[PRL] Joel on Software - Making Wrong Code Look Wrong

[Johan]

Paul A. Steckler wrote:
> In my current project, we have some too-clever code that only returns
> a meaningful value through an exception, and otherwise returns unit on
> a normal return.  During a code review, I pointed this out as a misuse
> of exceptions.  But we have lots of good uses of them, too.

Small sidetrack:

Can you do any meaningful strictness analysis in the face of exceptions?
Or do you have to thread some exception monad through it and serialize 
execution that way?