A slow day

The plan from yesterday was to continue with the liveness analysis. I did get a start on the gen function which is needed for the liveness analysis. It's mostly done, although there are some todos and questions:

• How to handle the opaque operation? It tells the compiler not to optimize a particular expression, so we would probably like to avoid merging allocations using opaque as well? Needs more thought.
• Blocks, in if- and loop-expressions for instance, are handled quite crudely, by just taking the gen of all the inner statements. I don't think that'll actually give us the result we want.
• There's an Op in BasicOp that I'm not quite sure to handle. In KernelsMem it includes allocations, but maybe it also holds more stuff? I don't know.
• What about variables in index functions? I don't think they are being reported correctly at the moment. I need to figure out why.

I didn't get to do my end-of-week review today. I'll try to write something over the weekend, possibly updating this post.