15:07:11 <API> #startmeeting
15:07:11 <Services> Meeting started Thu Dec 18 15:07:11 2014 UTC.  The chair is API. Information about MeetBot at http://wiki.debian.org/MeetBot.
15:07:11 <Services> Useful Commands: #action #agreed #help #info #idea #link #topic.
15:07:40 <API> #topic Progress towards 3.16
15:08:01 <API> #info this week is 3.15.3 release week
15:08:39 <API> #info there are new releases for atk and at-spi2-atk, that adds the (infamous) role ATK_ROLE_STATIC
15:08:57 <API> #info so in summary, during this early releases, the new API is mostly new roles/states
15:09:16 <clown> is it what you get when you walk on carpet with socks?
15:09:21 * clown ducks.
15:09:30 <API> #info during past webengine hackfest (former webkitgtk hackfest) I chatted a little about mathml and roles/subroles with joanie
15:10:07 <API> #info roles/subroles bug here: bug 739140
15:10:07 <Services> 04Bug http://bugzilla.gnome.org/show_bug.cgi?id=739140 normal, Normal, ---, atk-maint, UNCONFIRMED, Consider adding subroles
15:10:07 <API> and done
15:10:10 <API> I was mostly watching
15:10:16 <API> so probably joanie has more to report
15:10:20 <joanie> #info Joanie has been busy with WebKit hacking (in general plus MathML related) and is behind on doing unstable releases.
15:10:27 <joanie> #info Joanie will be resuming hacking on Orca (has actually already done so), and will resume doing proper releases (including a 3.15.3 release this week).
15:10:37 <joanie> #info Joanie really needs subroles for the MathML support and is waiting for feedback from Piñeiro.
15:10:40 <joanie> done
15:11:59 <clown> a thought that just occurred to me:  the ePub people want to add a lot of new roles, specific to eBooks.
15:12:10 <joanie> clown: they'll be subroles too
15:12:11 <clown> that might be worth paying attention to.
15:12:18 <joanie> you'll see chapter in the list already I think
15:12:23 <clown> and it might dovetail with the new atk bug.
15:12:28 <clown> ah, okay.
15:12:34 <joanie> already on my radar
15:12:39 <clown> and I am catching up...
15:12:56 <joanie> also added glossary, table of contents
15:13:05 <joanie> chart
15:13:12 <joanie> although we had chart as a role
15:13:17 <joanie> anyhoo, we can continue to add subroles
15:13:21 <joanie> as they come in from dpub
15:14:18 <clown> ok
15:15:14 <joanie> oh, I added endnote too (footnote is, if memory serves me, not new. but either way it's there as a proposed subrole)
15:15:34 <joanie> also map -- the non imagemap/you-are-here flavor
15:15:46 <joanie> anyhoo, the ball is in API's court
15:15:47 <API> #action API will review the last comment on the bug as soon as possible
15:16:03 <joanie> I was hoping we'd have had more time when I was there to discuss, but oh well
15:16:23 <joanie> thanks API
15:16:32 <API> ok, so anythin gelse on this topic?
15:17:04 <clown> not from me.
15:17:39 <API> #topic w3c updates
15:17:40 <API> clown, ?
15:17:48 <clown> two things.
15:18:23 <clown> #info Last week, the latest public working drafts of the ARIA spec, core mappings, and accessilbe name computation documents were published.
15:18:34 <clown> #info the announcement is here:  http://www.w3.org/blog/news/archives/4236
15:19:23 <clown> #info  Secondly, regarding the roles FF exposes for listitems or grid cells in a list or table with role="presentation".
15:19:45 <clown> #info See https://bugzilla.mozilla.org/show_bug.cgi?id=1013584
15:19:45 <Services> 04Bug 1013584: normal, --, mozilla36, surkov.alexander, RESOLVED FIXED, Children of tables and lists with role='presentation' should each be exposed as if they were div elements
15:19:56 <clown> #info Joseph tested with nightly of FF.
15:20:26 <clown> #info the findings are noted in ACTION-1542:  https://www.w3.org/WAI/PF/Group/track/actions/1542 (requires password).
15:20:48 * joanie apologizes for being busy
15:21:04 <joanie> paragraph is not a div element
15:21:09 <clown> #info  here they are:  list item -> ROLE_PARAGRAPH, table cell->ROLE_TEXT + AccessibleText interface.  A plain <DIV> is mapped to ROLE_SECTION
15:21:12 <joanie> role_text is wrong
15:21:46 <joanie> if that's really what they did, then the implementation is not correct
15:21:55 <clown> #for comparison, under IA2 on Win7, list item is ROLE_PARAGPH, table cell is ROLE_TEXT_FRAME, and DIV is ROLE_SECTION.
15:21:59 * joanie sighs
15:22:07 <clown> yeah, I thought you might.
15:22:15 <clown> s/#for/#info for/
15:22:24 * joanie downloads a nightly
15:22:50 <clown> I'm willing to publish those results on the FF bugzilla — is that the way to go?
15:23:04 <joanie> the results as you found them are wrong
15:23:11 <joanie> that is not the correct mapping for ATK
15:23:22 <joanie> if that is the mapping they f'ed up
15:23:39 <clown> joanie, I mean to add a comment to but 1013584 documenting what I found.
15:23:47 <joanie> aha
15:23:49 <clown> not, that I will change the core-aam mapping tables.
15:23:51 <joanie> that's fine then :)
15:23:53 <joanie> thank you
15:23:56 <joanie> you read my mind :)
15:23:58 <clown> no problem.
15:24:09 <joanie> we're still recovering from mapping bugs
15:24:21 <joanie> in which the "correct" mapping is whatever crazy thing Gecko does
15:24:31 <joanie> </rant>
15:24:43 <joanie> anyhoo, yes, clown if you comment your findings on the bug
15:25:03 <clown> #info At the last AAPI teleconference, Cynthia (Microsoft) thought the mapping of list item to paragraph was reasonable.
15:25:05 <joanie> noting that they repurposed my bug (was just for ATK) and then "fixed" it incorrectly
15:25:14 <joanie> it's not a paragraph
15:25:15 <clown> #info but it's still under discussion.
15:25:21 <joanie> and navigation by paragraph will be broken
15:25:38 <joanie> and Orca users will complain that Orca's navigation by paragraph moves them to objects which are not paragraphs
15:25:54 <clown> how do they know they are not paragraphs?
15:25:58 <clown> just curious.
15:26:35 <clown> Is the bullet or number lost when they "lose" their list item role?
15:27:07 <joanie> clown: my guess is not if the author doesn't remove it
15:27:12 * clown should check that with the test file.
15:27:21 <joanie> the listmarker is a rendered text node
15:27:24 <joanie> for ul and ol
15:27:37 <joanie> when nothing special has been done to suppress them
15:27:39 <clown> here's the test file that you wrote, joanie:  https://bug1013584.bugzilla.mozilla.org/attachment.cgi?id=8425789
15:28:06 <clown> I don't think the bullets will vanish.
15:28:07 <joanie> and I did nothing to suppress them
15:28:15 <joanie> so there's no reason to expect the bullets will vanish
15:28:17 <clown> right.
15:28:23 <joanie> ARIA (as you know) doesn't do that sort of thing
15:28:27 <clown> right.
15:28:33 <joanie> but that doesn't bother me
15:28:37 <joanie> it's just a character
15:28:42 <joanie> what bothers me is bogus roles
15:30:11 <clown> well, my curiosity was:  how do they know they are not paragraphs?  Maybe because there this weird bullet character at the front of each of them.
15:30:30 <joanie> probably because they are wicked short
15:30:37 <joanie> and then they wonder why orca is doing that
15:30:41 <joanie> and then they look at the source
15:30:44 <joanie> and then they blame orca
15:30:45 <joanie> ;)
15:30:54 <joanie> i.e. it's not the bullet
15:31:09 <clown> here's where it gets wonky:  I *think* one can create a <li> as a container for paragraphs.
15:31:11 <joanie> an inline list functioning as a nav bar might have no bullets
15:31:19 <joanie> one could
15:31:33 <joanie> and if one did you'd have a <div><p>$STRING</p></div>
15:31:34 <joanie> and that's ok
15:31:46 <joanie> if they made a paragraph then a paragraph is fine
15:31:47 <joanie> :)
15:31:53 <clown> but, as far as I know a <p> cannot contain another <p>
15:32:04 <joanie> perhaps it should not, but I've seen it
15:32:09 <clown> so, mapping <li> to <p> in that case is very odd.
15:32:10 <joanie> and Gecko does it for things like dropcaps
15:32:14 <joanie> it's very sad
15:32:25 <joanie> anyhoo, "exposed as if they were divs" means just that
15:32:26 <clown> dropcaps are paragarphs?
15:32:36 <joanie> once Gecko a11yifizes them
15:32:54 <joanie> the span in a paragraph that's a dropcap becomes a child paragraph
15:32:54 <clown> well, my guess is, the gecko team didn't agree that mapping to divs was the way to go.
15:33:00 <joanie> I've already wtfed that to Surkov
15:33:16 <joanie> and because they disagree, they are not implementing support for ATK correctly
15:33:35 <joanie> fortunately we already have a correct implementation, namely WebKitGtk
15:33:44 <clown> ok.
15:34:34 <joanie> anyhoo, we deep-dove
15:34:48 <API> yes
15:34:49 <API> probably
15:34:50 <clown> yes.  I just realized that, and decided to "shut up"
15:35:11 <API> do you want to summarize this last converstaion on  a info or just move?
15:35:17 <clown> #action:  Joseph will report his results on the FF bug sometime today.
15:36:36 <clown> I'm fine with just moving.
15:37:23 <API> ok
15:37:38 <API> joanie mentioned before that jjmarin is not coming
15:37:45 <API> (so I conclude that he is just lurking)
15:37:51 <API> so I will skip marketing
15:37:53 <API> so that means
15:37:58 <API> #topic Miscellaneous time
15:38:28 <API> #info bug 730505 was finally closed
15:38:28 <Services> 04Bug http://bugzilla.gnome.org/show_bug.cgi?id=730505 normal, Normal, ---, at-spi-maint, RESOLVED FIXED, Draft for atspi tests
15:38:32 <API> #info that means that we have now a basis for tests on at-spi2-atk
15:39:00 <API> #info eventually it would be good if we are able to move to a more general at-spi2-core, but it is true that having the possibility to create atk objects is convenient
15:39:09 <API> #info Next steps: 1. add more tests
15:39:44 <API> #info 2. probably nuke the similar ones that are at pyatspi2, at least the c ones, as doesn't fit there
15:39:45 <API> done
15:39:48 <API> some other misc stuff?
15:42:16 <joanie> sorry, I'm in the mozilla channel politely asking them wtf
15:42:18 <joanie> ;)
15:42:43 <joanie> #info Joanie's next priority for this cycle is doing full caret-nav support for WebKitGtk/Epiphany
15:42:49 <clown> sheesh.  Almost no time to add my comment to the bugzilla.
15:42:57 <joanie> so that we don't have to deal with incorrect mappings by Gecko
15:43:03 <joanie> sorry clown, but I have more meetings
15:43:08 <joanie> and this is just p*ssing me off
15:43:18 <joanie> I file a bug saying what the correct mapping is
15:43:24 <joanie> they close it by making it incorrect
15:43:34 <joanie> using apparently ATK_ROLE_TEXT of all things
15:43:51 <joanie> even though they know that is not correct in an extreme fashion
15:44:36 <joanie> so I apologize to you clown for jumping the gun
15:45:00 <joanie> but you shouldn't have to comment because they should have fixed it correctly for our platform as described in the bug
15:47:33 <clown> okay.
15:47:48 <API> so
15:47:50 <API> closing meeting?
15:50:04 <clown> sorry, I was adding my comment to the FF bug.  That's done.
15:50:26 <clown> And I have not other misc items other than I'm not here next week.
15:50:45 <clown> So, Merry Christmas and a Happy New Year, all.
15:50:52 <clown> (not here the week after, either).
15:52:03 * API looking calendar
15:52:08 <API> hmm true
15:52:16 <API> probably we should just cancel next wek meeting
15:52:17 <API> *week
15:52:31 <API> I will not be here
15:52:35 <API> and joanie is saying the same
15:52:45 <API> clown, would you like to met yourself next week?
15:53:25 <clown> API, I meet with myself every waking moment.  :-)
15:53:53 <API> ok, then I will send a canceling meeting email later
15:53:56 <API> and having said so
15:53:58 <API> #endmeeting