On Thu, 31 Jul 2025 16:21:46 GMT, Abhishek Kumar <abhis...@openjdk.org> wrote:

> > > If I am right, the main bug is about underline drawing and not about 
> > > mnemonic. Yet mnemonic is used to check underline drawing under the 
> > > string, I guess you can think of using the code mentioned in 
> > > [JDK-4967768](https://bugs.openjdk.org/browse/JDK-4967768) in the test 
> > > which would be fine for all L&F.
> > 
> > 
> > Good point. I second this if possible.
> 
> Yes, the main bug is about the underline drawing and not mnemonic but the 
> underline i.e. mnemonic visibility depends on the platform the test is 
> running. It depends on the current L&F as well. In case of Metal, Motif and 
> Nimbus, the underline is visible always whereas if the test ran with Windows 
> or GTK L&F, underline visibility depends on the how the underlying platform 
> handle the underline. If the underline drawing is invisible then I don't see 
> any way to verify it's position.
> 
> > I guess you can think of using the code mentioned in 
> > [JDK-4967768](https://bugs.openjdk.org/browse/JDK-4967768) in the test 
> > which would be fine for all L&F.
> 
> The code you mentioned is already implemented in [SwingUtilities's 
> drawStringUnderlineCharAt](https://github.com/kumarabhi006/jdk/blob/04c0b130f09c093797895cc928fe020d7e584cb9/src/java.desktop/share/classes/sun/swing/SwingUtilities2.java#L694)
>  method and I checked on Windows and macOS that this code path is executed 
> and that impact the underline drawing. If it is commented then there is no 
> underline drawing for the mnemonics and vice-versa.
> 
> The underline drawing is correct as it has drawn just below the baseline and 
> not below the character "P".
> 
> So, I think the fix requires an instruction change to let the tester know 
> that the "ALT" key press is a requirement to make the mnemonic visible and 
> then the underline position can be verified further.

I got the point. I'm suggesting that instead of relying on "ALT" key press 
which in turn draw underline that has some dependency on L&F, drawing underline 
to verify the test would be better.

-------------

PR Comment: https://git.openjdk.org/jdk/pull/26546#issuecomment-3142183359

Reply via email to