RFR: 8299553: Make ScaledEtchedBorderTest.java comprehensive

[Alexey Ivanov]

On Tue, 3 Jan 2023 23:05:53 GMT, Harshitha Onkar <honkar at openjdk.org> wrote:

>> The `ScaledEtchedBorderTest.java` test ([JDK-8279614](https://bugs.openjdk.org/browse/JDK-8279614)) verifies that the thickness of lines of `EtchedBorder` is consistent. The component inside had the same background color as the frame. Yet if there's another component the background of which is different, that background could be seen.
>> 
>> The updated test covers the scenario reported in [JDK-8294921](https://bugs.openjdk.org/browse/JDK-8294921).
>> 
>> The bug itself is fixed by refactoring [JDK-8294680](https://bugs.openjdk.org/browse/JDK-8294680) the border rendering code which uses a more precise coordinate calculation, see #11571. Thus, I contribute the test under a separate bugid.
>> 
>> I also modified `ScaledLineBorderTest.java` so that the location of the child panels is saved when they're created as done in `ScaledTextFieldBorderTest.java` in #11498. I removed the unused `Component` import after this modification from both tests.
>
> test/jdk/javax/swing/border/EtchedBorder/ScaledEtchedBorderTest.java line 256:
> 
>> 254:         throw new Error(
>> 255:                 String.format("Unexpected color at %d, %d: %08x",
>> 256:                               x, y, color));
> 
> Since `throwUnexpectedColor()` is called commonly for different state values, including the state value in the error msg could be useful for debugging purpose.

I don't think it's needed: the stack trace of the exception contains the line number which uniquely identifies the state.

For example, an excerpt of test output when it fails:

Scaling: 1.00
java.lang.Error: Unexpected color at 62, 24: ffffffff
        at ScaledEtchedBorderTest.throwUnexpectedColor(ScaledEtchedBorderTest.java:255)
        at ScaledEtchedBorderTest.checkBorder(ScaledEtchedBorderTest.java:231)
        at ScaledEtchedBorderTest.checkVerticalBorders(ScaledEtchedBorderTest.java:131)
        at ScaledEtchedBorderTest.verifyBorderRendering(ScaledEtchedBorderTest.java:99)
        at ScaledEtchedBorderTest.testScaling(ScaledEtchedBorderTest.java:87)
        at ScaledEtchedBorderTest.lambda$main$0(ScaledEtchedBorderTest.java:77)
	at …


https://github.com/openjdk/jdk/blob/b78389e8c1e23cb83338387ba20091bf705480e1/test/jdk/javax/swing/border/EtchedBorder/ScaledEtchedBorderTest.java#L221-L232

Line 231 corresponds to the `RIGHT_HIGHLIGHT` state. If you open the `test1.00.png` image, you quickly see, there's _a white line_ below the highlight line which shouldn't be there.

Therefore, passing the additional parameter doesn't add much value.

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

PR: https://git.openjdk.org/jdk/pull/11836