There's something about loading larger images from disk...

[Kawa-oneechan]

You link a 640×480 256-color image into a program directly and pass it to DRAW->DisplayPicture, you're fine.

DRAW->DisplayPicture((TImageFile*)&pic);

I temporarily changed the image in twohundred from 200-16.api to 480-256.api for testing and it consistently appeared perfectly well.

However, slideshow and disktest load their images from disk. I have not the capacity nor the crayons to properly figure this out but it seems like overly-large files might not load right. In the latter's case even the low-resolution 256-color image is consistently broken.

[Kawa-oneechan]

So from the Infinite Snake bug I've learned that there is something about malloc, possibly free, that messes things up at some point. Since loading images from disk involves a fair amount of data even with RLE, this might be related.

This calls for more research.

[Kawa-oneechan]

Well, I looked into it, and I'm stumped. Rarely this would break in the second part with the consecutive large allocations but the numbers add up.

Not shown: the linked list originally used in the snake game seeming perfectly okay.

[Kawa-oneechan]

It was interrupts.

They, well, interrupted the file loader and broke it. The rest was just timing.