Paul Sprencz | 1 Sep 11:22 2003
Picon

Fwd: Request to apply patch for bug #108664 in gnopernicus


We have the attached patch for bug #108664 in gnopernicus. This patch used
to adapt the changes made in at-spi to work a feature which is very
important for gnopernicus. 
Without this feature in gnopernicus can not be activate some feature on 
Solaris and in java applications.
This bug blocks a number of gnopernicus bugs and has been tested by the 
gnopernicus and the at-spi team.

Problem description:
Under Solaris and with java application the numpad keys are not working. This 
patch is a workaround to solve this.

Permission to apply is requested.

Regards
Paul Sprencz
BAUM Engineering

Attachment (libke.patch): text/x-diff, 9 KiB
Murray Cumming | 1 Sep 13:49 2003
Picon
Picon

Re: Fwd: Request to apply patch for bug #108664 in gnopernicus

On Mon, 2003-09-01 at 11:22, Paul Sprencz wrote:
> We have the attached patch for bug #108664 in gnopernicus. This patch used
> to adapt the changes made in at-spi to work a feature which is very
> important for gnopernicus. 
> Without this feature in gnopernicus can not be activate some feature on 
> Solaris and in java applications.
> This bug blocks a number of gnopernicus bugs and has been tested by the 
> gnopernicus and the at-spi team.
> 
> Problem description:
> Under Solaris and with java application the numpad keys are not working. This 
> patch is a workaround to solve this.
> 
> Permission to apply is requested.

OK, (1 of a necessary 2 approvals), with the usual "please be careful".
Thank you for testing the patches thoroughly.

--

-- 
Murray Cumming
murrayc <at> usa.net
www.murrayc.com
Mark McLoughlin | 1 Sep 14:05 2003
Picon

Re: Request to apply patch for bug #108664

Hi Padraig,

On Fri, 2003-08-29 at 14:33, padraig o'briain wrote:
> We have the attached patch for bug #108664. This bug blocks a number of 
> gnopernicus bugs and has been tested by the gnopernicus team.
> 
> Permission to apply is requested.

	#108664 is a pretty huge bug to read. Please try and help us make a
decision by giving us the following information:

 * What does this patch fix and what are the consequences to the user
   if this patch doesn't get in ?

 * In a nutshell what does the patch do ?

 * How well tested is this patch ?

 * In your estimation how much risk is there that this patch, will
   cause serious regressions/bugs in other areas ?

Thanks,
Mark.
Mark McLoughlin | 1 Sep 14:10 2003
Picon

Re: Request to apply patch for bug #120725

Hey,

On Thu, 2003-08-28 at 18:04, Frederic Crozat wrote:
> > Hi:
> > 
> > We have a tested patch for bug #12075 in AT-SPI:
> > http://bugzilla.gnome.org/show_bug.cgi?id=120725
> > 
> > The bug can trigger crashes in AT clients like gnopernicus under high
> > event loading, the fix addresses a re-entrancy/race.  This bug also
> > appears to be the root cause of other crashes, so it is fairly
> > high-impact.
> > 
> > Permission to apply is requested; thanks.
> 
> Patch is ok for me WITHOUT the debug stuff :)
> 
> Please, next time, attach the patch to the mail, it is easier to review.

	(Sorry for keeping ye waiting for a second approval)

	Patches fixes for a crash; Fix looks straightforward[1]; Looks safe to
go in.

	Approval 2 of 2.

Good Luck,
Mark.

[1] - ye gods it looks like a horrible hack, but I won't get any further
(Continue reading)

padraig o'briain | 1 Sep 14:26 2003
Picon

Re: Request to apply patch for bug #108664

This patch suppresses window:deactivate and window:activate event pairs 
and a focus event notification which occur when using the numeric keypad 
in gnopernicus to issue command to gnopernicus.

Gnopernicus requests that at-spi-registryd process listen for certain 
key strokes. This requires that at-spi-registryd call XGrabKey, unless 
XEVIE extension is available, which is currently only on Solaris.When 
the listened for keystrokes occur FocusOut and FocusIn X events. These 
cause window:deactivate and window:activate to be generated. The 
at-spi-registryd prevents these events being notofied to a listener, 
i.e. gnopernicus.

Essentially, it makes using the numeric keypad in gnopernicus, a key 
part of its functionality usable.

The patch has been thoroughly tested by the Baum development team.

I do not think that there is any risk of regressions or bugs in other areas.

Padraig

Mark McLoughlin wrote:
> Hi Padraig,
> 
> On Fri, 2003-08-29 at 14:33, padraig o'briain wrote:
> 
>>We have the attached patch for bug #108664. This bug blocks a number of 
>>gnopernicus bugs and has been tested by the gnopernicus team.
>>
>>Permission to apply is requested.
(Continue reading)

Mark McLoughlin | 1 Sep 14:37 2003
Picon

Re: Request to apply patch for bug #108664

Hey,
	Okay, I'm fine with that ...

	Approval 1 of 2.

Good Luck,
Mark.

On Mon, 2003-09-01 at 13:26, padraig o'briain wrote:
> This patch suppresses window:deactivate and window:activate event pairs 
> and a focus event notification which occur when using the numeric keypad 
> in gnopernicus to issue command to gnopernicus.
> 
> Gnopernicus requests that at-spi-registryd process listen for certain 
> key strokes. This requires that at-spi-registryd call XGrabKey, unless 
> XEVIE extension is available, which is currently only on Solaris.When 
> the listened for keystrokes occur FocusOut and FocusIn X events. These 
> cause window:deactivate and window:activate to be generated. The 
> at-spi-registryd prevents these events being notofied to a listener, 
> i.e. gnopernicus.
> 
> Essentially, it makes using the numeric keypad in gnopernicus, a key 
> part of its functionality usable.
> 
> The patch has been thoroughly tested by the Baum development team.
> 
> I do not think that there is any risk of regressions or bugs in other areas.
> 
> Padraig
> 
(Continue reading)

Paul Sprencz | 1 Sep 14:58 2003
Picon

Re: Fwd: Request to apply patch for bug #108664 ingnopernicus

Thank you!

Best Regards
Paul Sprencz
----- Original Message -----
From: "Murray Cumming" <murrayc <at> usa.net>
To: "Paul Sprencz" <ps <at> baum.ro>
Cc: "gnome-release-team" <release-team <at> gnome.org>;
<gnome-accessibility-devel <at> gnome.org>
Sent: Monday, September 01, 2003 2:49 PM
Subject: Re: Fwd: [g-a-devel]Request to apply patch for bug #108664
ingnopernicus

> On Mon, 2003-09-01 at 11:22, Paul Sprencz wrote:
> > We have the attached patch for bug #108664 in gnopernicus. This patch
used
> > to adapt the changes made in at-spi to work a feature which is very
> > important for gnopernicus.
> > Without this feature in gnopernicus can not be activate some feature on
> > Solaris and in java applications.
> > This bug blocks a number of gnopernicus bugs and has been tested by the
> > gnopernicus and the at-spi team.
> >
> > Problem description:
> > Under Solaris and with java application the numpad keys are not working.
This
> > patch is a workaround to solve this.
> >
> > Permission to apply is requested.
>
(Continue reading)

gnome-accessibility-devel | 1 Sep 18:59 2003

multiple registryd running ?

Bonjour,
I'm a at-spi & java-bridge CVS user. I can back from holidays and I 
noticed some strange errors in my at-spi using code.
Sometimes, my logging app. does not "see" another application. If I 
relaunche the taget  application (eog or abiword for instance) it 
sometimes is ok.
Investigating, I see that multiple at-spi-registryd are running !
I am launching two X servers (one for at-spi testing, the other to do 
other stuff that I do not want to lose if at-spi steals any mouse & 
keyboard input). I would expect 2 at-spi-registryd, but I sometime 
see many more ( five are running right now).

bash-2.05b$ ps xau |fgrep registryd
hugueney 20113  0.2  1.5  9272 3860 ?        S    16:43   0:02 
/var/UserSpace/Build//libexec/at-spi-registryd 
--oaf-activate-iid=OAFIID:Accessibility_Registry:1.0 --oaf-ior-fd=21
hugueney 21509  0.1  1.4  9268 3816 ?        S    16:45   0:01 
/var/UserSpace/Build//libexec/at-spi-registryd 
--oaf-activate-iid=OAFIID:Accessibility_Registry:1.0 --oaf-ior-fd=21
hugueney 23035  0.0  1.4  9268 3804 ?        S    16:50   0:00 
/var/UserSpace/Build//libexec/at-spi-registryd 
--oaf-activate-iid=OAFIID:Accessibility_Registry:1.0 --oaf-ior-fd=21
hugueney 23075  0.1  1.4  9268 3840 ?        S    16:51   0:01 
/var/UserSpace/Build//libexec/at-spi-registryd 
--oaf-activate-iid=OAFIID:Accessibility_Registry:1.0 --oaf-ior-fd=21
hugueney 26008  0.1  1.4  9268 3828 ?        S    16:57   0:00 
/var/UserSpace/Build//libexec/at-spi-registryd 
--oaf-activate-iid=OAFIID:Accessibility_Registry:1.0 --oaf-ior-fd=21
hugueney  1230  0.0  0.1  3244  484 pts/2    R    17:05   0:00 grep -F 
registryd
(Continue reading)

padraig o'briain | 1 Sep 17:38 2003
Picon

Re: multiple registryd running ?

Bernard,

There should not be multiple copies of at-spi-registryd.

The call to bonobo_activation_activate_from_id() in 
spi_atk_bridge_get_registry in at-spi/atk-bridge/bridge.c should cause 
the at-spi-registryd process to be started if it not already running.

The bonobo-activation-server process should deal with requests to start 
the process. If there is a problem it is probably in that process but as 
I have not have to analyse a problem such as this I cannot be certain.

Padraig

gnome-accessibility-devel <at> bernard-hugueney.org wrote:
> Bonjour,
> I'm a at-spi & java-bridge CVS user. I can back from holidays and I 
> noticed some strange errors in my at-spi using code.
> Sometimes, my logging app. does not "see" another application. If I 
> relaunche the taget  application (eog or abiword for instance) it 
> sometimes is ok.
> Investigating, I see that multiple at-spi-registryd are running !
> I am launching two X servers (one for at-spi testing, the other to do 
> other stuff that I do not want to lose if at-spi steals any mouse & 
> keyboard input). I would expect 2 at-spi-registryd, but I sometime 
> see many more ( five are running right now).
> 
> bash-2.05b$ ps xau |fgrep registryd
> hugueney 20113  0.2  1.5  9272 3860 ?        S    16:43   0:02 
> /var/UserSpace/Build//libexec/at-spi-registryd 
(Continue reading)

gnome-accessibility-devel | 1 Sep 20:02 2003

Re: multiple registryd running ?

Le Lundi 1 Septembre 2003 17:38, padraig o'briain a écrit :
> Bernard,
>
> There should not be multiple copies of at-spi-registryd.
That's what I thought.
It is true even if I have two X servers at once ?

>
> The call to bonobo_activation_activate_from_id() in
> spi_atk_bridge_get_registry in at-spi/atk-bridge/bridge.c should
> cause the at-spi-registryd process to be started if it not already
> running.
>
> The bonobo-activation-server process should deal with requests to
> start the process. If there is a problem it is probably in that
> process but as I have not have to analyse a problem such as this I
> cannot be certain.

I may investigate this. Could you tell me the prefered way to deal 
with a seemingly "malfunctionning" registryd ?
 bonobo-slay
or killall at-spi-registryd &&rm -rf /tmp/orbit-username 
or something else ?

Thanks for your help,
Bernard

Gmane