[win32] Adapt GC zoom when using plain images

[akoch-yatta]

This PR adapts the zoom passed to the GC when it is initialized for a plain Image created via Image#Image(Device, int, int) with a given width and height. It will fall back to DPIUtil#getDeviceZoom() again to revert to original bevahiour and prevent blurry rescaling of old usages.

[github-actions]

Test Results

   539 files   -  6     539 suites   - 6   33m 28s ⏱️ + 4m 6s
 4 361 tests  - 37   4 345 ✅  - 35   15 💤  - 3  1 ❌ +1 
16 682 runs   - 37  16 543 ✅  - 35  138 💤  - 3  1 ❌ +1 

For more details on these failures, see this check.

Results for commit 7f4f512. ± Comparison against base commit 9d3227d.

This pull request removes 37 tests.

AllWin32Tests org.eclipse.swt.graphics.ImageWin32Tests ‑ testImageDataForDifferentFractionalZoomsShouldBeDifferent
AllWin32Tests org.eclipse.swt.graphics.ImageWin32Tests ‑ testImageShouldHaveDimesionAsPerZoomLevel
AllWin32Tests org.eclipse.swt.tests.win32.Test_org_eclipse_swt_dnd_DND ‑ testByteArrayTransfer
AllWin32Tests org.eclipse.swt.tests.win32.Test_org_eclipse_swt_dnd_DND ‑ testFileTransfer
AllWin32Tests org.eclipse.swt.tests.win32.Test_org_eclipse_swt_dnd_DND ‑ testHtmlTransfer
AllWin32Tests org.eclipse.swt.tests.win32.Test_org_eclipse_swt_dnd_DND ‑ testImageTransfer_fromCopiedImage
AllWin32Tests org.eclipse.swt.tests.win32.Test_org_eclipse_swt_dnd_DND ‑ testImageTransfer_fromImage
AllWin32Tests org.eclipse.swt.tests.win32.Test_org_eclipse_swt_dnd_DND ‑ testImageTransfer_fromImageData
AllWin32Tests org.eclipse.swt.tests.win32.Test_org_eclipse_swt_dnd_DND ‑ testImageTransfer_fromImageDataFromImage
AllWin32Tests org.eclipse.swt.tests.win32.Test_org_eclipse_swt_dnd_DND ‑ testRtfTransfer
…

♻️ This comment has been updated with latest results.

[HeikoKlare]

The issue was introduced with this PR:

• [win32] Remove Image#initialNativeZoom #2064

It can, e.g., be seen in double-buffering scenarios. One such scenario can be reproduced with the WindowBuilder project when going back to this commit: eclipse-windowbuilder/windowbuilder@f69b7dd

Starting the application on Windows without monitor-specific scaling at 175% (or above) and creating a sample projects leads to this wrong scaling:

With the proposed fix, it will look like this:

With monitor-specific scaling enabled, it will like the above instead of having blurry scaling as before:

[HeikoKlare]

This change should go in the upcoming release (and also in RC1 to have some remaining time of testing) to preserve the behavior of users of the Image(device, width, height) constructor on Windows.

I consider this change rather safe to merge as it consistutes a partial revert of the recently merged change:

• #2064

So the risk that consumers already adapted to the changed behavior is low. At the same time, the risk that consumers that did not adapt will face unintended behavior is probably not that low.

[ptziegler]

You guys really are the best! ❤️

[HeikoKlare]

The thanks are yours, Patrick! ❤️ You have pointed us to that potential issue, otherwise we would have probably missed it.
Do you have any objections on merging this for RC1, @ptziegler?

[ptziegler]

Do you have any objections on merging this for RC1

Go for it.

[jonahgraham]

Do you have any objections on merging this for RC1

IIUC the RC1 is already complete and about to be submitted to SimRel - see https://www.eclipse.org/lists/eclipse-dev/msg12407.html

[iloveeclipse]

Do you have any objections on merging this for RC1

IIUC the RC1 is already complete and about to be submitted to SimRel - see https://www.eclipse.org/lists/eclipse-dev/msg12407.html

Not really, see eclipse-platform/eclipse.platform.releng.aggregator#3055

[merks]

We've been spinning a bit...

We should contribute something today, and then we/I can contribute something newer tomorrow.

I have to run now though...

[HeikoKlare]

Thank you all! I will merge this now and then trigger another I-Build, which we may then contribute as (an update for) RC1.

The failing test is unrelated and already documented: #2098