mozilla :: #accessibility

13 Sep 2017
14:14aaronlevalexander, eeejay, MarcoZ, davidb -- do any of you know of other tools besides AViewer/AccProbe for inspecting/watching IA2?
14:14davidbaaronlev: hey, sorry I didn't reply to email yet, I was hoping someone else would :)
14:14aaronlevnp
14:15davidbI would normally go to accprobe but haven't used in a long time
14:15davidbdoes it still need you to command line regsvr
14:15aaronlevmTBE
14:15aaronlevmaybe
14:15aaronlevi can't even get it to run
14:15aaronlevaccprobe.exe does nothing and returns for me
14:16aaronlevdmazzoni got it running though, but i believe it's known to be slow and buggy
14:17aaronlevtoo bad i don't have my bastardized msaa sdk tools anymore
14:17davidbaaronlev: regsvr info here https://github.com/IBMa/AccProbe
14:17MarcoZI use NVDAs object navigator and element info, havent used AccProbe in 5-6 years.
14:18aaronlevMarcoZ: that sounds interesting
14:18aaronlevmarcoz, what about for an event stream?
14:30firebotNew Firefox - Disability Access bug 1399497 filed by roxana.leitan@softvision.ro.
14:30firebothttps://bugzil.la/1399497 NEW, nobody@mozilla.org Title bar is not visible in High Contrast mode on Ubuntu
14:41jimmhttps://sql.telemetry.mozilla.org/queries/26633/source
14:41jimmhttps://sql.telemetry.mozilla.org/queries/16551/source
14:41jimmhttps://sql.telemetry.mozilla.org/queries/14971
14:43jimmhttps://sql.telemetry.mozilla.org/queries/12947
15:53MarcoZaaronlev: Haven't had to deal with one or could delegate it. Events are hard to monitor with a screen reader playing into the same browser at the same time.
16:01aaronlevmarcoz: yeah, it would be great if we could monitor from a different computer
16:17eeejayi tried accprobe the other day, and yeah. it won't start
16:18eeejaythe download page says it was designed with ease of use in mind, so its an eclipse component
16:18eeejaythat made me lol
17:17firebotNew Core - Disability Access APIs bug 1399557 filed by dbolter@mozilla.com.
17:17firebothttps://bugzil.la/1399557 NEW, surkov.alexander@gmail.com Crash in mozilla::a11y::DocAccessible::DoInitialUpdate
17:20davidbMarcoZ: bug 1380331 says firefox 56 branch... are you testing on beta accidentally?
17:20firebothttps://bugzil.la/1380331 NEW, nobody@mozilla.org [e10s a11y] Letting the machine go into hibernation and waking it up results in performance degradat
17:22MarcoZdavidb: No, this was when I filed that bug. Nightly was on 56 then. Feel free to change it to Trunk.
17:22davidbah ok
17:55surkovyzen: ping and hey
17:55surkov(checking connection)
17:55surkovdavidb: ^
17:56davidbsurkov: just got this
17:56surkovok, cool
18:05aklotzsurkov: About bug 1399557: The problem is that DocAccessible::DoInitialUpdate returns void, so any failures that happen there do not propagate up to NotificationController::WillRefresh. Even if they did, we would need to figure out what to do when DoInitialUpdate does return a failure
18:05firebothttps://bugzil.la/1399557 NEW, nobody@mozilla.org Crash in mozilla::a11y::DocAccessible::DoInitialUpdate
18:08surkovaklotz: what NotificationController::WillRefres could do with the failure?
18:08aklotzsurkov: Well, that's why we wanted to talk to you ;-)
18:08aklotzsurkov: We need to figure out whether / what we can do from there
18:09surkovwhat this asseriton is about?
18:09surkovand could it be moved right to WIllRefresh()?
18:09surkovat least you woulndt need to change DoIntialUpdate() :)
18:10aklotzsurkov: The assertion is essentially an IPC failure
18:11firebotNew Core - Disability Access APIs bug 1399580 filed by jdiggs@igalia.com.
18:11firebothttps://bugzil.la/1399580 NEW, nobody@mozilla.org When aria-haspopup contains an unrecognized value, it should be exposed as false; not true
18:14surkovaklotz: so how to approach to it? apparently CreateHolderFromAccessible fails but it says nothing to me
18:15aklotzsurkov: Well that's something that I need to do more investigation on and want to fix. But our concern in the meeting was that we should probably be able to gracefully handle failures from DoInitialUpdate
18:16surkovaklotz: ok, say DoIntialUpdate throws an elegant exception or something, then what optiosn do we have how to process that nicely?
18:16surkovit appears crash in debug build is not the best option?
18:17surkovak
18:18surkovaklotz: ^
18:18aklotzsurkov: Well, our concern is on release builds. What do we do then? Do we just MOZ_RELEASE_ASSERT, or do we try to do something better? So we wanted to ask you whether you think there is a way that we can handle this apart from asserting?
18:19surkovso we want to catch a problem but we dont want to crash, correct?
18:20surkovso is it a fatal error, can we continue if it failed?
18:27surkovaklotz: Im not really sure what options we have. If you want to cover release builds, then it probably a fatal error, and what you ideally want is to obtains steps to reproduce. It appears its a high volume crash according to crash reports, so making it failing it on release builds probably wont make you closer to solving it or obtaining strs. Is there an option to move the assertion somewhere up on a stack, i.e. into
18:27surkovCreateHolderFromAccessible, in case if it will say more?
18:31davidbdo we think the IPC fail would happen on subsequent attempts or is it flakey fail?
19:07aklotzdavidb: These failures look like they're consistently from the same one or two users. I think they will continue to see it on subsequent attempts.
19:08aklotzsurkov: Yes, I'm going to be moving that assertion higher up the stack
19:14surkovaklotz: ok, any improvements for DoInitialUpdate() you can think of or just leave it for now?
19:14aklotzsurkov: Probably not my call. davidb?
20:03jimmhow would one navigate to browser toolbar buttons using the keyboard? or do a11y users rely on keyboard shortcuts?
20:21surkova11y users rely on screen readers navigation I think
20:29davidbF6 is a thing
20:29davidbto try
20:30davidbbut our sighted keyboard user story needs work
20:30davidbon the plus side, Enn is back and might know more
20:35jimmdavidb: so trying to repro this surface 4 jank issue. so far no luck, seems to be scxrolling fluidly right after page reload or refresh..
20:35davidbjimm: clean profile?
20:35jimmI'll keep trying for a bit
20:35jimmyeah I never use this device
20:35davidbcool
20:35jimmfresh nightly install
20:36jimmloading up the wwI page
14 Sep 2017
No messages
   
Last message: 6 days and 18 hours ago