Fix issue #2446: rotozoom keeps the colorkey flag.

[igordsm]

Issue #2446 reported that using rotozoom looses the colorkey information and provided code to reproduce the issue. We modified rotozoomSurface to copy colorkey from src to rz_dst and added a test to test/transform_test.py.

Closes #2446

Before:

After:

[ankith26]

Hello, thanks for contributing to pygame-ce! 🎉 🍰

This seems like a nice fix, but we have CI failing on lint errors. To fix that run setup.py format and commit resulting changes (make sure you have the appropriate tools like black/clang-format installed)

[MyreMylar]

Just added a note here about an RLE thing that likely needs fixing in general across the whole of transform.c before we add another version of it in here.

We probably also need to make some tests for it based on the first post in #1668 just to check if it still is, or isn't a problem, as SDL has likely changed since this issue was first posted.

[MyreMylar]

This will also need to merge in the latest main branch to pick up the grub-efi fix.

[igordsm]

I'll get back to this in a few days

[MyreMylar]

LGTM. Can confirm it fixes the issue and works with RLE and non-RLE colorkey surfaces. Thank you for the contribution!

👍 🎉

[MyreMylar]

I realised while investigating the failing test that the function now known as SDL_SurfaceHasRLE that was renamed for SDL 3, was also only added to SDL 2 in version 2.0.14 so we are still 5 patch versions behind in our lowest supported SDL to use it. notably our wheels are ahead of this version so that's why I wasn't seeing this error locally.

I've just added a patch commit to vendor this function in for the case where SDL is less that 2.0.14. Not the most elegant solution (if it eventually works), but it should hold up OK as the function itself is super basic and we can delete it eventually when our bottom end catches up with our current main target.

[MyreMylar]

Getting that to compile via Github web version interactions required a few compromises... I'm planning to do another PR on the topic of Surface RLE flags for the whole transform module once this gets merged, so I can almost certainly clean this up a bit in that PR.

[ankith26]

LGTM, thanks for the PR 🎉

[ankith26]

Noticed the original commit has many co-authors, so thank you all! Unless anyone minds, I'd like to squash-and-merge this PR but I will leave the co-author tags intact.

[MyreMylar]

Absolutely squash and merge away. Igor did 99% of the work here. I was just wrestling with the compiler to get it over the line.

…

On Sat, Oct 28, 2023 at 12:56 PM Ankith ***@***.***> wrote: Noticed the original commit has many co-authors, so thank you all! Unless anyone minds, I'd like to squash-and-merge this PR but I will leave the co-author tags intact. — Reply to this email directly, view it on GitHub <#2491 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/ADGDGGUP2WZ557IVUA44OODYBTXIJAVCNFSM6AAAAAA5PJ3OU2VHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMYTOOBTG44DSNJXGQ> . You are receiving this because your review was requested.Message ID: ***@***.***>