/irc-logs / mozilla / #accessibility / 2014-12-08 / end
Options:
- # Session Start: Mon Dec 08 00:00:00 2014
- # Session Ident: #accessibility
- # [00:20] * yzen is now known as yzen_
- # [00:24] * Quits: @yzen_ (yzen@moz-ql1r6k.cpe.pppoe.ca) (Ping timeout: 121 seconds)
- # [03:18] * Joins: yliao (yliao@moz-4kr52a.wa.comcast.net)
- # [03:40] * Joins: surkov (surkov@moz-v8g78c.cpe.teksavvy.com)
- # [03:40] * ChanServ sets mode: +o surkov
- # [04:14] * Quits: yliao (yliao@moz-4kr52a.wa.comcast.net) (Connection closed)
- # [04:14] * Joins: yliao (yliao@moz-4kr52a.wa.comcast.net)
- # [04:18] * Quits: yliao (yliao@moz-4kr52a.wa.comcast.net) (Connection closed)
- # [04:18] * Joins: yliao (yliao@moz-4kr52a.wa.comcast.net)
- # [04:25] * Quits: yliao (yliao@moz-4kr52a.wa.comcast.net) (Connection closed)
- # [05:52] * Joins: icaaq (icaaq@moz-n1f1o3.cust.bredbandsbolaget.se)
- # [06:12] * Joins: yliao (yliao@moz-tuhs4n.hfc.comcastbusiness.net)
- # [06:44] * Quits: a-865 (fmcz@moz-6psm3k.cable.mindspring.com) (Quit: ChatZilla 0.9.91 [SeaMonkey 2.31/20141125174516])
- # [06:49] * Joins: yliao_ (yliao@moz-tuhs4n.hfc.comcastbusiness.net)
- # [06:49] * Quits: yliao (yliao@moz-tuhs4n.hfc.comcastbusiness.net) (Connection closed)
- # [06:52] * Joins: a-865 (fmcz@moz-6psm3k.cable.mindspring.com)
- # [07:43] * Quits: yliao_ (yliao@moz-tuhs4n.hfc.comcastbusiness.net) (Connection closed)
- # [07:49] * Tomcat|afk is now known as Tomcat|sheriffduty
- # [07:57] * Quits: MrMazda (fmcz@moz-6psm3k.cable.mindspring.com) (Ping timeout: 121 seconds)
- # [07:57] * Quits: a-865 (fmcz@moz-6psm3k.cable.mindspring.com) (Ping timeout: 121 seconds)
- # [07:57] * icaaq is now known as icaaq|afk
- # [08:01] * icaaq|afk is now known as icaaq
- # [08:25] * Joins: a-865 (fmcz@moz-6psm3k.cable.mindspring.com)
- # [08:25] * Joins: MrMazda (fmcz@moz-6psm3k.cable.mindspring.com)
- # [10:04] * Joins: fredw (Icedove@moz-lnbnpf.igalia.com)
- # [10:11] * Quits: @surkov (surkov@moz-v8g78c.cpe.teksavvy.com) (Client exited)
- # [10:53] * Joins: agibson (agibson@moz-j04gi9.cable.virginm.net)
- # [12:29] * icaaq is now known as icaaq|afk
- # [12:31] * Joins: Justin_o (uid14648@moz-c6vipl.ealing.irccloud.com)
- # [12:49] * agibson is now known as agibson|afk
- # [13:05] * icaaq|afk is now known as icaaq
- # [13:05] * icaaq is now known as icaaq|afk
- # [13:07] * icaaq|afk is now known as icaaq
- # [13:33] * Joins: aurelien_imac (chatzilla@moz-tj6t2q.rev.sfr.net)
- # [13:33] * aurelien_imac is now known as goetsu
- # [13:58] * Joins: yzen (yzen@moz-ql1r6k.cpe.pppoe.ca)
- # [13:58] * ChanServ sets mode: +o yzen
- # [14:05] <goetsu> Hi, i'm a newbie here I try to fix some a11y bug in gaia
- # [14:06] <goetsu> for some of them I need to add an aria-label, can anyone tell me the way to do it well to handle i18n
- # [14:19] <goetsu> if I understand it well it something like <span data-l10n-id="mykey"> and mykey.ariaLabel in the locales
- # [14:21] <goetsu> I also see that there is a specific locale file for accessibility, accessibility.en-US.properties, is my ariaLabel going here or not ?
- # [14:27] <@yzen> goetsu hi
- # [14:28] <@yzen> yes regarding localization and .properties files
- # [14:29] <@yzen> accessibility file is for screen reader specific strings like roles, element types etc
- # [14:29] <goetsu> ok so not for my case
- # [14:30] <@yzen> goetsu correct, in your case , all new strings should go into the .properties file that corresponds to the app you are working on
- # [14:31] <goetsu> ok
- # [14:31] <@yzen> so for music it will be in music/locales
- # [14:31] <@yzen> and then just pick music.en-US.properties, other locales will be triggered when that file is updated
- # [14:33] <@yzen> goetsu also, make sure you add a comment like this - https://github.com/mozilla-b2g/gaia/blob/master/apps/ftu/locales/ftu.en-US.properties#L164-L165 in front of you entry
- # [14:33] * Joins: Gijs (chatzilla@moz-pnd.q6m.80.5.IP)
- # [14:33] <@yzen> if you are just adding a single string it should be mentioned in that comment as such: https://github.com/mozilla-b2g/gaia/blob/master/apps/ftu/locales/ftu.en-US.properties#L164-L165
- # [14:33] <goetsu> ok
- # [14:34] <@yzen> oops wrong link: https://github.com/mozilla-b2g/gaia/blob/master/apps/ftu/locales/ftu.en-US.properties#L294-L295
- # [14:34] <goetsu> any pattern to follow in the value name ?
- # [14:36] <@yzen> goetsu also regarding the list semantics, i think the original comment mentioned grid but after going over it again, i dont think it fits either that or list/listbox that's why I think we should drop it and just use button on the actual item
- # [14:37] <goetsu> ok as you like
- # [14:37] * Joins: slee (chatzilla@moz-93hjmp.range81-132.btcentralplus.com)
- # [14:39] <@yzen> goetsu value name ?
- # [14:43] <goetsu> the value I use in data-l10n-id
- # [14:44] <@yzen> goetsu gotcha, anything sensible goes, the only thing i avoid is actually having a label/name/title/[type] prefix, so correct would be "next" and not "nextBotton" or "nextLabel"
- # [14:44] <@yzen> that's about it
- # [14:44] <@yzen> i mean postfix
- # [14:44] <@yzen> goetsu that accessibility- prefix you saw is really a pattern for the accessibility properties file
- # [14:44] <@yzen> only
- # [14:45] <@yzen> goetsu so you also wanted to know about rebasing ?
- # [14:45] <goetsu> yes
- # [14:45] <goetsu> https://github.com/mozilla-b2g/gaia/pull/26568 is a good candidate for this I think
- # [14:48] <goetsu> I use the github app and terminal (but If you know a GUI where we can do this kind of things and fetching/merging upstream without the terminal I'm interested)
- # [14:52] <goetsu> what do I need to do to rebase my local branch to my up to date master branch and merge all the commit in the PR to one
- # [14:53] <@yzen> right
- # [14:53] <@yzen> so first thing i guess is to get your local master branch up to date with the upstream(official one)
- # [14:54] <goetsu> I know how to do that one
- # [14:54] <@yzen> ok goo
- # [14:54] <@yzen> d
- # [14:54] <goetsu> git fetch upstream
- # [14:54] <goetsu> git merge upstream/master
- # [14:54] <@yzen> goetsu so you want to have 5 commits merged into 1 right ?
- # [14:54] <@yzen> goetsu just and fyi, there's 42 files changed there
- # [14:56] <goetsu> to update my branch I need to do git merge upstram/nameofmybranch ?
- # [14:57] <@yzen> you generally should not need to do that, only if that branch was updated upstream from somewhere else other than your local branch
- # [14:58] * agibson|afk is now known as agibson
- # [14:59] <@yzen> goetsu, so here's what you need
- # [14:59] * Quits: slee (chatzilla@moz-93hjmp.range81-132.btcentralplus.com) (Ping timeout: 121 seconds)
- # [15:00] <@yzen> goetsu when you merged master into your branch you need to call git rebase -i upstream/master
- # [15:01] <goetsu> ok
- # [15:01] <@yzen> goetsu you ll see a prompt with a list of your commits
- # [15:01] <@yzen> pick bb2b3b7 Bug-1069614
- # [15:01] <@yzen> pick 8d539cb add label to cover without visible artiste/album text
- # [15:01] <@yzen> pick a3b6282 respect coding style
- # [15:01] <@yzen> pick 49c28a3 update
- # [15:01] <@yzen> pick 025adf2 cleaning up PR
- # [15:03] <@yzen> goetsu actually before that i think that 'update' commit needs to be removed
- # [15:03] <@yzen> so cancel rebase
- # [15:03] <@yzen> and call git rebase --abort
- # [15:04] <@yzen> goetsu let me know when you aborted the rebase
- # [15:05] <@yzen> so the SHA for that commit is 49c28a3
- # [15:08] <goetsu> i'm a little lost between commande line and the github app, do I need first to select my working branch on the app because I also can go in the history and rollback or revert
- # [15:10] <goetsu> sorry
- # [15:10] <@yzen> goetsu no worries
- # [15:10] <@yzen> lets see
- # [15:10] <@yzen> if you are in your branch
- # [15:10] <@yzen> perhaps the app will let you delete a particular commit
- # [15:10] <@yzen> you need to get rid of the one with the message 'update'
- # [15:10] <@yzen> does the app have that option ?
- # [15:11] <@yzen> goetsu one sec ill download it too
- # [15:11] <goetsu> no but I can revert or rollack to the previous one
- # [15:16] <@yzen> goetsu sure do revert of that commit
- # [15:16] <@yzen> (it will show up as another commit)
- # [15:19] <goetsu> done but the PR still announce 42 files changed
- # [15:20] <@yzen> goetsu yeah, it looks like "respect coding style" was reverted instead of "udpate"
- # [15:20] <@yzen> so now we need to revert both "Revert "respect coding style"" commit and "update", which is fine
- # [15:21] <goetsu> I don't think I can do that with the app
- # [15:21] <goetsu> (both at the same time I mean)
- # [15:21] <@yzen> goetsu no problem, i think the best app i tried is source tree, it's free but it might ask to create an account within 30 days
- # [15:22] <@yzen> also see if you can do one by one
- # [15:28] <goetsu> ok I had it in sourcetree
- # [15:28] <goetsu> ok I have it in sourcetree
- # [15:29] * Joins: MarcoZ (marco.zehe@moz-tb35md.dip0.t-ipconnect.de)
- # [15:29] * ChanServ sets mode: +ao MarcoZ MarcoZ
- # [15:30] <&MarcoZ> Good day!
- # [15:30] <@yzen> goetsu cool, so first right click on your last revert commit and click reverse
- # [15:30] <@yzen> goetsu then pick 'update' commit and click reverse
- # [15:31] <goetsu> done
- # [15:33] <@yzen> ok great
- # [15:33] <@yzen> if you dont mind switch to command line right now as i dont know for sure how to do it in source tree
- # [15:33] <goetsu> ok
- # [15:33] <@yzen> and type git rebase -i upstream/master
- # [15:33] <goetsu> ok
- # [15:33] <goetsu> done
- # [15:33] <@yzen> paste here your list of commits
- # [15:34] <goetsu> pick bb2b3b7 Bug-1069614
- # [15:34] <goetsu> pick 8d539cb add label to cover without visible artiste/album text
- # [15:34] <goetsu> pick a3b6282 respect coding style
- # [15:34] <goetsu> pick 49c28a3 update
- # [15:34] <goetsu> pick 025adf2 cleaning up PR
- # [15:34] <goetsu> pick 0b6bdb6 Revert "respect coding style"
- # [15:34] <@yzen> cool
- # [15:34] * Joins: surkov (surkov@moz-v8g78c.cpe.teksavvy.com)
- # [15:34] * ChanServ sets mode: +o surkov
- # [15:34] <@yzen> so there are instructions underneath but i can explain what you need
- # [15:34] <@yzen> you need pick for the first one (pick bb2b3b7 Bug-1069614)
- # [15:35] <@yzen> all the other ones , switch the first word pick to s , that stands for squash
- # [15:35] <@yzen> also
- # [15:36] <&MarcoZ> Gijs: Hi! Not sure if you still look/work at ChatZilla, but if you do, you should also be aware of bug 1108568. :)
- # [15:36] <@firebot> https://bugzil.la/1108568 — NEW, nobody@mozilla.org — Runtime error "TypeError: client.currentObject is undefined @ <chrome://chatzilla/content/static.js>
- # [15:36] <@yzen> for the one we are picking, add a message, so instead of just Bug-1069614, write Bug 1069614 - making albums accessible in the main view. or something like that
- # [15:36] <@firebot> https://bugzil.la/1069614 — ASSIGNED, aurelien.levy@free.fr — Albums are not accessible in main view.
- # [15:36] <@yzen> goetsu^
- # [15:37] <goetsu> ok
- # [15:37] <@yzen> so what will happen is that all changes from the commits you listed will be there, but they will all show up as if they were done in the one we are picking
- # [15:37] <@yzen> mind you there might possibly be a conflict so watch the logs if it says rebase failed
- # [15:38] <@yzen> you would need to fix that before continuing
- # [15:38] <goetsu> but the update commit is still in the list
- # [15:38] <goetsu> and my wrong revert too
- # [15:39] <@yzen> right so we did not revert taht too
- # [15:40] <goetsu> it look like the sourcetree changes isn't seen
- # [15:40] <@yzen> goetsu hmm, i wonder if source tree repo is linked to the your original one, or another copy was made
- # [15:46] <goetsu> I will abort this command and try again
- # [15:48] <@yzen> goetsu sounds good, ill go offline for 15 min and will be right back
- # [15:48] * yzen is now known as yzen_
- # [15:53] * Quits: @yzen_ (yzen@moz-ql1r6k.cpe.pppoe.ca) (Ping timeout: 121 seconds)
- # [15:57] <@firebot> fred.wang@free.fr requested needinfo from jwei@jonathanwei.ca on bug 1001634.
- # [15:57] <@firebot> https://bugzil.la/1001634 — NEW, jwei@jonathanwei.ca — Add MathMLAccessible class
- # [16:01] * Joins: clown (clown@moz-v6m.c3o.211.205.IP)
- # [16:04] <goetsu> ok I don't find anyway to clean this so i'm stuck for now
- # [16:07] * Joins: yzen (yzen@moz-i5m.05u.207.66.IP)
- # [16:07] * ChanServ sets mode: +o yzen
- # [16:08] <@yzen> goetsu back
- # [16:09] <@yzen> goetsu see you have all commits revered now
- # [16:09] * Quits: clown (clown@moz-v6m.c3o.211.205.IP) (Quit: Leaving.)
- # [16:09] <@yzen> goetsu have you tried git rebase -i upstream/master yet ?
- # [16:10] <goetsu> yes but now I have
- # [16:10] <goetsu> pick c565996 Bug 1055357 remove role search and replace styling
- # [16:10] <goetsu> pick a535f0f BUG 1068713 add textbox
- # [16:10] <goetsu> pick 4086345 Revert c565996..a535f0f
- # [16:10] <@firebot> https://bugzil.la/1055357 — NEW, nobody@mozilla.org — Compose message form has a misleading role of "search"
- # [16:10] <@firebot> https://bugzil.la/1068713 — FIXED, nobody@mozilla.org — compose screen: email body should have role textbox
- # [16:10] <goetsu> pick 96582c5 Revert "Bug 1055357 remove role search and replace styling"
- # [16:12] * Joins: clown (clown@moz-v6m.c3o.211.205.IP)
- # [16:13] <goetsu> I think it's the mess in my local directory now
- # [16:16] * Joins: davidb (davidb@moz-5d550r.dsl.bell.ca)
- # [16:16] * ChanServ sets mode: +qo davidb davidb
- # [16:16] <~davidb> heyo
- # [16:18] <&MarcoZ> Heyo davidb!
- # [16:19] <~davidb> hi hi h i
- # [16:24] <~davidb> MarcoZ, how accessible is Trello? Is it usable?
- # [16:25] <@yzen> goetsu no worries
- # [16:26] <@yzen> we can try an alternative
- # [16:26] <@yzen> which should work
- # [16:26] <@yzen> so you are in your branch
- # [16:26] <@yzen> if you do git diff upstream/master, does it only show your necessary changes now ?
- # [16:27] <goetsu> no it show me some changes but not mine
- # [16:40] <&MarcoZ> davidb: The basics should be, like reading the texts and also creating new cards or so. Haven't actually ever had access to anything more advanced, so don't know if I could rearrange cards or such. It's...well...just another web app with only limited accessibility.
- # [16:40] <~davidb> MarcoZ, can you get info about the current column? Cards are arranged in columns.
- # [16:41] <&MarcoZ> davidb: I actually wouldn't know where to look for that. I am currently not part of any project on Trello.
- # [16:41] <&MarcoZ> Last time was a while ago while coordinating something for a podcast I participated in.
- # [16:41] <~davidb> yzen, can you add MarcoZ to the ffos a11y one?
- # [16:42] <@yzen> of course
- # [16:42] <&MarcoZ> yzen: I should probably get an account with my MoCo e-mail address first.
- # [16:42] <~davidb> yeah
- # [16:42] <@yzen> MarcoZ ill send you a linke
- # [16:42] <goetsu> is there a way to reset my local master branch to fresh source of the remote without breaking the open PR i've done using specific branch ?
- # [16:43] <@yzen> sent
- # [16:46] <@yzen> goetsu so what you can do is (since you have SHA for all the commits that you actually need) :
- # [16:47] <@yzen> goetsu: switch to you local master and make sure it's up to date with upstream
- # [16:47] <@yzen> goetsu make a new branch : git checkout -b tmp
- # [16:47] <@yzen> goetsu then do the following:
- # [16:47] <@yzen> git cherry-pick bb2b3b7
- # [16:48] <@yzen> git cherry-pick 8d539cb
- # [16:48] <goetsu> it's up to date but with 6 commits ahead due to previous wrong manipulation
- # [16:48] <@yzen> git cherry-pick a3b6282
- # [16:48] <@yzen> ok
- # [16:48] <@yzen> so then switch to branch other than master
- # [16:48] <@yzen> and do git branch -D master
- # [16:49] <&MarcoZ> davidb: Are the columns these things that have headings such as "Gaia To Do", "Accessibility" etc.?
- # [16:49] <goetsu> any other one or a new one ?
- # [16:49] <@yzen> any one existing
- # [16:49] <~davidb> MarcoZ, probably. cc yzen
- # [16:49] <@yzen> just temporarily
- # [16:49] <@yzen> then do: git fetch upstream master:master
- # [16:49] * ~davidb is perusing platform trello ATM
- # [16:50] * Joins: maxli (maxli@moz-fs8bmd.cs.uwaterloo.ca)
- # [16:50] <@yzen> MarcoZ you're admin now
- # [16:50] <goetsu> done
- # [16:50] <&MarcoZ> yzen: Thanks!
- # [16:51] <@yzen> goetsu ok great, now go back to that new master branch: git checkout master, it should be up to date
- # [16:51] <goetsu> Already on 'master'
- # [16:51] <@yzen> goetsu, then proceed starting from where i wrote: "goetsu make a new branch : git checkout -b tmp"
- # [16:51] <~davidb> surkov, are you following gecko <picture> implementation?
- # [16:53] * icaaq is now known as icaaq|afk
- # [16:53] <@yzen> goetsu last cherry pick after the 3 above: git cherry-pick 025adf2
- # [16:53] <goetsu> ok done
- # [16:53] <@yzen> this new tmp branch should now have just your 4 commits if all applied well
- # [16:54] <goetsu> yes
- # [16:54] <@yzen> goetsu great
- # [16:54] <@yzen> now
- # [16:54] <@yzen> we try our rebase :)
- # [16:54] <@yzen> git rebase -i master
- # [16:55] <goetsu> pick 9ef697b Bug-1069614
- # [16:55] <goetsu> pick 30ae477 add label to cover without visible artiste/album text
- # [16:55] <goetsu> pick 64ec1f1 respect coding style
- # [16:55] <goetsu> pick a329774 cleaning up PR
- # [16:55] <goetsu> pick the first and s for the oters ?
- # [16:55] <@yzen> goetsu, great, mark all except the top one with s instead of pick and change the message for the top one
- # [16:56] <@yzen> yes
- # [16:56] <@yzen> then proceed
- # [16:58] <@surkov> davidb: I didn’t, bug number?
- # [16:59] <goetsu> done
- # [16:59] <~davidb> surkov, e.g. bug 1017875
- # [16:59] <@firebot> https://bugzil.la/1017875 — ASSIGNED, john@pointysoftware.net — Pref on the picture element by default
- # [17:00] <@yzen> goetsu if you then do git log you will only see your new commit, does ?
- # [17:00] <@yzen> it look like that
- # [17:02] <goetsu> I see mine and three other one but not mine
- # [17:02] <goetsu> I see mine and three other one but not comming from me
- # [17:03] <goetsu> commit 65e97a2624638c726cce27f0b7af81952a5f312b
- # [17:03] <goetsu> Author: Douglas Sherk <github@sherk.me>
- # [17:03] <goetsu> commit 913580b0bf4caa5bf6dd2828aebf3e8387833a29
- # [17:03] <goetsu> Author: Doug Sherk <github@sherk.me>
- # [17:03] <goetsu> commit 1c636b94778d83cabccfa877400941ab4ef776f5
- # [17:03] <goetsu> Author: Ryan VanderMeulen <rvandermeulen@mozilla.com>
- # [17:04] <@yzen> that's fine
- # [17:04] <goetsu> ok
- # [17:04] <@yzen> goetsu just to make sure, can you push it so we dont loose anything ?
- # [17:05] <goetsu> done
- # [17:05] <@yzen> goetsu ok looks good
- # [17:05] <@yzen> so lets get rid of the old branch
- # [17:06] <@yzen> git branch -D Bug-1069614---Albums-are-not-accessible-
- # [17:06] <@yzen> git checkout -b Bug-1069614---Albums-are-not-accessible-
- # [17:06] <@yzen> git branch -D tmp
- # [17:06] <@yzen> git push -f origin Bug-1069614---Albums-are-not-accessible-
- # [17:08] <goetsu> done
- # [17:08] * Tomcat|sheriffduty is now known as Tomcat
- # [17:09] <@yzen> goetsu nice, so you can see now you only have one commit
- # [17:09] <@yzen> with the right changes
- # [17:09] * Tomcat is now known as Tomcat|afk
- # [17:09] <goetsu> \o/
- # [17:12] <@yzen> goetsu cool i think you can ask for a review now again, but i think you would still have to update the commit message before we put it in, if that;s ok with you
- # [17:12] <goetsu> ok
- # [17:15] <goetsu> How can i do that ?
- # [17:17] * Joins: slee (chatzilla@moz-93hjmp.range81-132.btcentralplus.com)
- # [17:18] <@firebot> New Core - Disability Access APIs bug 1108607 filed by jmathies@mozilla.com.
- # [17:18] <@firebot> https://bugzil.la/1108607 — NEW, nobody@mozilla.org — Touch input devices have e10s disabled due to accessibility
- # [17:26] <@yzen> goetsu so if you want to change last commit (which is the case here) simply do the following:
- # [17:27] <@yzen> (from your branch)
- # [17:27] <@yzen> git reset --soft 'HEAD^'
- # [17:27] <@yzen> then , optionally, make the changes to the commit, stage them (here not necessary)
- # [17:27] <@yzen> and then do
- # [17:28] <@yzen> git commit -c ORIG_HEAD
- # [17:28] <@yzen> and make sure to change the message up top to something like: Bug 1069614 - making albums accessible in the main view
- # [17:28] <@firebot> https://bugzil.la/1069614 — ASSIGNED, aurelien.levy@free.fr — Albums are not accessible in main view.
- # [17:28] <@yzen> and then do git push -f origin {your branch name}
- # [17:31] * Quits: slee (chatzilla@moz-93hjmp.range81-132.btcentralplus.com) (Ping timeout: 121 seconds)
- # [17:51] <goetsu> after git reset nothing happen
- # [17:52] * Quits: &MarcoZ (marco.zehe@moz-tb35md.dip0.t-ipconnect.de) (Quit: Curing jetlag...)
- # [17:53] <~davidb> surkov, hi
- # [17:54] <goetsu> in the app the commit appear as uncommited
- # [17:55] <goetsu> I can change the name in the app or how to change it with command line ?
- # [17:55] <@surkov> hi, davidb
- # [17:56] <@yzen> goetsu lets see, so did you do any of the steps starting from git reset --soft 'HEAD^' (in terminal)
- # [17:56] <@yzen> ?
- # [17:56] <goetsu> i've done the git reset
- # [17:57] <@yzen> ok yes
- # [17:57] <@yzen> after that step your changes will look like they are uncommitted
- # [17:57] <goetsu> yes
- # [17:57] <@yzen> so all you need to do now, in terminal is the git commit command
- # [17:57] <@yzen> and you ll get another prompt to change the commit message, the same one you'd get if you were just committing
- # [17:58] <@yzen> e.g. git commit -c ORIG_HEAD
- # [17:59] <goetsu> ok but for info in the app there is two others commits who appear I don't know why
- # [18:00] <goetsu> it look to come from previous commits in other branches I don't know why they show up here
- # [18:00] <@yzen> i m not entirely sure what state the app shows, but i would trust the terminal that if all goes well , you will only have your one commit on top
- # [18:00] <goetsu> ok so i do git commit -c ORIG_HEAD ?
- # [18:01] <goetsu> only one appear
- # [18:01] <goetsu> Bug-1069614 final cleanup off PR
- # [18:01] <goetsu> I change it to Bug 1069614 - making albums accessible in the main view
- # [18:01] <@firebot> https://bugzil.la/1069614 — ASSIGNED, aurelien.levy@free.fr — Albums are not accessible in main view.
- # [18:02] <@yzen> goetsu nice
- # [18:02] <@yzen> i think that should do it
- # [18:02] <goetsu> done
- # [18:02] <@yzen> goetsu nice, after that just do git push -f origin Bug-1069614---Albums-are-not-accessible-
- # [18:03] <@yzen> to update the remote branch
- # [18:03] <goetsu> done
- # [18:03] <@yzen> goetsu nice : https://github.com/mozilla-b2g/gaia/pull/26568/commits
- # [18:04] <@yzen> goetsu mark your pull request attachment with r? for jim porter and a11y-r? for me, ill test it on device then
- # [18:06] <goetsu> done too
- # [18:08] <@yzen> goetsu \o/ thanks!
- # [18:08] <goetsu> thanks for your patience
- # [18:09] <@yzen> no worries, the git workflow is a little tricky, i still dont remember many things myself, and rely on terminal history big time :)
- # [18:09] <goetsu> I think I will ask you again for the next one for the rebase thing hopping this time I will have less trouble
- # [18:14] <@yzen> goetsu any time
- # [18:26] <@yzen> goetsu are you familiar with python ?
- # [18:30] <goetsu> not much I'm more a front end a11y expert than backend
- # [18:32] * Quits: maxli (maxli@moz-fs8bmd.cs.uwaterloo.ca) (Quit: Leaving.)
- # [18:32] <~davidb> tbsaunde, does it make sense to assign bug 1098386 to you?
- # [18:32] <@firebot> https://bugzil.la/1098386 — NEW, nobody@mozilla.org — talos a11y crashes when run in e10s mode
- # [18:32] * Joins: slee (chatzilla@moz-93hjmp.range81-132.btcentralplus.com)
- # [18:32] <goetsu> and most of the time I do only audit, teaching and consulting on a11y not the fix myself so still a bit rusty as you see with git and javascript
- # [18:38] <tbsaunde> davidb: sure
- # [18:41] <~davidb> thnx
- # [18:47] <@yzen> goetsu one thing we try to do on gaia atm is to write some ui integration tests to make sure our a11y fixes do not regress, they are however written in python (though there are plenty of examples already). would you be interested to try for this bug ?
- # [18:48] <goetsu> I could try
- # [18:49] <@yzen> goetsu i think we can actually open a follow up bug for that to not stop this one from getting in
- # [18:51] <joanie> surkov: ping
- # [18:54] <@surkov> joanie: pong
- # [18:54] <joanie> surkov: Hey, so.... subroles! coming to Atk. :)
- # [18:54] <joanie> Piñeiro does not want AtkRole explosion
- # [18:54] <@surkov> makes sense
- # [18:54] <joanie> and what Apple does for AX API seems to work
- # [18:54] <joanie> cool
- # [18:55] <joanie> so later this week -- maybe even tomorrow
- # [18:55] <@surkov> I don’t like strings but it seems nobody invented any better
- # [18:55] <@surkov> and it’s extensible
- # [18:55] <joanie> We'll they will not be strings
- # [18:55] <joanie> they will be added to an enum
- # [18:55] <@surkov> ok L)
- # [18:55] <joanie> just like is done for AtkRole
- # [18:55] <@surkov> that’s c++ish :)
- # [18:55] <joanie> The exact API remains to be seen
- # [18:55] <@surkov> is there proposal?
- # [18:56] <joanie> ish, yes
- # [18:56] <@surkov> cool
- # [18:56] <joanie> lemme find the bug
- # [18:56] <@surkov> is it subroles or many roles?
- # [18:56] <joanie> it's not a full proposal yet
- # [18:56] <@surkov> that’s ok
- # [18:56] <joanie> there's not a patch yet
- # [18:56] <joanie> but ....
- # [18:56] * joanie looks
- # [18:56] <@surkov> I need make sure I cath the idea
- # [18:56] <joanie> https://bugzilla.gnome.org/show_bug.cgi?id=739140 is the full thing
- # [18:57] <joanie> btw, it's not really going to be atk_role_stuff_...
- # [18:58] <joanie> Piñeiro and I didn't know what to call it
- # [18:58] <joanie> but the idea is that there will be an atk_object_get_role_foo()
- # [18:58] <joanie> and an AtkRoleFoo
- # [18:58] <@surkov> what is Foo?
- # [18:58] <joanie> stuff
- # [18:58] <joanie> info
- # [18:59] <joanie> name we don't yet have
- # [18:59] <joanie> the problem is we have one free pad left for AtkObject
- # [18:59] <joanie> and we need both subrole and localized role name
- # [18:59] <@surkov> like atk_object_get_subrole?
- # [18:59] <joanie> the latter for ARIA
- # [18:59] <joanie> surkov: that was the original plan
- # [18:59] * icaaq|afk is now known as icaaq
- # [19:00] <joanie> but we need subrole and we need (for aria) the ability for a "customized" "localized" role name
- # [19:00] <joanie> the latter I'm not jazzed about
- # [19:00] <tbsaunde> why does localizing the string need to happen on our side? its the same strings for all apps right?
- # [19:00] <joanie> but others are
- # [19:00] <joanie> no sir
- # [19:00] <joanie> the idea is:
- # [19:00] <joanie> if we have "chapter" role (dpub)
- # [19:01] <joanie> but the publishing industry wants jaws to say "canto" instead of "chapter"
- # [19:01] <tbsaunde> where chapter is something atk doesn't now about
- # [19:01] <joanie> but just in Dante's Inferno....
- # [19:01] <joanie> there will be a chapter subrole
- # [19:01] <joanie> this came up at TPAC btw
- # [19:02] <tbsaunde> I'm very tempted to just say no
- # [19:02] <joanie> anyhoo, the idea is not role registration or role replacement
- # [19:02] <joanie> it's a way to provide an speakable string without stomping on a role
- # [19:02] <joanie> done by authors via aria
- # [19:03] <joanie> but only in specific cases
- # [19:03] <joanie> apparently AX API supports it, UIA supports it, MSAA supports it
- # [19:03] <joanie> and a lot of people want it
- # [19:03] <joanie> so we're going to be stuck with it
- # [19:03] <@surkov> are there implementations?
- # [19:03] <joanie> not yet because I don't think it's in the spec *yet*
- # [19:04] <tbsaunde> well, afaik msaa just uses strings so I think you stomp on the "real" role?
- # [19:04] <@surkov> I see
- # [19:04] <joanie> tbsaunde: I am not sure
- # [19:04] <joanie> but the MS person is fine with it
- # [19:04] <tbsaunde> joanie: doesn't mean I have to be fine with us implementing it ;)
- # [19:04] <joanie> surkov: anyhoo, for now you would just return an empty/null value for localized role name
- # [19:04] <@surkov> yeah, that’s what we do for other APIs :)
- # [19:05] <joanie> tbsaunde: then you can continue to return an empty/null value
- # [19:05] <joanie> what I really need is subroles
- # [19:05] <joanie> and the only reason we are now talking about this other crap is because I wanted to explain why it is probably not going to be atk_object_get_subrole
- # [19:05] <joanie> but if making it atk_object_get_subrole would make you give me subroles
- # [19:05] <joanie> I'll make it that
- # [19:06] <joanie> :)
- # [19:06] <joanie> so are we ok with subroles tbsaunde ?
- # [19:06] <@surkov> I still don’t follow about naming
- # [19:06] <joanie> naming what?
- # [19:06] <@surkov> subrole vs role_stuff
- # [19:06] <~davidb> (the MSAA role hack is because callers can check variant.vt == VT_BSTR)
- # [19:07] <joanie> we have one free pad
- # [19:07] <joanie> in AtkObject
- # [19:07] <joanie> if we want to have atk_object_get_subrole()
- # [19:07] <joanie> AND
- # [19:07] <joanie> atk_object_get_localized_rolename()
- # [19:07] <joanie> and we have one free pad
- # [19:07] <joanie> we cannot have both
- # [19:07] <joanie> furthermore
- # [19:07] <@surkov> it sounds like you have to deal with it sooner or later
- # [19:08] <joanie> why have multiple calls (atk_object_get_role(), atk_object_get_subrole(), and atk_object_get_localized_rolename()
- # [19:08] <joanie> ?
- # [19:08] <tbsaunde> joanie: I have no objection to the idea of subroles
- # [19:08] <@surkov> I meant you will need a new mehtod in a while
- # [19:08] <joanie> we could instead have atk_object_get_all_the_freaking_role_crap()
- # [19:08] <@surkov> while you evolve
- # [19:08] <joanie> which would provide role, subrole, and localized rolename (aka NULL)
- # [19:08] <joanie> ;)
- # [19:09] * Quits: slee (chatzilla@moz-93hjmp.range81-132.btcentralplus.com) (Ping timeout: 121 seconds)
- # [19:09] <~davidb> heh
- # [19:09] <@surkov> one call I think is preferable if you need to fetch all at once
- # [19:09] <joanie> davidb: you approve of my new method name?
- # [19:09] <~davidb> sure
- # [19:09] <@yzen> goetsu bug 1108662 if you are interested :)
- # [19:09] <@firebot> https://bugzil.la/1108662 — NEW, nobody@mozilla.org — Write gaia-ui tests for click on album from main view.
- # [19:10] <~davidb> joanie, subrole?
- # [19:10] <joanie> surkov: ok, one call it is. And david says I can include the word "crap" in the method.
- # [19:10] <joanie> davidb: like AX API has
- # [19:10] <@surkov> joanie: yeah, I think I would name it internally somehow else :)
- # [19:10] <joanie> Piñeiro does not want role explosion
- # [19:10] <~davidb> wait a minute :)
- # [19:10] <~davidb> oh atk_object_get_all_the_freaking_role_crap, yeah perfect.
- # [19:11] <joanie> davidb++
- # [19:11] <joanie> ;)
- # [19:11] <~davidb> :)
- # [19:11] <goetsu> is there an easy way to know what file to change when you have found the bug in the source code ?
- # [19:12] <~davidb> joanie, is the idea that subrole is easily extensible?
- # [19:12] <~davidb> (custom roles)
- # [19:12] <goetsu> for exemple I try to find where is the <span class="remove"></span> used on the homescreen app in edit mode to remove the app
- # [19:12] <~davidb> or just that roles.h is too big
- # [19:12] <@surkov> by Piñeiro yes :)
- # [19:12] <~davidb> :)
- # [19:12] <~davidb> roles1.h roles2.h ...
- # [19:13] <joanie> davidb: easily, no more so than role
- # [19:13] <~davidb> poop
- # [19:13] <joanie> the idea is that we make role a more general (smaller set)
- # [19:13] <joanie> and then refine via subrole
- # [19:13] <joanie> so given ATK_ROLE_TABLE
- # [19:13] <joanie> subroles might be:
- # [19:14] <~davidb> i wonder if AT would just end up always checking
- # [19:14] <joanie> ATK_SUBROLE_GRID, and
- # [19:14] <@surkov> that’s good for web/app extensibility
- # [19:14] <@surkov> when no new widgets but all widgets gets more specialized
- # [19:14] <joanie> ATK_SUBROLE_LAYOUT_TABLE
- # [19:15] <joanie> and ATK_SUBROLE_DATA_TABLE
- # [19:15] <joanie> and ...
- # [19:15] <~davidb> ATK_ROLE_COMBOBOX, ATK_SUBROLE_FUGLY_CAROUSEL
- # [19:15] <~davidb> COMBOBOX/RADIOGROUP
- # [19:15] <joanie> for buttons:
- # [19:15] <joanie> role: button
- # [19:15] <@yzen> goetsu im afraid it's based on what your editor can do, i use sublime so i would search for .remove or even remove across the whole apps/verticalhome
- # [19:16] <joanie> subroles: push, toggle, checkbox, radiobutton, ....
- # [19:16] <joanie> also color chooser, file upload, ....
- # [19:16] <~davidb> ok
- # [19:16] <joanie> if you think of all the new HTML input types
- # [19:16] <~davidb> what about control patterns?
- # [19:16] <joanie> do we really want new entry roles for all of those?
- # [19:17] <joanie> you mean UIA stuff?
- # [19:17] <~davidb> yeah
- # [19:17] <joanie> or AtkAction?
- # [19:17] <~davidb> no
- # [19:17] <joanie> we're thinking about looking at what UIA does, but it won't happen this release cycle
- # [19:17] <joanie> (if I had to bet)
- # [19:17] <~davidb> i guess a larger question... should we step back and look at all the desktop APIs, ARIA, IndieUI and go drinking
- # [19:18] <tbsaunde> davidb: we shouldn't look and go directly to the gas station
- # [19:18] <joanie> yes to booze. I need to work on the stuff I'm in spain for.
- # [19:18] <tbsaunde> (to get gas to burn it all down with)
- # [19:19] <~davidb> oh right you are in spain
- # [19:19] <~davidb> tbsaunde :)
- # [19:19] <~davidb> ok well getting surkov's thumb up is important
- # [19:19] <~davidb> (if we want convergence)
- # [19:19] <goetsu> that what I do but something like more than 700 results :(
- # [19:19] <~davidb> (and i know you know this and i am interrupting)
- # [19:20] * icaaq is now known as icaaq|afk
- # [19:21] <@surkov> I think I like the idea of subroles. I would show the proposal to Jamie to see if it’s something we want on IA2
- # [19:23] <fredw> Hi everybody. I'm also in Spain with Joanie. The other thing we discussed was about creating "fake accessibles" for the MathML scripted elements.
- # [19:23] <fredw> I think tbsaunde said yesterday that it would be doable
- # [19:24] <~davidb> mphantom?
- # [19:24] <~davidb> or is that something different?
- # [19:25] <fredw> davidb: no not mphantom. For mphantom and other invisible things, we probably don't want to expose them at all (at least for ATK)
- # [19:25] <~davidb> ok
- # [19:27] <fredw> It's for example for <msubsup>BASE SUBSCRIPT SUPERSCRIPT</msubsup> and <msup>BASE SUPERSCRIPT</msup>. We want to create only a single "scripted" role for the two elements and create a fake empty "subscript" for the latter so that they both have the same structure
- # [19:27] <fredw> http://krijnhoetmer.nl/irc-logs/accessibility/20141207#l-64
- # [19:37] <fredw> davidb surkov tbsaunde these are the notes taken by joanie:
- # [19:37] <fredw> Had a very long, but ultimately productive, discussion regarding
- # [19:37] <fredw> script-related elements. Under consideration: subrole-only approach,
- # [19:37] <fredw> using AtkRelation, fake accessible/parent usage.
- # [19:37] <fredw> The problem with <mmultiscripts> is primarily its significant
- # [19:37] <fredw> reliance upon DOM child position:
- # [19:37] <fredw> * first child (index 0) is the base
- # [19:37] <fredw> * odd-indexed children prior to <mprescripts> are post subscripts
- # [19:37] <fredw> * even-indexed children prior to <mprescripts> are post superscripts
- # [19:37] <fredw> * to preserve the above indexing, <none /> is used for missing post
- # [19:37] <fredw> {sub,super}scripts
- # [19:37] <fredw> * the presence of <none /> means you have useless/empty accessibles
- # [19:37] <fredw> * <mprescripts> cannot have children so the element is a marker
- # [19:37] <fredw> * if you do not expose <mprescripts>, the missing marker means ATs
- # [19:37] <fredw> don't know when post ends and pre begins
- # [19:37] <fredw> * if you do expose <mprescripts>, the even/odd indexing is reversed
- # [19:37] <fredw> for prescripts (odd == pre superscript; even == pre subscript)
- # [19:37] <fredw> * related aside, there is no <mpostscripts>
- # [19:37] <fredw> Given the above, one (ugly) option:
- # [19:37] <fredw> * ATK_SUBROLE_MATH_SCRIPTED (there would be no sub/sup subrole)
- # [19:37] <fredw> * base is always first child
- # [19:37] <fredw> * New AtkRelation types (multiple targets with some targets possibly
- # [19:37] <fredw> being null):
- # [19:37] <fredw> - ATK_RELATION_PRE_SUBSCRIPT
- # [19:37] <fredw> - ATK_RELATION_PRE_SUPERSCRIPT
- # [19:37] <fredw> - ATK_RELATION_POST_SUBSCRIPT
- # [19:37] <fredw> - ATK_RELATION_POST_SUPERSCRIPT
- # [19:37] <fredw> An alternative solution is:
- # [19:37] <fredw> * Expose two fake accessibles: One for pre, one for post.
- # [19:37] <fredw> * The children of these fakes would be the actual pre and posts
- # [19:37] <fredw> * Subroles: scripted, prescripts, postscripts
- # [19:37] <fredw> * Additional fake accessibles for missing sub/sup
- # [19:37] <~davidb> feel free to http://pastebin.mozilla.org/
- # [19:37] <joanie> guys, all the above notes are not what you want/need
- # [19:38] <joanie> the problem with multiscripts is all you care about
- # [19:38] <joanie> sorry for the noise
- # [19:38] <@surkov> fredw: do you have example of it somewhere
- # [19:38] <fredw> surkov: example of mmultiscripts?
- # [19:38] <@surkov> yes
- # [19:38] <fredw> surkov: https://developer.mozilla.org/en-US/docs/Web/MathML/Element/mmultiscripts
- # [19:39] <joanie> the tl;dr version is: the MathML <mprescripts /> screws everything up
- # [19:39] <joanie> and the MathML spec should be fixed
- # [19:39] <joanie> but until that happens.... ;)
- # [19:40] <@surkov> is there proposal for mahtmL somwhere?
- # [19:40] <fredw> surkov: proposal for MathML ATK support?
- # [19:40] <@surkov> no for changing mathml spec agains mprescripts stuff
- # [19:41] <fredw> surkov: no
- # [19:41] <@surkov> I’m curious if that’s somethign that should be really changed
- # [19:41] <joanie> not yet
- # [19:41] <joanie> I plan to email the list soon (MathML WG list)
- # [19:41] <@surkov> ok, good
- # [19:41] <joanie> I discovered this horrible situation yesterday
- # [19:41] <joanie> and fredw doesn't think it will change overnight
- # [19:41] <@surkov> what’s wrong with it?
- # [19:41] <@surkov> it wont’ I bet
- # [19:41] <fredw> surkov: at least for the visual rendering it does not really matter because we browse the whole list of children to place the scripts anyway. That might be different for a11y
- # [19:42] <joanie> lemme pastebin just the relevant parts of the notes
- # [19:42] <@surkov> it sounds like relations approach works fine, no?
- # [19:42] <joanie> surkov: http://fpaste.org/157733/
- # [19:43] <joanie> surkov: We do NOT need relations for anything other than -- potentially -- mmultiscripts
- # [19:43] <joanie> and if we do NOT do the inverse/reciprocal AtkRelations
- # [19:43] <joanie> but just the one-way ones
- # [19:43] <joanie> bare minimum
- # [19:43] <joanie> we have 4 new relations
- # [19:43] <joanie> which we have to implement in Atk
- # [19:43] <joanie> and Atspi
- # [19:43] <joanie> just for this stupid broken spec
- # [19:44] * Joins: slee (chatzilla@moz-93hjmp.range81-132.btcentralplus.com)
- # [19:44] <@surkov> technically you can traverse the tree without relations?
- # [19:44] <joanie> surkov: when we have proper roles and subroles
- # [19:44] <@surkov> ok
- # [19:44] <joanie> which is not, btw ATK_ROLE_UNKNOWN
- # [19:45] <joanie> any more than it's ATK_ROLE_TEXT ;)
- # [19:45] <@surkov> :)
- # [19:45] <joanie> i.e. the intern's patch might need some fixin'
- # [19:45] <@surkov> sure, I just finally realized that we need to have a base and then polish it
- # [19:46] <@surkov> I wasn’t really sure about tree hierarchy approach
- # [19:46] <@surkov> I liked to see some text based stuff
- # [19:46] <@surkov> like map everything into offsets and then provide sets of attributes for the given offset
- # [19:46] <@surkov> like describe where I am
- # [19:48] <joanie> surkov: not sure I follow you
- # [19:48] <@surkov> joanie: I meant I liked to see different kind of API for mathml
- # [19:49] <joanie> like....
- # [19:49] <@surkov> like no accessibles for mo mb stuff
- # [19:49] <@surkov> just say you deal with raw data x+7=something
- # [19:49] <joanie> they might ultimately be "foldable" like <span>
- # [19:50] <joanie> but until we know that, I think they are ATK_ROLE_STATIC
- # [19:50] <joanie> that is what we're going to do in WebKitGtk
- # [19:50] <@surkov> I would write here a more comlex expression if I was allowed to :)
- # [19:50] <joanie> later we might flatten
- # [19:50] <@surkov> but don’t you need as AT to know this is an operator
- # [19:50] <fredw> surkov: like <mfrac><mi>x</mi><mn>7</mn></mfrac>
- # [19:50] <@surkov> yeah
- # [19:51] <joanie> fredw: good example
- # [19:51] <joanie> so what do we do with the x and 7?
- # [19:51] <joanie> and how do we speak the localized fraction line?
- # [19:51] <@surkov> you would have one accessible object for all of that, but then asking for offset stuff you would receive you’re in fract, numberator, at x
- # [19:51] <@surkov> something like that
- # [19:52] <joanie> surkov: so AtkMath iface?
- # [19:52] <@surkov> yeah, that could be
- # [19:52] <@surkov> it would be close to text attributes stuff
- # [19:52] <joanie> maybe....
- # [19:52] <@surkov> but I never moved far on this way
- # [19:53] <joanie> uhhhh.... text attributes?
- # [19:53] <@surkov> sort of, just bunch of description for some offsets
- # [19:53] <joanie> that sounds like AtkHyperlinkImpl
- # [19:53] <joanie> and friends
- # [19:53] <@surkov> anyway it sounds we deal with hierarchical approach
- # [19:53] <joanie> aka the worst API ever
- # [19:54] <joanie> at least for now
- # [19:54] <joanie> I'm interested in a possible AtkMath iface
- # [19:54] <@surkov> I can try to wrtie down the idea
- # [19:55] <@surkov> at least to see if it’s reaonable on the paper :)
- # [19:55] <joanie> but anything that resembles what happens with hypertext now in Atk (and object attributes and child offsets and ....) sounds icky
- # [19:55] <@surkov> still reaosnable :)
- # [19:55] <joanie> brand new iface sounds nice
- # [19:55] <joanie> ;)
- # [20:00] <joanie> surkov: so back to the fake accessibles.... ;)
- # [20:00] <@surkov> ok, what is that?
- # [20:00] <joanie> the <mprescripts> really should be the parent of the prescripts
- # [20:01] <joanie> right now it's an element that is not rendered and functioning as a marker
- # [20:01] <@surkov> is fake accessible supposed to be a replacement of relations?
- # [20:02] <joanie> not quite
- # [20:02] <joanie> functionally the <mprescripts /> says: "The elements which follow are presecripts"
- # [20:03] <joanie> the fact that it's not <mprescripts> my prescripts here</mprescripts> completely boggles my mind
- # [20:03] <joanie> and means that the child ordering is different before and after that element
- # [20:04] <joanie> before the element, odd indexed children are subscripts and even indexed children are superscripts
- # [20:04] <joanie> after the element, odd indexed children are superscripts and even indexed children are subscripts
- # [20:04] <joanie> thanks MathML WG!
- # [20:04] <joanie> :P
- # [20:04] <@surkov> it doesn’t look so from example https://developer.mozilla.org/en-US/docs/Web/MathML/Element/mmultiscripts
- # [20:05] <@surkov> the order is the same or you talk about something else?
- # [20:05] <@surkov> I mean d and b are subsripts tehre
- # [20:05] <@surkov> they are both at 1st index
- # [20:06] <@surkov> markup is rather crazy I agree
- # [20:06] <@surkov> but I still dint’ cath fake accesisbles idea
- # [20:06] <joanie> surkov: look at how Gecko exposes Example 1
- # [20:06] <joanie> child at index 0 is the base
- # [20:07] <joanie> child at index 1 is d (sub)
- # [20:07] <joanie> child at index 2 is c (sup)
- # [20:07] <joanie> child at index 3 is the prescripts
- # [20:07] <joanie> child at index 4 is b (sbu)
- # [20:07] * Quits: goetsu (chatzilla@moz-tj6t2q.rev.sfr.net) (Connection closed)
- # [20:07] <joanie> child at index 5 is a (sup)
- # [20:07] <joanie> so subs are 1 and 4
- # [20:07] <joanie> sups are 2 and 5
- # [20:08] <joanie> what I think example 1 should really have is
- # [20:08] <@surkov> right
- # [20:08] <joanie> ...
- # [20:08] <joanie> <mprescripts>
- # [20:08] <joanie> <mi>b</mi>
- # [20:08] <joanie> <mi>a</mi>
- # [20:09] <joanie> </mprescripts
- # [20:09] <joanie> >
- # [20:09] <joanie> if that were how it worked, the prescripts would be children of the mprescripts element
- # [20:09] <tbsaunde> joanie: its rather odd, but it does have the advantage of being easier to parse
- # [20:09] <joanie> you'd have a groupy-role for the mprescripts
- # [20:09] <joanie> and the sub, sup order would be consistent
- # [20:09] <tbsaunde> well, maybe not if you require the mprescripts to be the last child I guess
- # [20:10] <@surkov> joanie: I still not sure what is plan a) file proposal to change mathml or b) try to find ways to map exisitng mathml inot a11y
- # [20:11] <joanie> it's a+b
- # [20:11] <joanie> I'm going to do A
- # [20:11] <joanie> fredw: thinks it will be a hard (impossible) sell
- # [20:11] <joanie> in the meantime, I think we can try to make the a11y tree look like what the MathML tree should look like but fails to look like ;)
- # [20:12] <joanie> namely a parent element (mprescripts) with children (the prescripts)
- # [20:12] <joanie> until MathML WG does my bidding, we can (I think) create a fake accessible or a fake parent
- # [20:13] <@surkov> I would avoid changing a11y tree so hard
- # [20:14] <joanie> yeah, I know
- # [20:14] <joanie> that was my initial reaction
- # [20:14] <@surkov> ok :)
- # [20:14] <joanie> but then we looked at the *minimum* required relations
- # [20:14] <joanie> and it was 4
- # [20:14] <joanie> to hack around this specific MathML Spec bug
- # [20:14] <@surkov> still, relations are much easiser to implement I think
- # [20:14] <joanie> if you want to sell those to Piñeiro go for it
- # [20:15] <joanie> ;)
- # [20:15] <@surkov> :)
- # [20:15] <@surkov> I know he fights for spec purity
- # [20:15] <tbsaunde> surkov: what do you do about multiple pre super scripts say?
- # [20:15] <joanie> plus, the target list of these relations would potentially be null AtkObjects
- # [20:15] <@surkov> but if it makes implementation a nightmare it’s probably not worth it
- # [20:16] <joanie> so it's all sorts of crazy extra work to hack around a single, and not widely used, spec
- # [20:16] <@surkov> tbsaunde: sorry, what do you mean?
- # [20:16] <tbsaunde> I have no clue what that looks like on screen, or why you'd support it in the first place
- # [20:16] <@surkov> joanie: maybe keep it inaccessible for a while :) at least unitl mathml is fixed the way you like?
- # [20:17] <joanie> tbsaunde: on screen it looks as bad as it sounds here. But lemme try:
- # [20:17] <joanie> given a base X
- # [20:17] <joanie> a presuperscript a
- # [20:17] <joanie> a presubscript b
- # [20:17] <joanie> a post superscript c
- # [20:17] <joanie> and a post subscript d
- # [20:17] <joanie> on screen you have:
- # [20:18] <joanie> X in the middle. A in the upper left, B in the bottom left, C in the upper right, D in the lower right.
- # [20:18] <joanie> make (some) sense?
- # [20:18] <@surkov> it sounds with me those scripts are sort of common stuff in math
- # [20:18] <tbsaunde> joanie: yeah, that makes sense
- # [20:18] <@surkov> when math run out of letters :)
- # [20:18] <@surkov> symbols
- # [20:18] <tbsaunde> joanie: I wonder about case where you have multiple pre super scripts
- # [20:18] <joanie> it gets uglier
- # [20:18] <tbsaunde> which this spec seems to allow
- # [20:18] <joanie> fredw: remind me where the second and third prescripts go
- # [20:20] <joanie> the subsequent post ones definitely extend to the right
- # [20:20] <joanie> so with c and d being the first post super and first post sub respectively,
- # [20:20] <joanie> the second post super C2 would be immediately to the right of C
- # [20:20] <joanie> the second post sub D2 would be immediately to the right of D
- # [20:21] <tbsaunde> joanie: ok, I guess that kind of makes sense
- # [20:21] <joanie> I forget which way the pres extend, and and fredw isn't sure either so he's making a test case so we can look
- # [20:21] <tbsaunde> joanie: doesn't really matter
- # [20:21] <joanie> ok
- # [20:21] <joanie> like I said, it's as ugly spatially as it is spec-y
- # [20:22] <tbsaunde> I'm confused that "a b" as one superscript isn't enough and you need to be able to have "a" and "b" as separate scripts, but anyway it exists
- # [20:22] <joanie> a is a superscript
- # [20:22] <joanie> b is a subscript
- # [20:22] <tbsaunde> which seems problematic for relations idea since AT might want to know order of pre superscripts
- # [20:22] <joanie> well, we did figure that out, but it's sad
- # [20:23] <joanie> btw, you can have empty scripts
- # [20:23] <joanie> like no a, but still b, c, and d
- # [20:23] <tbsaunde> sure
- # [20:23] <joanie> and because it relies upon element child order
- # [20:23] <joanie> missing scripts are <none />
- # [20:23] <joanie> so the AtkRelation approach needs multiple targets (already supported)
- # [20:24] <tbsaunde> yeah, I wasn't really talking in markup that was too much effort with this syntax, and I don't think tex can express this
- # [20:24] <joanie> with the target order being precise and corresponding with the element order
- # [20:24] <joanie> not sure if that's guaranteed, but I think it might be in practice
- # [20:24] <joanie> and some targets can be null
- # [20:24] <tbsaunde> yea, I guess its doable, but seems rather ugly
- # [20:25] <joanie> and at that point, we have four new relations, target order matters, targets can be null, and all because the MathML WG was smoking crack when they define <mprescripts>
- # [20:25] <joanie> or.... we can have a .... (wait for it) single fake parent
- # [20:25] <tbsaunde> joanie: well, its just an empty target set, which already exists
- # [20:25] <joanie> not target set
- # [20:26] <joanie> single target at a specified index
- # [20:26] <joanie> i.e. if you have 4 post super scripts
- # [20:26] <joanie> and the 3rd post super script is empty
- # [20:26] <joanie> getting the target at index 2 would return a null AtkObject
- # [20:27] <tbsaunde> hm, wouldn't you try to get RELATION_PRE_SUPERSCRIPT_OF and get the empty set?
- # [20:27] <joanie> no
- # [20:27] <joanie> well
- # [20:27] <joanie> what if we have 3 pre superscripts
- # [20:27] <joanie> and the 2nd one is null
- # [20:28] <tbsaunde> I'm not sure this is worth discussing, but why do you need the null?
- # [20:28] <joanie> btw the super script and sub script pairs on screen are laid out spatially
- # [20:28] <joanie> so let's just talk about post scripts
- # [20:28] <joanie> and say we have two pair
- # [20:28] <tbsaunde> yeah, just though of needing it for alignment info
- # [20:28] <joanie> correct
- # [20:28] <joanie> but they get elements
- # [20:29] <joanie> and if where the superscript is matters
- # [20:29] <joanie> the superscript at position 3 needs to be at position 3
- # [20:29] <joanie> not at position 2
- # [20:29] <joanie> so if position 2 is empty
- # [20:29] <tbsaunde> I'm not absolutely sure, the code to update trees is kind of complex, but I think the fake parent stuff should be much easier
- # [20:29] <joanie> ok
- # [20:29] <joanie> I think it will be as well
- # [20:30] <joanie> and I think that convincing Piñeiro to add four relation types to hack around the MathML spec is not going to go well
- # [20:30] <tbsaunde> I just wish we could go back 20 years and find a way to use tex instead of inventing mathml
- # [20:30] <joanie> and if the MathML WG does what I think they should, the end result will match the proposed accessible hierarchy accomplished via fake parent
- # [20:30] <joanie> which means that MathML fixing their bug won't break ATs
- # [20:32] * Quits: slee (chatzilla@moz-93hjmp.range81-132.btcentralplus.com) (Ping timeout: 121 seconds)
- # [20:41] * Quits: fredw (Icedove@moz-lnbnpf.igalia.com) (Client exited)
- # [21:17] * agibson is now known as agibson|afk
- # [21:20] * agibson|afk is now known as agibson
- # [21:21] * agibson is now known as agibson|afk
- # [21:28] * Quits: ~davidb (davidb@moz-5d550r.dsl.bell.ca) (Quit: Blah blah blah)
- # [21:31] * Joins: kalli_ (fabian@moz-ifjfid.res.insa-lyon.fr)
- # [21:32] * icaaq|afk is now known as icaaq
- # [21:35] * Joins: icaaq_ (icaaq@moz-n1f1o3.cust.bredbandsbolaget.se)
- # [21:35] * Quits: kalli_ (fabian@moz-ifjfid.res.insa-lyon.fr) (Ping timeout: 121 seconds)
- # [21:36] * Quits: icaaq (icaaq@moz-n1f1o3.cust.bredbandsbolaget.se) (Ping timeout: 121 seconds)
- # [21:43] * Joins: davidb (davidb@moz-5d550r.dsl.bell.ca)
- # [21:43] * ChanServ sets mode: +qo davidb davidb
- # [22:03] * icaaq_ is now known as icaaq|afk
- # [22:05] * davidb is now known as davidb|afk
- # [22:10] * icaaq|afk is now known as icaaq_
- # [22:16] * davidb|afk is now known as davidb
- # [22:44] * Parts: clown (clown@moz-v6m.c3o.211.205.IP)
- # [22:50] * Quits: @surkov (surkov@moz-v8g78c.cpe.teksavvy.com) (Client exited)
- # [22:54] * agibson|afk is now known as agibson
- # [22:59] * Quits: icaaq_ (icaaq@moz-n1f1o3.cust.bredbandsbolaget.se) (Quit: 10-79 Notify coroner, over and out)
- # [23:02] * davidb is now known as davidb|afk
- # [23:02] * Joins: surkov (surkov@moz-v8g78c.cpe.teksavvy.com)
- # [23:02] * ChanServ sets mode: +o surkov
- # [23:10] * Joins: Yakulu (fabien@moz-8a4.fkm.194.78.IP)
- # [23:12] * Quits: @yzen (yzen@moz-i5m.05u.207.66.IP) (Ping timeout: 121 seconds)
- # [23:14] * agibson is now known as agibson|afk
- # [23:28] * davidb|afk is now known as davidb
- # [23:29] * Quits: ~davidb (davidb@moz-5d550r.dsl.bell.ca) (Quit: Blah blah blah)
- # Session Close: Tue Dec 09 00:00:00 2014
The end :)