[Digikam-devel] Mousepointer glued to rubber, steps to reproduce.

[Peter Heckert]

Hello,

Here the steps to reproduce the phenomena:

Open noisereduction.
Move any of the rubbers. Dont release the left mousebutton.
While the filter  is busy press right mousebutton. The "Whats this" box will
appear.
When the filter is not busy then release the buttons. Do /not/ click
onto the
"whats this" box. Click offside, and the box disappears.

Now the Mouse is "glued" to the rubber. When the mouse is moved over the
rubber then the rubber gets focus and moves, without any mousebutton
pressed.
Reproducibility: 100%

I found only one way to resolve this situation: Move mouse over the
glued rubber.
Press left button, release,press middle mousebutton,release, dont move
mouse.
Possibly repeat it until success.

Sometimes this happens randomly when I never pressed the right mousebutton.
Possibly it has to do with my mouse. It is a optical mouse and somtimes
it makes
sudden jumps (errornous position data). It's a Logitech premium optical
wheel Mouse.

This can also be observed with other plugins, e.g. the refocus plugin.
The preview
window must be made so large, that it is busy long enough to press the
right mouse
button.

Peter
_______________________________________________
Digikam-devel mailing list
[hidden email]
https://mail.kde.org/mailman/listinfo/digikam-devel

[Peter Heckert]

Peter Heckert wrote:

I believe, that sometimes the "Mousebutton released" Message is lost,
due to data corruption and this will "glue" the Mouse to the rubber.

Peter
_______________________________________________
Digikam-devel mailing list
[hidden email]
https://mail.kde.org/mailman/listinfo/digikam-devel

[Gilles Caulier-2]

Le Dimanche 30 Avril 2006 12:40 AM, Peter Heckert a écrit :
> Peter Heckert wrote:
>
> I believe, that sometimes the "Mousebutton released" Message is lost,
> due to data corruption and this will "glue" the Mouse to the rubber.
>
> Peter

data corruption done by what ? You filter code ?

In all case, check your code using valgrind. It will decrease digikam speed
(/15) but the report on the console are very instructive.

Look in HACKING file from digikam folder.

Gilles
_______________________________________________
Digikam-devel mailing list
[hidden email]
https://mail.kde.org/mailman/listinfo/digikam-devel

[Peter Heckert]

Hi Gilles,

hope you had luck and success.

Caulier Gilles wrote:

> Le Dimanche 30 Avril 2006 12:40 AM, Peter Heckert a écrit :
>  
>> Peter Heckert wrote:
>>
>> I believe, that sometimes the "Mousebutton released" Message is lost,
>> due to data corruption and this will "glue" the Mouse to the rubber.
>>
>> Peter
>>    
>
> data corruption done by what ? You filter code ?
>
>  

No. As I mentioned before, the effect also happens in the refocus
plugin, when
the window is made so large, that the filter is busy long enough, after
moving
a rubber.

Even under windows my mouse "jumps" sometimes. So I believe the mouse
events
can be lost or corrupted by hardware or due to lost interrupts, when the
processor
is busy.

> In all case, check your code using valgrind. It will decrease digikam speed
> (/15) but the report on the console are very instructive.
>
>  
I will do. However before I do it, I must learn it and this needs time.

> Look in HACKING file from digikam folder.
>  
I already did ;-)

wish you a nice weekend,

Peter

_______________________________________________
Digikam-devel mailing list
[hidden email]
https://mail.kde.org/mailman/listinfo/digikam-devel

[Peter Heckert]

Caulier Gilles wrote:

> Le Dimanche 30 Avril 2006 12:40 AM, Peter Heckert a écrit :
>  
>> Peter Heckert wrote:
>>
>> I believe, that sometimes the "Mousebutton released" Message is lost,
>> due to data corruption and this will "glue" the Mouse to the rubber.
>>
>> Peter
>>    
>
> data corruption done by what ? You filter code ?
>  

It happens apparently with any plugin, that needs some calculation time
and that starts
calculation while the mousepointer is pressed.

I spent a lot of time to find a way to reproduce, because this happens very
often in noisereduction for me, it is disturbing.

Here is a better way and more reasonable way to reproduce:

Move a rubber and hold mouse button down, until the progress bar starts
running.
While the progressbar is running, release mousebutton and move mousepointer
away quickly.
After this, the mouse is "glued" to the rubber.
This can be reproduced in Noisereduction plugin and in refocus plugin
(make window large enough, so that it is busy for one second).
I believe, that the rubber doesnt catch the "Mousebutton Up" Event, when
the mouse
is moved away while the plugin is busy calculating.

Because noisereduction parameters need accurate adjustment, and I move the
rubbers slowly, this happens quite often to me.

regards,

Peter





_______________________________________________
Digikam-devel mailing list
[hidden email]
https://mail.kde.org/mailman/listinfo/digikam-devel

[Peter Heckert]

This is one of the reasons why I left linux some time ago:
When you start to become a "mouse wizard" like some photoshop professionals,
and you move the mouse more quickly than you can see ;-) then often buttons
or rubbers move, that you didnt want to move.
This happened mostly under gnome, while I developed the noisereduction code.

peter

_______________________________________________
Digikam-devel mailing list
[hidden email]
https://mail.kde.org/mailman/listinfo/digikam-devel