Attachment '20110210_log.txt'
Download 1 (15:05:09) API: well, we gave 5 gentle minutes
2 (15:05:16) API: I think that we can start the meeting
3 (15:05:21) ***clown waves
4 (15:05:25) API: agenda here:
5 (15:05:27) API: http://live.gnome.org/Accessibility/Meetings
6 (15:05:30) ***slee thumbs up
7 (15:05:44) API: first item
8 (15:05:47) API: 641869
9 (15:05:57) API: https://bugzilla.gnome.org/show_bug.cgi?id=641869
10 (15:06:06) ***clown reads
11 (15:06:09) API: mgorse, I guess that this is your point
12 (15:06:41) ***slee still waiting for bug to load
13 (15:07:30) mgorse: ok
14 (15:07:52) mgorse: So in at-spi2 we've had an --enable-relocate option, to allow AT-SPI and AT-SPI2 to co-exist
15 (15:08:09) mgorse: and the version of at-spi could be switched with a gconf option
16 (15:08:35) mgorse: and pyatspi would load the alternate version of pyatspi of the relocated version was in effect, and GTK_PATH would be set
17 (15:08:54) fregl [~gladhorn@eckhart-woerner.de] entered the room.
18 (15:09:05) mgorse: Apparently setting GTK_PATH was causing problems. It may have been some transitional issue with upgrading a package in Fedora. I'm not sure.
19 (15:09:34) mgorse: Either way, it was intended to be there temporarily to help people test AT-SPI2, so I propose to remove it if no one objects
20 (15:09:52) joanie: +1
21 (15:09:56) mgorse: It would allow some gconf-related code to be removed, and I think it would generally make things less confusing
22 (15:09:57) fer: +1
23 (15:10:09) mgorse: I've talked with TheMuso and sshaw, and they seem okay with it
24 (15:10:11) API: mgorse, that would be remove the relocate stuff from both, right?
25 (15:10:17) API: I mean, from at-spi and at-spi2 too
26 (15:10:27) fer: yeah, and I guess that most people did for testing: at-spi1 on /usr from their distro, and at-spi2 on /opt from sources
27 (15:10:45) mgorse: API: That was my next question really. Do people think it is important to preserve the --enable-relocate behavior in AT-SPI?
28 (15:10:56) API: hmm
29 (15:11:00) API: well, as you said
30 (15:11:06) API: most of the people
31 (15:11:07) mgorse: AT-SPI-CORBA I mean
32 (15:11:09) API: AFAIK
33 (15:11:14) API: were not using the relocate thing
34 (15:11:22) API: (not sure if this is a pity or not)
35 (15:11:35) API: IMHO, it would be better to remove it from both
36 (15:11:45) API: and as you said, make things simpler
37 (15:12:43) mgorse: okay. I'll plan on just removing from both then, since no one is objecting / feeling like it's important to keep it around
38 (15:12:49) slee: +1 having old options around is only confusing - especially for newcomers
39 (15:13:27) fer: I know that F14 installed both at-spi and at-spi2, but I don't really know why
40 (15:13:31) mgorse: so the packages will just conflict with each other
41 (15:13:39) API: fer, really ?
42 (15:13:48) fer: yeah, pretty weird
43 (15:13:53) API: but I guess that one compiled with relocate, right?
44 (15:13:55) ***joanie chuckles
45 (15:13:58) fer: yes
46 (15:14:00) mgorse: unless we made --enable-relocate always-on with AT-SPI (which would mean preserving some code in AT-SPI2 to ensure that it's disabled when necessary)
47 (15:14:02) API: although thiis is a good point
48 (15:14:14) fer: there was something depending on at-spi2 package, let me check it
49 (15:14:14) API: removing relocate would make installing both incompatible
50 (15:14:35) fer: but I don't know the rationale for having at-spi2 on GNOME2 on F14 :)
51 (15:14:38) fer: mclassen should know
52 (15:15:35) fer: oh, at-spi2-core in F14 is a dependency for orca and pyatspi :)
53 (15:15:58) fer: maybe it was because they picked up a recent version of pyatspi doing the gconf relocation check?
54 (15:17:33) fer: ok
55 (15:17:58) fer: <mclasen> because we were aiming to have gnome3 in f14
56 (15:17:58) fer: <mclasen> and then gnome3 got delayed and we had to quickly shift gears
57 (15:17:58) fer: <fer> so just a preview, ok
58 (15:17:58) fer: * _ke se ha marchado (umount /mnt/me)
59 (15:17:58) fer: <mclasen> and leave behind a few useless packages...
60 (15:17:58) fer: <fer> is it safe for F15 to remove relocation and at-spi1, right?
61 (15:18:01) fer: <mclasen> yes, should be
62 (15:18:30) mgorse: ok
63 (15:19:50) API: well, so concluding ...
64 (15:20:04) mgorse: so I'll plan on just removing it. Seems not worth keeping around if most people weren't using it anyhow
65 (15:20:23) API: well, and in some place it would be required to say that both are incompatible
66 (15:20:31) API: both at the same time I mean
67 (15:20:42) API: unless you still plan to maintain relocate stuff on at-spi2
68 (15:20:43) API: sorry
69 (15:20:46) API: at-spi1
70 (15:21:21) fer: we can just add it to the NEWS file
71 (15:22:01) bnitz: fer:Just a context question here, F14, F15? Are you referring to builds?
72 (15:22:14) fer: like "at-spi2-*" is incompatible with at-spi, so packagers can notice it and mark their at-spi2 packages with Obsoletes: at-spi
73 (15:22:24) fer: Fedora 14, Fedora 15
74 (15:22:26) fer: releases
75 (15:22:28) clown: bnitz, I think they are release versions of Fedora
76 (15:22:35) bnitz: That would have been my second guess, thanks!
77 (15:22:42) mgorse: 15 is the up-coming release
78 (15:23:32) API: well, anything else in this point?
79 (15:26:34) API: I guess that the silence means "no" ;)
80 (15:26:41) API: lets move to the next point
81 (15:26:43) bnitz: Listinging to John Cage 4'33.
82 (15:26:48) ***API looking
83 (15:26:59) API: FOSDEM summary (What was cool? What did those not in attendance miss?)
84 (15:27:08) API: well, I was on FOSDEM
85 (15:27:17) slee: bnitz: that's a bit to repetitive for me
86 (15:27:20) API: fer planned to go, but I guess that he didn't go
87 (15:27:38) fer: yup, had to do some stuff here in madrid :(
88 (15:27:41) ***slee is keen to hear about FOSDEM
89 (15:27:46) ***joanie too
90 (15:27:59) API: well, I guess that it is my point
91 (15:28:00) bnitz: did anyone go?
92 (15:28:03) API: me
93 (15:28:08) API: <API> well, I was on FOSDEM
94 (15:28:13) joanie: :-)
95 (15:28:21) API: so, there was a devroom about accessibility
96 (15:28:27) slee: API: one man show?
97 (15:28:32) API: I missed some of then
98 (15:28:46) slee: Someone told me about the room
99 (15:28:46) joanie: slee: aleiva and others were there
100 (15:28:50) API: and although we have some public
101 (15:29:04) API: it is clear that we weren0't the most popular devroom
102 (15:29:08) slee: joanie: thanks
103 (15:29:22) API: Mario couldn't go there, so they replace him with a
104 (15:29:27) API: talk by
105 (15:29:28) API: hmm
106 (15:29:31) joanie: Malte
107 (15:29:34) API: Openoffice accessibility guy?
108 (15:29:37) API: yes, that one ;)
109 (15:29:47) API: explaining more or less the status
110 (15:29:58) API: and there were some people asking questions
111 (15:30:03) joanie: Which is what? (the status)
112 (15:30:06) API: so at least some people get interest
113 (15:30:16) API: joanie, well, it is more about what they did
114 (15:30:18) slee: questions are good
115 (15:30:19) API: and some plans
116 (15:30:21) API: like IA2
117 (15:30:25) API: well, yes, but in my case
118 (15:30:30) API: one guy asked if
119 (15:30:38) API: instead of work on specific
120 (15:30:42) API: window managers
121 (15:30:49) API: if it would have sense to directly
122 (15:30:55) API: add some a11y support on X
123 (15:31:12) API: and forgot the specific bits
124 (15:31:15) API: anyway
125 (15:31:24) API: I talked with Malte there about the hackfest
126 (15:31:25) ***slee wishes life was that simple
127 (15:31:31) API: and he said
128 (15:31:42) API: that the atk/at-spi2 hackfest is interesting
129 (15:31:48) API: but right now he is not sure
130 (15:31:56) API: if he could go
131 (15:32:02) API: due company politics
132 (15:32:12) API: I was also on the LibreOffice presentation
133 (15:32:16) API: by Michael Meeks
134 (15:32:25) API: on the question turn, I asked about a11y
135 (15:32:31) ***clown brb
136 (15:32:34) API: in summary, OpenOffice has
137 (15:32:37) API: a11y support
138 (15:32:42) API: so LibreOffice also have
139 (15:32:53) API: "if you are interested in improve it, join us!"
140 (15:32:57) joanie: heh
141 (15:33:00) joanie: my question is this:
142 (15:33:15) joanie: Does Michael and crew plan to continue pulling what OOo A11y does?
143 (15:33:25) joanie: or are these forks truly separate now?
144 (15:33:52) API: hmm, good question
145 (15:33:58) slee: good point - were should the effort go?
146 (15:33:58) API: anyway, as far as I know
147 (15:34:16) API: LibreOffice doesn't have any problem to get patches from OpenOffice
148 (15:34:24) API: so
149 (15:34:31) API: if there are improvements on OpenOffice
150 (15:34:42) API: I guess that those could be used for LibreOffice
151 (15:34:57) API: although no idea if someone is willing to do that
152 (15:35:00) joanie: meaning they will actively look for them or meaning we (a11y team) need to keep up with what Malte's team does and then ping the LO team to include the change?
153 (15:35:01) mgorse: The licenses don't conflict then?
154 (15:35:29) API: well, I don't think so
155 (15:35:47) API: but no idea about the interrelations between both communities
156 (15:35:57) API: using my ignorance
157 (15:36:08) joanie: We need to figure out *exactly* what is going on (or not going on).
158 (15:36:09) API: I don't think that OpenOffice people notifies LO
159 (15:36:11) API: people
160 (15:36:17) joanie: :-)
161 (15:36:54) API: ok, makes sense
162 (15:36:56) ***joanie sees becoming an active member of the LO community in her future
163 (15:36:57) joanie: :-/
164 (15:37:00) ***fer likes forks like Xorg, where the original project just stops, so there is no need to sync patches :)
165 (15:37:31) ***clown doubts that Ooo is going to just stop.
166 (15:37:47) joanie: The problem here (beyond what clown just said) is our users expect to be able to use either
167 (15:37:47) API: I also doubt that
168 (15:37:59) joanie: and Michael apparently has that 'patch it or shut up' attitude
169 (15:38:04) ***slee thinks joanie is slacking if not already a member </ducks>
170 (15:38:08) API: and in fact some distributions started to move to libreoffice
171 (15:38:09) API: ie
172 (15:38:10) joanie: which I would agree with if we had tons of resources
173 (15:38:16) API: natty uses libreoffice by default
174 (15:38:16) joanie: but we don't
175 (15:38:23) ***joanie nods re natty
176 (15:38:35) joanie: the one good thing is that building LO is super easy
177 (15:38:41) fer: what do we prefer? to keep OOo and LO compatible? or trying to push changes on LO to get a11y improved and then have separate support for them (like orca stuff)
178 (15:38:43) joanie: much more open
179 (15:39:10) joanie: fer: I prefer they each are accessible AND that they make them accessible without us having to do the work to contribute patches
180 (15:39:15) slee: LO are a doign a great community engagement act - so are likely to win form that point of view
181 (15:39:17) joanie: I don't mind filing bugs
182 (15:39:32) slee: always be pretty hard to with OOo
183 (15:39:43) slee: mind you patch it or shut up is not good
184 (15:40:12) API: well, so trying to get some conclusions
185 (15:40:13) slee: if LO OOo diverge much dioing both will be huge effort.
186 (15:40:14) API: summary?
187 (15:40:15) API: actions?
188 (15:40:20) slee: might be OK for the short term
189 (15:40:22) joanie: I'll add LO to my list
190 (15:40:33) ***joanie looks at her list and sighs
191 (15:40:41) ***slee pats joanie on her back
192 (15:40:55) joanie: mind you I already asked in the LO a11y community
193 (15:41:09) joanie: time for me to become more visible in the non a11y LO communities.
194 (15:41:31) joanie: API thanks for attending that session and reporting
195 (15:42:26) API: I will try it again ;)
196 (15:42:31) API: <API> summary?
197 (15:42:31) API: <API> actions?
198 (15:42:36) joanie: Action
199 (15:42:43) joanie: I will put LO on my list
200 (15:42:47) joanie: (see above)
201 (15:43:36) ***joanie wonders what other actions API is looking for?
202 (15:43:55) API: well, this "put LO in my list"
203 (15:44:00) fer: some guys subscribing to LO mailing list and making noise :)
204 (15:44:04) API: also include investigate current relation between
205 (15:44:08) API: LO and OO?
206 (15:44:11) API: (new code and so on)
207 (15:44:15) joanie: yes all of the above
208 (15:44:25) joanie: 1. Making more noise on other LO lists
209 (15:44:46) joanie: 2. Trying to sort out *exactly* what the relationship between OOo and LO is or is not w.r.t. a11y fixes
210 (15:45:07) joanie: 3. Trying to sort out the expectations w.r.t. filing a11y bugs (which I assume is file duplicate bugs)
211 (15:45:21) joanie: 4. Trying to sort out if Michael really does mean patch it or shut up
212 (15:45:33) joanie: (i.e. what sort of support we can look to from LO)
213 (15:45:39) joanie: 5. Inform the respective communities
214 (15:45:56) slee: 6. trying to get them to do the donkey work
215 (15:46:00) joanie: i.e. if LO's a11y policy is patch it or shut up, that is not consistent with the Ubuntu commitment to a11y
216 (15:46:11) joanie: aka 'put it on my list'
217 (15:46:12) joanie: ;-)
218 (15:46:21) joanie: questions?
219 (15:46:55) fer: well, I'd like to assume it is "patch it or shut up" for new features, but not for fixing actual bugs on their code :)
220 (15:47:21) joanie: fer they are ignoring a crasher bug I filed
221 (15:47:31) ***clown wonders if moving to at-spi2 is a new feature.
222 (15:47:31) joanie: they responded immediately to say "could not reproduce it"
223 (15:47:50) joanie: and then when I reopened having built LO from master and provided a stack trace
224 (15:48:00) joanie: they couldn't be bothered to respond
225 (15:48:11) ***bnitz likes 5, would be great if "a11y support is too hard to do by yourself" is the catalyst to end the stalemate.
226 (15:49:02) API: ok, a lot of conclusions ;)
227 (15:49:03) API: thanks
228 (15:49:08) API: lets move
229 (15:49:20) API: 3. a11y testing towards GNOME3 using Fedora test day approach? ( https://fedoraproject.org/wiki/Test_Day:2011-02-03_GNOME3_Alpha )
230 (15:49:27) API: joanie, I guess that you added this point
231 (15:49:35) joanie: not I
232 (15:49:37) fer: no, I did
233 (15:49:46) ***joanie doesn't care what fedora does
234 (15:49:47) joanie: ;-)
235 (15:49:51) fer: ahaha
236 (15:49:54) fer: it's not about fedora
237 (15:50:01) fer: it's about the approach for testing
238 (15:50:16) fer: like having this kind of old bug days but for testing
239 (15:50:18) joanie: fer I know. I was being snarky
240 (15:50:20) API: well, yes I agree
241 (15:50:21) API: I mean
242 (15:50:33) API: that at least they provide a "gnome 3 compliant" distribution
243 (15:50:39) API: we can't say that from ubuntu natty
244 (15:50:45) API: that it is still a mixed thing
245 (15:50:51) API: it would be good to test
246 (15:50:54) API: a gnome 3 distribution
247 (15:51:00) API: and check that a11y is working
248 (15:51:02) joanie: and they are doing it strictly out of selfless motives, no doubt. :-P
249 (15:51:13) API: ie: I fear that all this gconf/gseetings thing creates some nightmares
250 (15:51:22) API: selfless motives?
251 (15:51:30) API: I thought that this was extinct ;)
252 (15:51:35) joanie: sorry snark again
253 (15:51:36) fer: probably try to run a a11y testing day would be hard, but we can try to add a11y testcases to Fedora testing days :)
254 (15:51:38) ***joanie behaves
255 (15:51:50) mgorse: gnome-shell a11y is still a work in progress, so I don't know how that effects things. We might have to test the "fallback mode" if anything
256 (15:52:29) API: mgorse, well, some patches related to gnome-shell a11y
257 (15:52:32) API: are now on the master
258 (15:52:37) API: although without solving this
259 (15:52:40) API: https://bugzilla.gnome.org/show_bug.cgi?id=640057
260 (15:52:42) ***bnitz wonders how to test gnome-shell magnification.
261 (15:52:45) API: it is still not really testable
262 (15:52:51) ***clown hides
263 (15:53:08) API: bnitz, well right now on gnome-shell you can activate magnification
264 (15:53:12) API: but you can't configure it :P
265 (15:53:27) API: at least without orca, although I didn't test it
266 (15:53:28) bnitz: API:I was wondering about automating that test.
267 (15:53:34) API: ah ok
268 (15:53:40) clown: bnitz, there is a patch for a GUI testing various mag settings...
269 (15:53:58) clown: https://bugzilla.gnome.org/show_bug.cgi?id=622414
270 (15:53:59) bnitz: clown:Good thanks.
271 (15:54:10) clown: but, right, that doesn't do auto testing.
272 (15:54:17) API: fer, anywat that link was for a testing day on 03-02
273 (15:54:21) API: I guess that we are late
274 (15:54:26) API: or I'm missing something?
275 (15:54:28) clown: bnitz, is there a framework to hook automatic testing into?
276 (15:54:34) fer: API: it was an example
277 (15:54:51) bnitz: clown:I'm building one based on mago.
278 (15:55:02) API: so your proposal is?
279 (15:55:07) API: fer, so your proposal is?
280 (15:55:24) clown: bntiz ETA? (no pressure, but I'm getting some from AEGIS/Peter to hook into your framework).
281 (15:55:29) bnitz: Right now it can use accerciser plugins and a few tests I've written. I think magnifier would have to be tested by comparing screenshots which mago can do.
282 (15:55:33) fer: not proposing anything, just asking you guys if you think it is a good idea to try to get some a11y related testcases there
283 (15:56:24) bnitz: clown:Planning for april but since my platform is on GNOME 2, I'm not focused on testing gnome 3 features such as gnome-shell mag.
284 (15:56:48) API: fer, for the next one?
285 (15:56:50) clown: bnitz, well I have toyed with the idea of unit tests like -- setting one of the mag settings, and then reading it back to see if it "took". Of course, that doesn't show one what actually happend on screen.
286 (15:57:10) fer: API: if we add them now, they would be tested on any future testing day, I guess
287 (15:57:24) clown: bnitz, good to know re: GNOME3 vs. GNOME2
288 (15:57:26) clown: thanks
289 (15:57:27) API: imho it makes sense
290 (15:57:29) API: but
291 (15:57:35) API: who have time to track this?
292 (15:58:01) joanie: and would the Fedora people actually be responsive
293 (15:58:02) ***fer hides
294 (15:58:14) fer: redhat people don't care about it
295 (15:58:21) fer: but community does
296 (15:58:22) ***clown sees fer behind that tree over there...
297 (15:58:43) API: well, the issue is that it would not be really "polite"
298 (15:58:44) fer: I could try to poke someone to check if they are positive about this
299 (15:58:53) API: to add a a11y thing
300 (15:59:00) joanie: personally I think that this would be a model to follow for 3.2
301 (15:59:00) API: and then
302 (15:59:02) joanie: and beyond
303 (15:59:06) API: not to test that by ourselves
304 (15:59:20) joanie: and through some other mechanism than piggy backing onto what Fedora is doing
305 (16:00:06) joanie: For this cycle I think we might be stuck with testing on our own. But that's just my opinion.
306 (16:00:34) fer: it makes sense
307 (16:00:43) fer: as that kind of testing is more focused on regressions
308 (16:00:44) joanie: For instance, partnering with the Vinux folks might make sense
309 (16:00:51) fer: rather than "this is not working yet" :)
310 (16:00:58) API: well, yes, but as say, having a "GNOME 3 compliant" distribution would make our live easier
311 (16:01:16) API: but, afaik, vinux are still based on ubuntu
312 (16:01:29) API: see above this "mixture gnome2 and gnome3" thing
313 (16:01:29) joanie: yup
314 (16:01:38) fer: what about OpenSuSE?
315 (16:01:42) joanie: See above the 3.2 reference
316 (16:02:17) mgorse: 11.4 isn't going to ship GNOME 3. I think there will be an external repository to get it, and there's a GNOME 3 livecd for it
317 (16:02:30) mgorse: but its release schedule is going to miss GNOME 3
318 (16:02:34) joanie: mgorse: really re live CD?
319 (16:02:42) joanie: is it available externally yet?
320 (16:02:53) joanie: I pulled the dev CD a couple of weeks ago
321 (16:03:17) mgorse: joanie: I'm not sure off-hand. I'll need to look. I just know that it gets built, since I was asking a question the other day and someone mentioned it
322 (16:03:28) joanie: the other possibility might be Foresight
323 (16:03:44) joanie: but I cannot get that to install completely
324 (16:03:52) joanie: (not that I've dug into it yet)
325 (16:04:29) joanie: anyhoo, since API is going to ask for summary and actions soon. :-P
326 (16:04:35) API: well, 5 minutes over time, so, could we try to
327 (16:04:36) API: yep
328 (16:04:38) API: ;)
329 (16:04:42) joanie: I propose we add this sort of organized testing to the Roadmap
330 (16:04:46) API: get some summary, conclusions and actions
331 (16:04:59) joanie: i.e. to have our own sort of Fedora testing day (but without Fedora, necessarily)
332 (16:05:01) API: this roadmap?
333 (16:05:04) API: http://live.gnome.org/Accessibility/Roadmap
334 (16:05:12) joanie: but emulate their techniques/approach
335 (16:05:14) joanie: yessir
336 (16:05:18) joanie: re that roadmap
337 (16:05:22) mgorse: joanie: http://blog.crozat.net/2011/01/gnome-3-live-cd-usb-test-image.html
338 (16:06:04) joanie: mgorse: ah thanks.
339 (16:07:12) API: ok, makes sense
340 (16:07:15) API: sooo
341 (16:07:21) API: as we are over time
342 (16:07:26) API: can we conclude the meeting
343 (16:07:35) API: or someone requires miscellaneous time?
344 (16:07:45) joanie: no. the meeting must continue into perpetuity
345 (16:07:53) clown: API, I had an agenda item...
346 (16:07:53) joanie: oops. behaving NOW.
347 (16:07:54) ***slee just come back at the end
348 (16:08:17) API: clown, column thing?
349 (16:08:21) clown: yes.
350 (16:08:25) fer: yeah
351 (16:08:29) API: well....
352 (16:08:31) fer: and I have also an issue
353 (16:08:40) API: this seems too long for miscellaneous time
354 (16:08:49) API: lets place that as the first item on the next meeting
355 (16:08:49) bnitz: In case anyone is interested in a11y testing code within the mago framework, here is the latest: https://code.launchpad.net/~brian-nitz/+junk/a11ytesting
356 (16:09:13) clown: API, the working group who asked the question wanted the answer by next Mon...
357 (16:09:19) fer: well, quick info: about WAI User agent guidelines, some guy from IBM asked me to do a review of it regarding atk
358 (16:09:29) bnitz: It allows event-triggered running of some of the tests in Accerciser's basic validation plugin.
359 (16:09:29) clown: but, we don't have to do it now; we can discuss in the #a11y channel.
360 (16:09:51) clown: fer -- sounds like the issue they asked me.
361 (16:09:54) joanie: clown: Quick summary?
362 (16:10:08) joanie: (for the purpose of minutes)
363 (16:10:13) ***slee waves
364 (16:10:23) clown: joanie -- look at this table: http://www.w3.org/WAI/PF/aria-implementation/#mapping_role_table
365 (16:10:27) fer: clown: the many of us looking at it, the better the guide would be
366 (16:10:40) clown: should the column labelled "ATK" be labelled "AT-SPI" instead?
367 (16:10:42) joanie: oh yeah, that came up in open a11y too
368 (16:10:46) fer: well, they asked me about progressbar and so on, but I spoted some bugs there
369 (16:10:48) clown: or ATK/AT-SPI?
370 (16:11:08) API: fer, well that table already have a ATK column
371 (16:11:10) clown: and, does the at-spi-corba vs. at-spi-dbus have any relevance here?
372 (16:11:12) API: so whats missing?
373 (16:11:17) fer: the ATK / AT-SPI thing is tricky, because sometimes it is not 1 to 1 map
374 (16:11:41) API: well, in that case, it would make sense two columns
375 (16:11:43) API: I guess
376 (16:11:51) API: or a ATK/AT-SPI column
377 (16:11:58) API: and explain the differences if neccesary
378 (16:12:31) fer: yup
379 (16:12:44) fer: example, look at aria-checked (state)="false" / "true" / "mixed"
380 (16:12:46) clown: API, the question is as a browser developer or an AT developer, what a11y interface(s) should I consult when developing on GNOME with respect to ARIA
381 (16:12:49) clown: ?
382 (16:12:55) fer: ATK column
383 (16:13:15) API: well clown
384 (16:13:16) fer: it says nothing about setting/clearing ATK_STATE_CHECK
385 (16:13:33) API: you mean if you should check ATK or AT-SPI?
386 (16:13:38) fer: and says Expose ATK_STATE_INDETERMINATE/STATE_SYSTEM_MIXED, and we don't have such a MIXED state
387 (16:13:41) API: where check==consult
388 (16:14:25) API: hmm, yes it is true
389 (16:14:29) clown: API, right -- what GNOME (linux?) documentation/libraries should I look at?
390 (16:14:32) API: that state is put there
391 (16:14:33) joanie: but we do have STATE_INDETERMINATE. Where did MIXED come from?
392 (16:14:50) API: in my opinion at-spi2
393 (16:14:52) clown: joanie, I know the ARIA motivation for mixed.
394 (16:14:52) fer: ia2 copys / paste bug probably
395 (16:15:02) API: hmm
396 (16:15:04) joanie: clown: not asking about motivation
397 (16:15:05) clown: API does at-spi2 == at-spi-dbus?
398 (16:15:07) API: a question
399 (16:15:08) joanie: asking who put it there
400 (16:15:09) joanie: ;-)
401 (16:15:22) API: this table is a guide for browser developers, right?
402 (16:15:29) clown: joanie -- that may be lost to history ;-)
403 (16:15:32) fer: conclusion: the whole document
404 (16:15:32) API: in that case, for GNOME
405 (16:15:32) clown: aaron?
406 (16:15:36) API: ATK
407 (16:15:40) API: as the browsers
408 (16:15:45) API: would implement ATK roles/states
409 (16:15:50) clown: API this is a guide for browser AND AT developers.
410 (16:15:51) API: not at-spi roles/states
411 (16:15:56) API: urgh
412 (16:16:02) API: well, in this case
413 (16:16:03) clown: AT developers since they are using the a11y API.
414 (16:16:08) API: both
415 (16:16:26) API: and stating that ATK is for the browser developers and at-spi API is for the AT tools
416 (16:16:28) fer: yeah, but they implement that trough atk, instead of directly talking to at-spi
417 (16:17:06) ***clown likes API's "stating that ATK is for the browser developers and at-spi API is for the AT tools".
418 (16:17:31) joanie: although that won't apply in the Qt world
419 (16:17:38) ***clown notes that browser write to the a11y API, and ATs read from it (at a first approximation).
420 (16:17:38) API: well
421 (16:17:52) API: joanie, right now there isn't a11y support on the Qt world
422 (16:18:02) joanie: it's coming though.
423 (16:18:06) clown: joanie, yes, QT/KDE is another wrinkle.
424 (16:18:25) fer: what api should linux Qt based browsers use?
425 (16:18:29) joanie: And documents should, when possible, be flexible enough to continue to be relevant in the near future
426 (16:18:40) clown: but where there is an a11y api (atk or atk/at-spi) that should be noted.
427 (16:18:42) API: I mean that we should worry about Qt when we have something to check
428 (16:18:55) API: in the case of Qt, if the qt-bridge is finished
429 (16:19:02) API: it would be just about at-spi
430 (16:19:05) joanie: right
431 (16:19:10) API: afaik, and after my conversations with fregl
432 (16:19:21) API: they are not using the internal qt bits related to ia2 or msaa
433 (16:19:28) API: they are directly using qt api
434 (16:19:28) joanie: my point is, I'm not suer that we should state explicitly what ATK is for versus what AT-SPI is for
435 (16:19:36) joanie: do a 'ATK/AT-SPI' and be done with it
436 (16:19:45) API: why not?
437 (16:19:50) joanie: if the implementors don't know which is which we have a bigger problems
438 (16:20:01) joanie: because Qt won't implement ATK
439 (16:20:03) joanie: (right?)
440 (16:20:14) clown: joanie, dunno
441 (16:20:26) joanie: thus your dichotomy would fall apart in that case
442 (16:20:41) clown: to be really concrete what would Qt use for an aria role of, say "alert"?
443 (16:20:47) fer: but would QT expose directly at-spi bits? or their own QAccessible classes?
444 (16:20:49) joanie: if Qt implements AT-SPI
445 (16:21:06) joanie: I *assume* (always a bad idea) AT-SPI
446 (16:21:12) joanie: rather than some sort of QAccessible
447 (16:21:21) ***joanie pokes fregl
448 (16:21:28) API: yes, AFAIU
449 (16:21:37) API: qt-bridge is literally that
450 (16:21:51) API: without a QAccessible implementing any accessibility API
451 (16:21:58) API: anyway, I think that we are losing the focus
452 (16:22:06) joanie: therefore, the way to make this document continue to apply is to NOT specify the aforementioned differences between ATK and AT-SPI
453 (16:22:07) API: as clown initially asked about GNOME world
454 (16:22:15) ***joanie shuts up
455 (16:22:17) clown: API, right
456 (16:22:35) API: well, joanie, IMHO yes
457 (16:22:42) API: ATK/AT-SPI for GNOME
458 (16:22:50) API: Qt/AT-SPI for KDE or whatever
459 (16:23:16) fer: at-spi should be hidden by toolkit bridges
460 (16:23:29) fer: and toolkit should expose only Atk or QAccessible or whatever
461 (16:23:57) clown: here was a question posed at the teleconference:
462 (16:24:28) clown: if I use Accerciser to look at the a11y info, am I seeing ATK info? or more than ATK?
463 (16:24:46) fer: oh, that is a good point
464 (16:25:08) fer: people using accerciser for testing do see at-spi interfaces
465 (16:25:16) API: well, but this is again going to the app vs at
466 (16:25:21) API: accerciser is a AT tool
467 (16:25:29) API: so it would be looking for at-spi interfaces
468 (16:25:34) clown: because the browser/AT developers are going to use Accerciser to "test" if things are properly written/read.
469 (16:25:37) API: but other equivalent question would be
470 (16:25:50) fer: that is the only thing implementors implementing ATK interfaces can use :)
471 (16:25:52) API: "if Im implement a11y support of my_browser, am I seeing ATK info"
472 (16:25:53) ***clown notes that he is echoing fer.
473 (16:25:55) API: and the answer is yes
474 (16:26:31) clown: so the column is properly labelled?
475 (16:26:36) clown: right now it's "ATK".
476 (16:26:58) fer: and they are filling it based on testing done with accerciser I guess :)
477 (16:26:58) slee left the room (quit: ChatZilla 0.9.86 [Firefox 3.6.13/20101203075014]).
478 (16:27:02) joanie: clown: I would (again) call it ATK/AT-SPI
479 (16:27:06) API: my vote it is still move it to a "ATK/AT-SPI " thing
480 (16:27:12) joanie: and (again) I would not distinguish what either is for
481 (16:27:25) clown: joanie, API that's how I'm leaning.
482 (16:27:33) joanie: I would leave researching that as an exercise for the reader
483 (16:27:46) fer: and then we have a mess when there is no 1 to 1 map from at-spi and Atk :)
484 (16:27:58) joanie: fer....
485 (16:28:01) clown: with a footnote that "AT-SPI" means AT-SPI-2 (Dbus).
486 (16:28:04) joanie: one of the goals of the hackfest
487 (16:28:13) joanie: is to bring those two more closely in alignment
488 (16:28:24) fer: agreed
489 (16:28:36) joanie: so if we (again) are looking at how to have a document which *remains* relevant for more than 6 months....
490 (16:29:04) joanie: I don't think that the few differences which currently exist (would be noticed) between the two are all that critical
491 (16:29:06) joanie: imho
492 (16:29:54) clown: pardon my ignorance, but why isn't there a 1-to-1 map between at-spi and atk? (perhaps a topic for another time...)
493 (16:30:12) ***joanie glances sideways at clown and chuckles
494 (16:30:21) fer: clown: that it the past! look at the future! :)
495 (16:30:53) clown: fer: the future is all transparent and animated (gnome shell and gnome3).
496 (16:31:01) joanie: lol
497 (16:31:07) API: clown, well the easy answer
498 (16:31:07) API: is
499 (16:31:13) API: that at-spi
500 (16:31:19) API: was though to be used not just only with atk
501 (16:31:26) clown: btw, that raises a related question: what does the ST toolkit a11y publish? ATK info?
502 (16:31:33) API: so some divergencies were not important
503 (16:31:40) API: clown,
504 (16:31:40) API: yes
505 (16:31:41) API: ATK
506 (16:31:53) API: I made some accessibility objects implementing ATK
507 (16:31:54) clown: API -- good point. at-spi is an a11y broker for any a11y API.
508 (16:31:56) API: like gail
509 (16:32:10) API: although it is also true, that right now
510 (16:32:13) API: due the bonobo thing
511 (16:32:30) API: ATK is mostly the only "client-side" user of at-spi
512 (16:32:38) API: although in theory, Qt is near
513 (16:32:44) fer: dudu, I had almost forgotten about bonobos! those monkeys!
514 (16:32:45) API: well, 30 minutes over time
515 (16:32:48) API: something else in this point?
516 (16:33:14) clown: API, not from me -- I think I have what I need to go back to the aria working group.
517 (16:33:32) API: well ok
518 (16:33:35) API: so meeting over
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.