Attachment '20131010_log.txt'
Download 1 16:01:18 <API> #startmeeting
2 16:01:18 <tota11y> Meeting started Thu Oct 10 16:01:18 2013 CET. The chair is API. Information about MeetBot at http://wiki.debian.org/MeetBot.
3 16:01:18 <tota11y> Useful Commands: #action #agreed #help #info #idea #link #topic.
4 16:01:28 <API> #topic 5-min of margin
5 16:06:35 <API> an now meeting time
6 16:06:42 <API> +5
7 16:07:05 <API> #topic "Boston" Summit
8 16:07:27 <API> #info API is configuring the laptop in order to do the pre-event homework as agreed
9 16:07:55 <API> #info talking with people led to conclude that ubuntu was not the best to try wayland-gnome, using f20 now
10 16:08:03 <joanie> yay
11 16:08:23 <API> #info building all the stack right now, so its a wip right now
12 16:08:24 <API> done
13 16:08:32 <API> I think that joanie have more to say about this
14 16:08:39 <joanie> #info Joanie did much of her pre-event homework.
15 16:08:47 <joanie> one sec
16 16:09:02 <joanie> #info Joanie is adding her findings to https://wiki.gnome.org/Accessibility/Hackfests/Montreal2013
17 16:09:20 <joanie> #info Joanie also updated the Application Status Page: https://wiki.gnome.org/Wayland/Applications
18 16:09:54 <joanie> #info Orca was sometimes crashing upon launch with the Wayland backend. Joanie has since fixed that with a workaround, and will have it totally fixed by today's end (hopefully).
19 16:10:07 <joanie> #info Accerciser reliably crashes upon launch with the Wayland backend.
20 16:10:26 <joanie> #info Joanie filed bug 709449 for that.
21 16:10:26 <tota11y> 04Bug https://bugzilla.gnome.org/show_bug.cgi?id=709449 major, Normal, ---, accerciser-devel, UNCONFIRMED, Accerciser segfaults when run with the wayland backend
22 16:10:41 <joanie> #info But accerciser has some other problems and has not be worked on in quite some time.
23 16:10:59 <joanie> #action Joanie will ping Javi about his plans w.r.t. his module.
24 16:11:06 <joanie> I think that's it.
25 16:11:14 <joanie> mgorse: you have a wayland environment?
26 16:11:58 <mgorse> I'm not sure. I ran jhbuild, and the Xorg that it built doesn't want to start.
27 16:12:13 <joanie> ugh
28 16:12:20 <mgorse> at least not if I run startx
29 16:12:33 <mgorse> Maybe there are instructions somewhere that I wasn't following
30 16:12:36 <joanie> I followed the instructions on the Wayland trying it page
31 16:12:47 <mgorse> okay; I need to look for that
32 16:12:55 <joanie> there's a link on our team's page
33 16:12:58 <joanie> but I'll get it
34 16:13:08 <joanie> https://wiki.gnome.org/Wayland/TryingIt
35 16:13:14 <joanie> mgorse: that is what I followed
36 16:13:38 <joanie> mgorse: you might also want to read my comments on our team's page
37 16:13:46 <joanie> https://wiki.gnome.org/Accessibility/Hackfests/Montreal2013
38 16:13:49 <joanie> in particular
39 16:13:57 <joanie> As is already known, global keyboard shortcuts do not work when you are in an app using the Wayland backend. If using a mouse to launch or quit applications is problematic for you, consider adding a non-Wayland terminal application to your session's start-up. From within that terminal, you not only can launch apps via the terminal (i.e. to use the Wayland backend), but you can also use gnome-shell's global keyboard shortcuts.
40 16:14:44 <API> joanie, we briefly talked about that the other day
41 16:14:46 <joanie> oh another key thing.... the shortcuts to get back to virtual consoles are broken
42 16:14:52 <API> our conclusion was "talk with gcampax"
43 16:14:59 <joanie> API: I know
44 16:15:01 <API> should I send a email to him before the event?
45 16:15:12 <joanie> my point is, once mgorse gets a working environment
46 16:15:12 <API> I mean, instead of just waiting for a face-to-face talk
47 16:15:15 <joanie> it won't work for him
48 16:15:19 <joanie> if he doesn't do the terminal trick
49 16:15:27 <joanie> because you need the mouse otherwise
50 16:16:04 <API> k
51 16:17:03 <joanie> anyhoo, the terminal trick works for me because the touchpad support is poor enough that I cannot reliably click on stuff
52 16:17:09 <joanie> due to hand tremors
53 16:17:24 <joanie> but I think we're digressing (sorry API)
54 16:17:33 <API> no problem
55 16:17:36 <API> in any case
56 16:17:45 <API> there isn't a lot of people here ;)
57 16:17:57 <API> btw, I had an action item about sending the agenda
58 16:17:59 <joanie> are we otherwise set for Montreal?
59 16:18:03 <API> I will do that after the meeting
60 16:18:27 <API> the delay will not be really important, because afaik, most of the people going to the summit were at the meeting
61 16:18:33 <API> last week meeting I mean
62 16:18:38 * joanie nods
63 16:18:46 <API> apart for that, no, I don't have anything else to say about the summit
64 16:19:00 <joanie> me neither
65 16:19:02 <API> so unless someone wants to add something, I will move to next point
66 16:19:50 <API> #topic GNOME 3.12 Planning
67 16:20:07 <API> well, I think that there isn't any big changes here, compared to last week
68 16:20:20 <API> probably just add the accerciser thing that joanie mentioned before
69 16:20:29 <mgorse> API: Thanks for your reply on g-a-devel. I need to re-read it and reply to it.
70 16:20:53 <joanie> yeah, that's a big 3.12er potentially
71 16:20:54 <joanie> :)
72 16:20:55 <API> mgorse, no problem,
73 16:21:05 <joanie> I have another one potentially
74 16:21:32 <API> sorry if you replied my reply, but I didn't have a proper environemnet to read emails
75 16:21:33 * API looking
76 16:21:35 <joanie> API from a while back you had an action item w.r.t. pinging the mousetweaks devs re wayland
77 16:21:47 <API> hmm, yeah
78 16:21:56 <API> but I also understood that magdalen was planning to do the same
79 16:21:59 <joanie> and if they need to do something and it isn't done
80 16:22:07 <joanie> then we'll have a regression
81 16:22:24 <joanie> Magdalen is a full-time student
82 16:22:28 <joanie> and novice developer
83 16:22:47 <joanie> I think that if we care about Mousetweaks we need to take the lead on this
84 16:24:51 <API> well, but she sent that email
85 16:25:03 <joanie> what email
86 16:25:03 <joanie> ?
87 16:25:12 <API> I didn't want to flood Francesco inbox about the same from different people
88 16:25:14 <joanie> anyhoo, if you would prefer, I can remove your action item
89 16:25:35 <API> magpie sent a forward to the list
90 16:25:37 <API> Subject:
91 16:25:37 <API> Re: Mousetweaks & A11y meeting regrets
92 16:25:44 <joanie> oh yeah
93 16:25:44 <API> unfourtunately she mixed topics
94 16:26:10 <joanie> ok, I'll remove your action item then
95 16:26:13 <joanie> sorry for the noise
96 16:27:44 <API> np
97 16:28:00 <API> btw, I also have as pending the ai of gtk+2 plans
98 16:28:06 <API> will do that asap
99 16:28:12 <API> anything else on this topic?
100 16:28:28 <joanie> not from me
101 16:29:27 <API> #topic W3C updates
102 16:29:32 <API> clown is not here
103 16:29:34 <API> joanie, ?
104 16:29:39 <joanie> nothing from me
105 16:30:50 <API> ok
106 16:30:55 <API> #topic Marketing
107 16:31:01 <API> jjmarin is not here
108 16:31:13 <API> #topic Miscellaneous time
109 16:31:26 <API> so anyone present want to mention something that was not scheduled?
110 16:31:31 <API> this time it doesn't need to be short
111 16:31:36 <joanie> heh
112 16:31:40 <API> because the meeting was going at full speed
113 16:31:50 <joanie> wnck in pygobject is apparently quite broken
114 16:31:57 <joanie> today's discovery
115 16:32:25 <API> fwiw, last time I tried to use accerciser
116 16:32:28 <joanie> I think accerciser uses it too
117 16:32:28 <API> failed due wnck
118 16:32:47 <API> so I0'm wondering if accerciser breakage is related with wnck
119 16:32:55 <joanie> my guess is yes
120 16:33:21 <API> joanie, do you know if wnck is used by a lot of projects?
121 16:33:27 <joanie> dunno
122 16:33:32 <API> or pygobject bindings at least
123 16:33:35 <joanie> and the question is: a lot of python projects
124 16:33:37 <joanie> exactly
125 16:33:48 <joanie> my immediate concern is Orca
126 16:34:09 <joanie> which can (and apparently has been) using Atspi's component as a fallback
127 16:34:15 <joanie> for what it tried to accomplish via wnck
128 16:34:23 <joanie> so I'm removing the wnck code from orca now
129 16:34:33 <joanie> if Orca keeps on keepin' on, then my problem is solved
130 16:34:36 <joanie> then I'll look at accerciser
131 16:34:47 <joanie> because the amount of breakage is getting severe
132 16:34:58 <joanie> anyhoo.....
133 16:35:07 <joanie> that's all my misc time stuff I think
134 16:35:08 <joanie> :)
135 16:35:18 <API> joanie, could you summarize it with pretty #info?
136 16:35:23 <joanie> bah
137 16:35:27 * joanie grumbles
138 16:35:29 <joanie> :P
139 16:35:30 <joanie> (sure)
140 16:36:12 <joanie> #info Joanie discovered that with pygobject, Wnck seems a bit broken. e.g. Wnck.Screen.get_default() in Wayland causes segfaults.
141 16:36:41 <joanie> #info And in the non-wayland world wnck_screen_get_windows_stacked() returns an empty list
142 16:36:57 <joanie> #info Joanie discovered that Orca doesn't actually need the wnck code in question.
143 16:37:16 <joanie> #info Joanie suspects this might be why Accerciser is segfaulting on launch with the wayland backend.
144 16:37:39 <joanie> #action Joanie will investigate this for Accerciser and (hopefully) provide a patch so we can use Accerciser to test.
145 16:37:45 <joanie> how's that fearless one?
146 16:38:09 <API> good, thanks
147 16:38:17 <API> so, anything else for misc time?
148 16:38:39 <joanie> msanchez: can talk about webkitgtk atktext wtf perhaps? :)
149 16:39:49 <joanie> or not
150 16:41:39 <joanie> meeting over then?
151 16:41:52 * joanie wants to resume her wnckectomy
152 16:41:53 <joanie> :)
153 16:42:34 <joanie> or is that wnckotomy
154 16:42:40 <joanie> I want to remove wnck anyway
155 16:43:41 <msanchez> not sure what to mention
156 16:43:48 <joanie> up to you :)
157 16:43:59 <msanchez> well, tehre's much stuff I believe
158 16:44:00 <joanie> sounds like a refactor might be coming soon?
159 16:44:01 <msanchez> but in a nutshell
160 16:44:22 <msanchez> I'm taking a deep look, step by step, to the implementation of atk_text_get_text in WebKitGTK+
161 16:44:32 <msanchez> it looks to me like there is much room for improvement in there
162 16:44:47 <msanchez> I doubt it will be a "destroy everything and start over" thing
163 16:45:00 * joanie hides the matches and gas can
164 16:45:17 <API> msanchez, fwiw, you also mentioned yesterday something about tests in webkit2
165 16:45:30 <msanchez> I'm more pro doing iterative steps, and at the moment that means I've analyzed the file and identified some API methods that need checking... being this one the most important one
166 16:45:36 <msanchez> but I don't have any results yet
167 16:45:44 <msanchez> API: ah, true
168 16:45:51 <API> as I was busy I quickly just answered that could be a task for next webkitgtk ahckfest
169 16:45:53 <API> *hackfest
170 16:46:05 <msanchez> my other idea to work on a11y now is making sure we move away from testatk.c at some point, which only runs in WK1
171 16:46:35 <msanchez> we need to at least migrate those tests to WebKit2 infrastructure, which I believe will mean to test by means of atspi, not atk
172 16:46:49 <msanchez> besides some other issues that we will probably find along the way
173 16:46:54 <msanchez> but again, it's too early
174 16:47:12 <API> msanchez, so right now
175 16:47:14 <API> on webkit2
176 16:47:20 <API> the only atk related tests
177 16:47:22 <msanchez> This week I landed the last patch I had pending on review and I'm now in the phase of fixing the AtkText mess
178 16:47:28 <API> are related with checking that the plug is working?
179 16:47:30 <msanchez> no ATK tests in WebKit2
180 16:47:48 <msanchez> only one AtSpi test to check that the hierarchies are connected
181 16:47:50 <msanchez> that's right
182 16:47:53 <msanchez> + the layout tests
183 16:48:12 <msanchez> and the fact that the bots still run WebKit1 unit tests (which check other stuff from ATK, not checked by the layout tests)
184 16:48:37 <API> ok, so I will try to summarize this:
185 16:48:55 <API> #info as an early idea of what we can do for webkitgtk hackfest, some ideas are appearing
186 16:49:06 <API> #info a) Review atk_text_get_text implementation
187 16:49:30 <API> #info b) Look about moving atk tests to Webkit2 tests, as right now it only have one at-spi test related with the hierarchy
188 16:49:38 <API> msanchez, is that good enough for a summary?
189 16:49:38 <msanchez> I can propose a third one
190 16:50:06 <API> shot
191 16:50:24 <msanchez> what about going through the mappings in http://www.w3.org/TR/wai-aria-implementation/#mapping_state-property and check which ones are correct (so can be safely implemented now already) and which ones are not?
192 16:50:30 <msanchez> shoot
193 16:51:05 <API> joanie, what do you think about the third one?
194 16:51:21 <joanie> sure
195 16:51:29 <joanie> I hope to have reviewed all the mappings by then
196 16:51:37 <joanie> and proposed API changes (should they be needed)
197 16:51:44 <msanchez> olrai
198 16:51:52 <joanie> olrai?
199 16:52:01 <msanchez> all right
200 16:52:03 <joanie> ah
201 16:52:04 <API> #info msanchez also propose review http://www.w3.org/TR/wai-aria-implementation/#mapping_state-property in order to check correctness and which needs new atk API
202 16:52:12 <API> #info disclaimer: all that are early ideas
203 16:52:17 <API> and having said so
204 16:52:23 <API> I think that we can finish the meeting
205 16:52:52 <API> #endmeeting
Attached Files
To refer to attachments on a page, use attachment:filename, as shown below in the list of files. Do NOT use the URL of the [get] link, since this is subject to change and can break easily.You are not allowed to attach a file to this page.