I like this comment:
>reusable code should be the target of decompilers<

but I don't necessarily see the need for EXACT/original source code re-creation, especially where variables are concerned. As long as the decompiler proggie keeps them straight, I'd think it's ok for the compiler to spit out a variable named "Var1", even though the source code was "MyVar", as long as "Var1" was consistantly named whenever that specific variable was actually used in the target proggie.

Do we want this decompiler to give us, for example, a structure definition too, or just give us the operations on the structures elements, and let the compiler (when operating on our recovered code) generate error messages telling us what (syntatically) is wrong that we poor humans would have to clean up (in this case, by defining that structure ourselves)? This would certainly result in useable, runnable code, but obviously not the EXACT/original source code. How close to the EXACT/original source code are we talking?

Further, if you actually reproduced runnable code, but it was only 80% (or 70%? or maybe only 50%) of the EXACT/original code, and therefore needed some additional user input, would people buy it?

Sarge