I tried to play with tensorflow in Haskell, but I found, as with all nontrivial libraries with external bindings, getting it to install on my system is a headac

I tried to play with tensorflow in Haskell, but I found, as with all nontrivial libraries with external bindings, getting it to install on my system is a headache and a half, better incompatibilities, and massive compile times and memory demands (8GB+ system) and hard disk usage / IO. Unless stuff is published as whole images for docker or virtual box, it's just exhausting to work with non-pure Haskell (and sometimes even pure Haskell with many dependencies).