Armando Ramirez

GPU_FlushBlitBuffer() didn't make a difference either.

I'll confirm I can make a minimal repro in C.

Here's a C repro: https://gitlab.com/macaroni.dev/sdl-gpu-hs/-/merge_requests/11/diffs

[That works!](https://gist.github.com/ramirez7/f9b0e7849ac83e7d6f3f7c06e8055890#file-mylib-hs) Is there any reason for `repeat` in `liquid-base` to not have that type as well?

I've implemented the fix [here](https://gist.github.com/ramirez7/f9b0e7849ac83e7d6f3f7c06e8055890/bbb9fd4ab7f0c3d52ecfed2fcb461b3cb8415a80)! I also added some a postcondition to `take'` [here](https://gist.github.com/ramirez7/f9b0e7849ac83e7d6f3f7c06e8055890/7424dbbe15c5b81fd741b718e77e1eb77f17052d): ```haskell {-@ take' :: xs:[a] -> {i:Int | i >= 0 && i < len xs}...

> I'm not aware of any reasons. There are plenty of opportunities to improve LH output. The resugaring step should be not too difficult and looks like the biggest improvement....

I guess `spine` of an infinite list is `_|_`, so you're asking for trouble :grin:

Thanks for the link! I'll look into it - I'm using nixpkgs-unstable + haskell.nix, but maybe it is lagging behind and hasn't picked up that PR yet.

Looks like my the `nixpkgs-unstable` in `haskell.nix` is from September https://github.com/NixOS/nixpkgs/commit/e14f9fb57315f0d4abde222364f19f88c77d2b79. I'lll try bumping and see if that fixes it

Not yet. I'll give it a go next week (I'm focused on preparing for Ludum Dare 52 this week!)