> but people are claiming that software A must be a derivitive of library B due to the fact that there is no other library that it works with, if there was another library then it may not be a derivitive of software B.
"Is there an alternative implementation?" is indeed not a rigorous way to express this "test". The real question is: "Is the API of library B an original, copyrighted creation of B?" (assuming that a copyright on just an API can be legal - but please let's not start over).
> this is what isn't logical, and as others point out, is a very dangerous precedence to set.
"Logical", "legal" and "dangerous" are different things. Please avoid mixing them in the very same sentence, otherwise it looks like your reasoning is not sheer but tainted by your agenda.
> it would make all free software that was written for windows derived from windows
Yes. But since Microsoft licenses are absolutely not using the concept of "derived work" like the GPL does, this would be a very different story. (about this difference see other posts above)
> (because that's the only OS it was written for)
Not really, but rather because: win32 and other MS APIs are original, copyrighted creations of Microsoft.