@ncoghlan on the issue suggested we could do this in the AST optimizer without changing the eval loop. Is that approach worth considering? It would involve transforming a and b or c into (b or c) if a else c.

@ncoghlan on the issue suggested we could do this in the AST optimizer without changing the eval loop. Is that approach worth considering? It would involve transforming a and b or c into (b or c) if a else c.

I'm not sure I see what the general solution is, but I'm all for trying it.

I'll merge this though - we can easily replace this with the alternative idea. The "sequence" pseudo-ops change is worth having anyway.