After an evaluation, GNOME has moved from Bugzilla to GitLab. Learn more about GitLab.
No new issues can be reported in GNOME Bugzilla anymore.
To report an issue in a GNOME project, go to GNOME GitLab.
Do not go to GNOME Gitlab for: Bluefish, Doxygen, GnuCash, GStreamer, java-gnome, LDTP, NetworkManager, Tomboy.
Bug 524263 - Sometime gnome-session show error dialogs about "accessibility registry was not found"
Sometime gnome-session show error dialogs about "accessibility registry was n...
Status: RESOLVED FIXED
Product: gnome-session
Classification: Core
Component: gnome-session
unspecified
Other All
: Normal major
: ---
Assigned To: Session Maintainers
Session Maintainers
: 485944 (view as bug list)
Depends on:
Blocks:
 
 
Reported: 2008-03-25 08:14 UTC by Li Yuan
Modified: 2008-06-17 19:39 UTC
See Also:
GNOME target: ---
GNOME version: ---


Attachments
patch (857 bytes, patch)
2008-03-25 08:20 UTC, Li Yuan
none Details | Review
new patch (848 bytes, patch)
2008-04-10 02:01 UTC, Li Yuan
none Details | Review

Description Li Yuan 2008-03-25 08:14:58 UTC
Normally happens when logout and login again, or kill X and login again. After input the password of an account, a dialog popup, says "Assistive technology support has been requested for this session, but the accessibility registry was not found. Please ensure that the AT-SPI package is installed. Your session has been started without assistive technology support."

But after login, at-spi-registryd is running and works fine.
Comment 1 Li Yuan 2008-03-25 08:20:54 UTC
Created attachment 107979 [details] [review]
patch

When timeout, we check the root window's property to see if the IOR string is there. If not, pop up the error dialog, if yes, just go on. This is because some machine is not fast enough, and the filter cannot be call in time. I don't think a longer time is a good idea because it make user wait longer if it really fails, and may still don't work on slower machines.
Comment 2 Li Yuan 2008-03-25 10:05:45 UTC
*** Bug 485944 has been marked as a duplicate of this bug. ***
Comment 3 Li Yuan 2008-03-28 05:05:00 UTC
Any body want to review the patch?
Comment 4 Li Yuan 2008-04-10 02:01:47 UTC
Created attachment 108962 [details] [review]
new patch

only test if the pointer is equal to 0
Comment 5 Willie Walker 2008-05-26 12:11:01 UTC
This is kind of a troublesome problem on slow machines.  In addition, the dialog seems to block further progress until it is clicked.  Furthermore, the dialog is positioned on the upper left of the screen, away from the pointer, requiring you to move the pointer to the dialog before being able to dismiss it.

All these reasons make it difficult for people who are blind or otherwise need assistive technologies to access the machine to get past this dialog.  
If possible, I wonder if another strategy besides a timeout should be used or if we could consider getting rid of the dialog altogether?
Comment 6 Li Yuan 2008-05-27 01:54:28 UTC
I just pinged Lucas about this bug yesterday.

Actually I have never seen a failure start of at-spi-registryd. All warning dialog popup because of the timeout. at-spi-registryd can be started in time as I debug, but seems signal emit is very slow in X sometime. My patch is to check IOR directly when timeout, and it works on my machine. Another idea is we just disable the error dialog, because a failure start of at-spi-registryd is unlikely.
Comment 7 Willie Walker 2008-05-27 13:13:45 UTC
(In reply to comment #6)
> I just pinged Lucas about this bug yesterday.
> 
> Actually I have never seen a failure start of at-spi-registryd. All warning
> dialog popup because of the timeout. at-spi-registryd can be started in time as
> I debug, but seems signal emit is very slow in X sometime. My patch is to check
> IOR directly when timeout, and it works on my machine. Another idea is we just
> disable the error dialog, because a failure start of at-spi-registryd is
> unlikely.

Given that the failure is unlikely, most likely because of Li's hard work on the AT-SPI infrastructure, I think the dialog has probably outlived its purpose.  Instead of serving as a useful tool to let you know the AT-SPI infrastructure has gone south, it is now kind of just getting in the way.

I'd propose getting rid of the dialog altogether, and I'd propose doing this for both GNOME 2.22 and trunk.

Thoughts?
Comment 8 Kjartan Maraas 2008-05-27 17:02:54 UTC
The only time I've actually seen this is when running under valgrind. Showing the dialog doesn't really help get the AT-SPI infrastructure up and running in that situation anyway I guess.
Comment 9 Lucas Rocha 2008-06-01 12:59:24 UTC
+1 for removing the dialog. I'd just like to hear from other experienced guys about it before doing it. Dan, Vincent, what do you think?
Comment 10 Dan Winship 2008-06-01 18:52:46 UTC
Well, it sounds like the dialog is wrong more often than it's right, so...
Comment 11 Lucas Rocha 2008-06-01 19:48:18 UTC
Removed from trunk. Vincent, should I do it in gnome-2-22 branch as well?

2008-05-31  Lucas Rocha  <lucasr@gnome.org>

        * compat/at-spi-registryd-wrapper.c (show_error, registry_ior_watch,
        registry_ior_watch_timeout, main): remove dialog shown after a timeout 
        in at-spi wrapper program. #524263, Li Yuan.
Comment 12 Willie Walker 2008-06-04 20:22:28 UTC
Unless I'm missing something, I'd really be for making this change to 2.22 as well.  Vincent, what are your thoughts?

For other improvement ideas, see also bug #408034.
Comment 13 Lucas Rocha 2008-06-05 22:23:14 UTC
Ok, commited to 2.22 as well. Closing.