 <API> well, as usually this lead to some kind of testing framework
14:39:27 * clown is gnome-contacts an app?
14:39:29 <joanie> but it is more the general issue
14:39:35 <API> and a lot of examples using it
14:39:43 <API> I mean
14:39:52 <API> and a lot of use cases filling it
14:39:54 <joanie> a11y breaks and we don't know it and the module maintainers don't know it
14:39:58 <API> it seems that bnitz is not here
14:40:08 <API> well, ideally
14:40:29 <API> as I said, it would be good to have something like a automatic a11y tests
14:40:34 <API> and run them
14:40:35 * joanie nods
14:40:36 <API> http://build.gnome.org/
14:40:41 <joanie> that's the long range idea
14:40:46 <API> and if a error happens, a mail is sent to a list
14:40:58 * API still have on todo the mail configuration
14:41:06 <joanie> but in the short run...
14:41:15 <joanie> What do we do about this general issue?
14:41:18 <API> #info long term solution: a11y test suit, executed on build.gnome.org, mail sent on any error
14:41:32 <joanie> Can we delegate this to someone or someones?
14:41:49 <joanie> make a list of core gnome apps
14:41:56 <prlw1> I think bnitz was going to even associate the error with the commit, so the developer responsible for the commit gets the email.
14:42:02 <joanie> and test them for a11y -- functional testing until we have something automated in place
14:43:09 <joanie> But do we know where bnitz is w.r.t. these tools?
14:43:24 <joanie> Last time I checked, he was saying Oracle doesn't ship GNOME 3
14:43:35 <joanie> so I'm not sure that he will be our best resource on this front
14:43:40 <joanie> though I could be wrong
14:43:55 <API> joanie, about that delegation thing
14:43:58 <API> what do you mean?
14:44:10 <API> are you talking about manual testing?
14:44:22 <joanie> I mean that someone (other than joanie and api) do manual testing of the *core* gnome apps
14:44:29 <joanie> until we have a better system in place
14:44:52 <joanie> where gnome == gnome 3 ;-)
14:44:56 <joanie> and not gnome 2
14:45:18 <joanie> And Aline isn't here either
14:45:25 <API> well, as usual I only see an option here
14:45:29 <joanie> I was thinking this might be a neat task for her
14:45:42 <API> and its the current state,
14:45:52 <API> developers detecting heavy changes
14:46:00 <API> and notify gnome a11y lists asking for testing
14:46:13 <joanie> actually that just gave me an idea
14:46:24 <API> I'm not sure if asking people to do smoketesting of all the plattform each X days viable
14:46:40 <joanie> we could subscribe and better monitor the commit ml
14:46:55 * API btw, still pending a platform a11y smoketesting after vuntz suggestion
14:47:23 * prlw1 thinks that bnitz has solved what you are thinking about
14:47:32 <joanie> really?
14:47:44 <joanie> prlw1: mind elaborating with an #info
14:47:50 <joanie> so I can follow up with him later
14:48:29 <prlw1> or maybe just get it from the horses mouth?
14:48:56 <joanie> What did he solve?
14:50:17 <joanie> or to put it another way prlw1, what are you talking about when you said 'what you are thinking about'?
14:50:21 <joanie> who is you
14:50:28 <joanie> and several of us shared thoughts
14:50:35 <prlw1> nevermind
14:50:49 <API> anyway, we have been in this point about 20 minutes
14:50:54 * joanie nods
14:50:56 <joanie> sorry
14:50:57 <API> so a conclusion is required, taking into account
14:51:03 <API> that we have other points
14:51:10 <API> probably a "ask bnitz"
14:51:35 <prlw1> correct
14:51:43 <API> #info Joanie ask if it would be possible if someone (other than joanie and api) do manual testing on the core gnome apps
14:51:57 <API> #info API says that we would require to do manual testing for a while
14:52:11 <API> #info Patrick comments that Brian Nitz have something solving those issues
14:52:34 <API> #info Brian is not here, so this is postponed, further investigation required
14:52:37 <API> anything else?
14:52:45 <joanie> nah
14:52:49 <API> ok
14:52:50 <API> so
14:53:10 <API> #topic 1.2 Testing. Gtk A11y dump tool
14:53:16 <API> http://git.gnome.org/browse/gtk+/commit/?id=b2cfbd0de9e4306ff65a8f9171ecc795b055d16a
14:53:29 <API> it seem that Company started a kind of function a11y testing
14:53:38 <API> joanie, do you have more details?
14:53:50 * joanie defers to Company if he is around
14:53:52 * API as Company already said that he will be busy in other meeting
14:54:05 <Company> not started yet
14:54:21 <Company> so i'll just talk here until it happens
14:54:45 <Company> so I did this tool that takes a ui file created in glade
14:55:07 <Company> grabs its accessible and dumps everything from it into a file
14:55:24 <Company> see the link from API for more details
14:55:29 * clown sounds nifty.
14:55:38 <clown> Company:  is the output plain text?
14:55:47 <Company> clown: lemme find links to an example
14:56:01 <Company> so a glade file like this: http://git.gnome.org/browse/gtk+/tree/tests/a11y/hello-world.ui?id=a11y
14:56:17 <Company> which is just a button saying "Hello World"
14:56:20 <clown> right
14:56:31 <Company> generates something like this: http://git.gnome.org/browse/gtk+/tree/tests/a11y/hello-world.txt?id=a11y
14:57:02 <Company> the idea for this kind of test is to do 3 things:
14:57:12 <Company> 1) regression-test
14:57:32 <Company> we want to notice if we messed something up while refactoring etc
14:57:37 <Company> 2) crash-test
14:58:00 <Company> if we can make a weird combination of widgets crash, just make a glade file that exposes it, and put it there
14:58:02 <API> well, orca has already several regression tests, so
14:58:14 <API> will this one replace orca ones?
14:58:23 <API> or can we have both?
14:58:31 <Company> i have no idea what orca tests
14:58:44 <API> joanie, ^
14:58:48 <Company> this is strictly for GTK's implementation of a11y
14:58:57 <API> but for example, it has a lot of firefox tests
14:59:00 <joanie> we can have both
14:59:04 <API> and I think that it have a lot of gtk tests
14:59:15 <joanie> Orca is designed to be sure that changes to Orca don't break access to, say, gtk
14:59:26 <joanie> that's independent of what Company is doing/talking about
14:59:53 <Company> anyway, point 3 and the most important one:
14:59:59 <joanie> s/Orca is/Orca tests are/
15:00:13 <Company> 3) specifying the intended behavior for GTK
15:00:29 <Company> so we want to create a ui file and the text file manually
15:00:43 <Company> and then make sure that the gtk implementation actually conforms to it
15:01:13 <Company> to take the treeview atktable example we discussed an hour ago: if we say we want row items in treeviews
15:01:24 <Company> we'll create a glade file with a treeview
15:01:49 <Company> and a text file that says "here are row objects, they have this role, implement these states and interfaces etc"
15:02:00 <API> Company, it is gtk specific? it could be easy to have something similar on clutter/st? I see the problem on those glade files
15:02:07 <API> and with this specific example of tables
15:02:12 <Company> and then we'll go and change the GailTreeView implementation
15:02:21 <API> it would be good to have something similar also on webkitgtk or mozilla
15:02:42 <Company> and this is what I told joanie about is what i need from a11y people
15:03:33 <Company> i need an idea of how you would like the output of gail to look for certain widgets/ui files
15:03:39 <joanie> This might have some implication for our ATK prioritizations
15:03:40 <Company> so we can make that happen
15:03:51 <joanie> for instance, today there is no such thing as ATK_ROLE_TABLE_ROW. But there will be
15:04:09 <API> Company, so you will provide glade files and a11y guys should say if the output is correct or not?
15:04:51 <Company> API: either that, or you guys can create glade files and say "it should look like this"
15:05:34 <API> aha
15:05:47 <Company> API: and about being gtk specific - the way the test is set up is very gtk specific (glade files etc), but the actual dumping of the accessible is not so much
15:06:01 <API> yes, this is the reason I asked
15:06:10 <Company> API: there are some things we don't dump (like the stuff that says "omitted") or component bounds
15:06:13 <API> as it would be good to have this in other places
15:06:24 <Company> because those things are very specific to the gtk theme you are using etc
15:06:52 <clown> Company, the dump shows ATK info, right?  As opposed to AT-SPI info.
15:07:01 <Company> clown: yes
15:07:22 <clown> Company:  ok, thanks.
15:07:26 <Company> clown: the dump currently just takes the AtkObject of the toplevel window and dumps it and its children
15:08:06 <Company> clown: i might in the future write a version that goes over dbus, and dumps the AT-SPI objects, too, just to see if they are equivalent
15:08:16 <clown> cool...
15:08:43 <clown> although that's kind of what accerciser does (in a gui).
15:08:52 <Company> yeah
15:09:19 <API> well, not sure if anything else is missing in this topic
15:09:22 <Company> i currentlt do not care a lot about the at-spi layer, but more about the GAIL layer
15:09:29 <API> makes sense
15:09:30 <Company> so that's what these tests focus on
15:09:41 <Company> but improving on that would be great
15:09:58 <Company> most of all: please write tests for us so we can fix things
15:10:09 <clown> sure.  it's targetted at ATK developers as way for them to make sure what they did is coming out right wrt ATK.
15:10:48 <API> and this goes again to "people are really busy" these days ..
15:10:58 <joanie> Company: what is the best way to get you tests? bugzilla patches?
15:11:30 <joanie> API yeah, but I really want to see if we can pitch in here. I'm hoping next week to dive in
15:11:37 <Company> joanie: as that ison a branch anyway, so just ping on irc + fpaste is fine usually
15:11:47 <joanie> k
15:11:55 <Company> joanie: i probably wouldn't even mind if you just committed them :)
15:12:14 <joanie> :-) After you see one or two and we're on the same page, absolutely!
15:12:47 <Company> and i'm off to the rh meeting now
15:12:55 <joanie> thanks Company!!
15:13:06 <API> well, this was  long point
15:13:11 <joanie> but a good one
15:13:15 <API> so it doesn't make sense to make #info points
15:13:17 <API> so
15:13:19 <API> joanie
15:13:25 <API> what about a action item
15:13:30 <API> ?
15:13:33 <joanie> :-)
15:13:40 <API> #action joanie will summarize 1.2 topic
15:13:50 <API> well, lets try to at least finish this point
15:14:17 <API> #topic 1.3 Testing. Testing Distro update?
15:14:26 <API> well, jhernandez is not here
15:14:28 <API> clown, ?
15:14:38 <API> I saw a lot of activity in your bugs these days
15:14:47 <API> any relation with the testing distro
15:14:53 <clown> indeed.
15:15:00 <clown> with respect to the testing distro...
15:15:07 * API in theory the last one of our custom one until we "merge" with fcrozat one
15:15:28 <clown> #info I supplied a patch to Orca to fix the DBus communication with gs-mag ...
15:15:44 <clown> #info joanie reviewed and made some suggestions ...
15:16:01 <clown> #info I'm modifying in light of those suggestions.
15:16:25 <clown> #info and hope to have a new patch by end of today (but probably end of week [tomorrow]).
15:16:33 <clown> done, anything to add joanie?
15:16:59 <joanie> #info Joanie will roll the new tarball for the testing distro
15:17:15 <joanie> (once the new patch is committed)
15:17:57 <API> ok, so we need to talk about anything else in this topic?
15:18:14 <clown> not from me.
15:18:16 <joanie> nope
15:18:48 <API> ok
15:18:55 <API> #topic GNOME 3.2 progress:
15:19:44 <API> #info API strill trying to get some time to work on current St bugs
15:19:59 <API> #info API, not a lot of new here
15:20:01 <API> so people
15:20:03 <API> your turn
15:20:06 <API> ;)
15:20:29 <mgorse> #info mgorse pushed a patch to make atk-bridge not listen for events when no ATs are running
15:20:31 <prlw1> AFAIK dasher works on gnome 3
15:20:48 <prlw1> please let me know if it breaks for you...
15:21:07 <joanie> prlw1: So you've tested it in GNOME 3?
15:21:12 <prlw1> congrats to mgorse!
15:21:25 <joanie> +1 to congrats to mgorse
15:21:59 <API> prlw1, tested on gnome-shell?
15:22:00 <joanie> I wonder how much closer that gets us to the goal of can enable a11y by default?
15:22:17 <mgorse> Not sure. I have no idea if it will noticably improve performance, but it (hopefully) won't hurt
15:22:40 <joanie> What I mean is that we've been told that we cannot enable a11y by default
15:22:50 <joanie> I'm wondering if this means we can
15:23:01 <joanie> i.e. because nothing bad will happen to non AT users
15:23:06 <joanie> API thoughts?
15:23:07 <API> well, we can add a action
15:23:17 <API> investigate if last change on atk-bridge
15:23:24 <API> help on setting a11y by default
15:23:37 <API> s/setting/setting on
15:24:00 <API> well, no sure who should test it
15:24:07 <API> but lets at least add the action
15:24:32 <API> #action check how last mgorse change improve the situation of the performance at the desktop with a11y on and no AT listening
15:25:01 <joanie> mgorse: And if we can get this on by default, you will be the team hero
15:25:06 <joanie> nice job!
15:25:18 <API> well, 5 minutes till the end
15:25:23 <API> anything else in this topic?
15:25:34 <mgorse> Do we know if there are specific performance issues that people are having / what they are? Opening large trees might be one (in Nautilus, for instance; no idea if that's still been an issue with AT-SPI2)
15:25:41 <API> msanchez, is right now
15:25:53 <API> talking so not sure if he can say anything ;)
15:26:02 <API> mgorse, yes, anything related with nautilius
15:26:08 <API> and gtktreeview are required to be tested
15:26:28 <API> so lets set that action
15:26:39 <API> and as no one else is saying anything
15:26:43 <API> 4 minutes till the end
15:26:47 <API> #topic miscellaneous time
15:27:05 <API> do you want so say anything sort not included on the meeting agenda?
15:27:15 <API> s/sort/short
15:28:26 <API> ...
15:28:42 * clown the silence is deafening.
15:29:03 * API is the sound ... of the silence ...
15:29:27 * clown one hand clapping?
15:30:08 <API> ...
15:30:10 * joanie refrains from commenting about people hearing without listening
15:30:11 <API> well
15:30:25 <API> no one used miscellanous time
15:30:26 * clown that's deep
15:30:27 <API> so
15:30:30 <API> meeting over
15:30:33 <API> #endmeeting