[prev in list] [next in list] [prev in thread] [next in thread] 

List:       openjdk-2d-dev
Subject:    Re: Kindly request your attention for a really annoying visual Java bug on Linux Mint
From:       Aleksei Ivanov <alexey.ivanov () oracle ! com>
Date:       2022-01-24 19:33:52
Message-ID: 379cce55-7bd5-7771-7241-34b0a25a1525 () oracle ! com
[Download RAW message or body]

Hello Stefan,

On 23/01/2022 18:44, Stefan Reich wrote:
> Hello Alexey,
> 
> thanks a lot for investigating and submitting the bug. I don't think I 
> can sign up at bugs.openjdk.java.net <http://bugs.openjdk.java.net> 
> directly, right?

No, you can't.

You can submit a bug report using https://bugreport.java.com/bugreport/
After evaluation, the bug will be moved to JBS at 
https://bugs.openjdk.java.net/

> I have also done some further testing, for example on an older version 
> of Peppermint Linux (also on my ThinkPad). Same kind of errors appear. 
> So it doesn't seem related to the flavor of Linux used.

Thank you for additional testing and for confirming other flavours of 
Linux are affected.

> I have also tried the source code you produced. It gives the same 
> artifacts as my original version.

That's good. At least, the modified test reproduces the same problem.

> 
> > Perhaps, the calculation of the grid has an error or something like 
> that.
> 
> I challenge you to show me the bug in that 10 line function... ^^ 
> (Don't think the problem is in there)

It's not the case. I referred to the change of colour that I saw. I 
posted in my comment: While all the windows were created as one go on 
the EDT, each window was displayed with the right colour initially. Once 
the EDT got unblocked, I saw something like animation where the window 
tiles on the screen changed colours. The colours remained stable 
afterwards. This effect made me think there could be something going on 
with the coordinates.

However, it looks as if processing of some events makes the windows 
change the colour.

A better test would be to create one window in the area that gets darker 
and see what events it receives. Another thing I'd verify is the number 
of the paint events. The window is neither opaque nor transparent. If 
it's painted several times, the same background colour could be painted 
over the existing colour, thus the window could become darker with each 
paint. It's a wild guess, it could be misleading yet it's worth checking.

> Originally I had simply tried to make a JWindow covering the whole 
> second screen and it was completely opaque. That's how I discovered 
> the bug in the first place.
> 
> So... judging from the official bug report it seems this is being 
> investigated by the proper "authorities", right? That's a very good 
> result for now.

It's assigned to an engineer. That's all to it at this very moment.

If you can help with pinpointing the root cause or provide the fix for 
the problem, your help and contribution will be appreciated.


Regards,
Alexey

> 
> Many greetings,
> Stefan
> 
> 
> 
> On Fri, 21 Jan 2022 at 22:38, Aleksei Ivanov 
> <alexey.ivanov@oracle.com> wrote:
> 
> Dear Stefan,
> 
> Thank you for reporting the bug.
> 
> You could have submitted the bug via
> https://bugreport.java.com/bugreport/
> <https://urldefense.com/v3/__https://bugreport.java.com/bugreport/__;!!ACWV5N9M2RV99hQ!cKbBJVPVj_k3vtI9ThVRGYo8rY4-KtkDz6pN2QBPrYLkgYlHOkZcTe-ALpBzkWBwWZg$>
>  
> 
> Anyway, I reproduced the issue and now I've submitted the bug for you:
> https://bugs.openjdk.java.net/browse/JDK-8280482
> "Window transparency bug on Linux Mint"
> 
> I attached a modified version of your testcase which creates all the
> windows on the EDT.
> 
> 
> Even though I see the problem, there could be something wrong with
> the
> test, or possibly window placement. If I add a delay after a
> window is
> created, I can see that all the screen is getting covered
> gradually with
> windows and all the windows have the same level of translucency. When
> the bottom of the screen is reached, the new windows get moved
> above so
> that the newly created window does not cover the taskbar. This
> creates a
> darker rectangle seen along the bottom of the screen on the main
> monitor.
> 
> In my environment, the dark square area that you have on your
> screenshot
> is not as large, it covers only the part of the screen on the main
> monitor. At the same time, this area appears only when all the
> windows
> are created. Again, it could be that some of the windows of the
> grid get
> misplaced for whatever reason. Perhaps, the calculation of the
> grid has
> an error or something like that.
> 
> Eventually, it may be a bug not in Java but in the sample app.
> I'll play
> around with it over the weekend if I have time.
> 
> On 21/01/2022 20:10, Stefan Reich wrote:
> > Dear Sirs or Madams,
> > 
> > I make a visual recognition tool <https://gaz.ai
> <https://urldefense.com/v3/__https://gaz.ai__;!!ACWV5N9M2RV99hQ!cKbBJVPVj_k3vtI9ThVRGYo8rY4-KtkDz6pN2QBPrYLkgYlHOkZcTe-ALpBzCtp8Vto$>>
>  in Java that would
> > tremendously benefit from the ability to show translucent
> windows. Indeed,
> > Java has such an ability but it is broken on my Linux Mint 19
> ThinkPad for
> > unknown reasons (screenshot
> <https://botcompany.de/images/1103064
> <https://urldefense.com/v3/__https://botcompany.de/images/1103064__;!!ACWV5N9M2RV99hQ!cKbBJVPVj_k3vtI9ThVRGYo8rY4-KtkDz6pN2QBPrYLkgYlHOkZcTe-ALpBz3GvDylE$>>).
> 
> > 
> > All the details are here:
> > 
> > https://github.com/adoptium/adoptium-support/issues/430
> <https://urldefense.com/v3/__https://github.com/adoptium/adoptium-support/issues/430 \
> __;!!ACWV5N9M2RV99hQ!cKbBJVPVj_k3vtI9ThVRGYo8rY4-KtkDz6pN2QBPrYLkgYlHOkZcTe-ALpBzss3HZh8$>
> 
> > 
> > I'd love to see this fixed. Not sure how reproducible it is or
> if there is
> > anything screwed in my Linux Mint installation. Seems to run
> fine otherwise
> > though.
> > 
> > Many greetings,
> > Stefan Reich
> -- 
> Regards,
> Alexey
> 
> 
> 
> -- 
> ==    Join the _https://t.me/HumanAscension 
> <https://urldefense.com/v3/__https://t.me/HumanAscension__;!!ACWV5N9M2RV99hQ!cKbBJVPVj_k3vtI9ThVRGYo8rY4-KtkDz6pN2QBPrYLkgYlHOkZcTe-ALpBzY2f3Z6g$>_ \
>  ==
> == https://bitchute.com/channel/LxVBAHP3O0Hj 
> <https://urldefense.com/v3/__https://bitchute.com/channel/LxVBAHP3O0Hj__;!!ACWV5N9M2RV99hQ!cKbBJVPVj_k3vtI9ThVRGYo8rY4-KtkDz6pN2QBPrYLkgYlHOkZcTe-ALpBzCwpb1Io$> \
>  ==


[prev in list] [next in list] [prev in thread] [next in thread]