IANAL, but here's my understanding of it: 1. Yes. 2. Yes. 3. Yes. No longer the same when all the original code is gone (see BSD Unix). Note, though, that

IANAL, but here's my understanding of it: 1. Yes. 2. Yes. 3. Yes. No longer the same when all the original code is gone (see BSD Unix). Note, though, that BSD is descried as a rewrite, not a refactor. There may be a line short of that, but I doubt it. 4. No. To be completely safe, though, it should be Chinese-walled. 5. I don't think so, so long as they do not distribute B'. Copyright law contains a statutory right to make modifications to enable it to run. Note that this is often done - VM hosts rewrite code for virtualization, the JVM does JIT compilation, and as an extreme example the Transmeta Crusoe processor rewrote x86 machine code into its instruction set on the fly. 6 and 7 - how is the reverse engineering done? If it is a translation of A', it is probably copyright infringement, but this situation is somewhat confusing to me. 8. I hope not! Further, on modern OSes, the non-modifiable executable memory pages should be shared between processes, so there's only actually one copy in memory but it's mapped into both processes. It might violate a EULA or other licensing terms, but that doesn't make it copyright infringement. Copyright itself isn't very clear on this kind of ephemeral copying so far as I know. 9. This one is getting shadier, but I don't think it's copyright infringment. It might violate a EULA, but not all EULA violations are copyright infringement (depending on a variety of factors). 10. I don't entirely understand this one. If the process is designed to be forked and distributed in this fashion, then absent other license terms I'd have to say probably not.