[win32] Set background color on all image handles

[akoch-yatta]

This PR adapts the win32 implementation of Image, so ensure that the background color of an image is always set to all existing of later created image handles. Currently all handles that are created after the background color was set, would not receive the background color.

Prerequisite for

• [win32] Create ImageData based image handles on demand #1909

[github-actions]

Test Results

   510 files  + 1     510 suites  +1   8m 26s ⏱️ -10s
 4 345 tests +37   4 331 ✅ +35   14 💤 +3  0 ❌  - 1 
16 613 runs  +37  16 502 ✅ +35  111 💤 +3  0 ❌  - 1 

Results for commit cc39aa1. ± Comparison against base commit 9440140.

♻️ This comment has been updated with latest results.

[fedejeanne]

I would like to have a concrete way to test this though.
Also:

• what was the bug that's being fixed?
• Is it a bug or simply an inconsistency in the API contract?
• Is there a test to make sure that an exception is thrown if setBackground(...) is called twice? If not, can you add it?
• Is there a test that makes sure that images created afterward maintain the background? If not, can you add it?

[akoch-yatta]

I would like to have a concrete way to test this though. Also:

• what was the bug that's being fixed?
• Is it a bug or simply an inconsistency in the API contract?

I didn't actively saw the issue somewhere as this functionality is not used very much, but we should be able to create an issue with a Snippet. I will have a look into that.

• Is there a test that makes sure that images created afterward maintain the background? If not, can you add it?

Yes, I can add one in the win32 specific tests

• Is there a test to make sure that an exception is thrown if setBackground(...) is called twice? If not, can you add it?

No, but I can add one

[akoch-yatta]

I would like to have a concrete way to test this though. Also:

• what was the bug that's being fixed?
• Is it a bug or simply an inconsistency in the API contract?

I didn't actively saw the issue somewhere as this functionality is not used very much, but we should be able to create an issue with a Snippet. I will have a look into that.

@fedejeanne After having a more thorough look into the implementation I ran into the problem, that this functionality doesn't seem to have any (visible) effect anymore. I assume that was originally added to fix issues with transparency with older windows versions. I was not able to trigger any visible effect by this -> transparencyPixel always worked.

• Is there a test that makes sure that images created afterward maintain the background? If not, can you add it?

Yes, I can add one in the win32 specific tests

I don't really see a way to test that automatically. As it was not possible to create a manual test as well, I don't see a chance here.

• Is there a test to make sure that an exception is thrown if setBackground(...) is called twice? If not, can you add it?

No, but I can add one

I would rather push that to a separate PR or even leave it out. If there is no real use case of this functionality I don't see value in adding this.

[fedejeanne]

The changes look OK. One final question: is transparentColor even needed anymore? I see that it's only used outside of Image (in GC) but it looks to me as if it can be replaced by the newly added backgroundColor.

WDYT @akoch-yatta ?

[akoch-yatta]

The changes look OK. One final question: is transparentColor even needed anymore? I see that it's only used outside of Image (in GC) but it looks to me as if it can be replaced by the newly added backgroundColor.

WDYT @akoch-yatta ?

There is quite a similar logic involved there. I am just a bit hesitant to do something like that, because I don't fully understand the whole logic there.