During the split I noticed that sometimes we tried to create an entry point to code that we decided was unreachable. (after try/catch with both having returns). We should investigate and fix that and make sure we can assert that again.

Putting P1 for now. I hope to do that soonish

This is the affected code: > for (;;) { > try { > throw 3; > } catch(e) { > break; > } > } > for (var i = 0; i < 15; i++) {} The issue is that we don't compile the catch block. As a result we don't know if it has cfg in it. Now we look at the the pc at the end of the catch to know if we can have code after the catch. But that code is never reached. The code is reached through the break. I looked at how we could make it smarter, without compiling the catch block. But I hit blockroads the whole time. I think the path forward here would be to compile the catch block and let prune branches do its job?