Pretty much what the title says. Is it ok to run DeadCodeElimination before NoPythonTypeInference? I see that DeadCodeElimination is placed inside typed_passes module. Thus, I wonder if it is only safe to do after type inference?
From memory, I seem to recall that it wasn’t safe due to issues with aliasing and that to guarantee safety a full type map was required. It may be that alias analysis has now been altered/fixed and it is working fine, @DrTodd13 or @ehsantn may know/recall more. Perhaps open an issue on github, if the pass doesn’t have a dependency set on type inference then perhaps it ought?