Point #1: Winsock is just that, Windows sockets; obviously it's tied down to the Windows platform, without a serious rewrite.
Point #2: "WSA functions" or no, the I/O completion port model is only used by Windows (or well, nothing exactly like it's found anywhere else -- [s]Solaris[/s] Slowaris (edit: pretend there's strikethrough, d'oh forums) is about as close as it gets). Other operating systems prefer alternate models, such as epoll (*nix) & kqueue (fbsd). (NOTE: I differentiated *BSD from regular old *nix!).
Point #3: You just realised it was heavily tied to Windows? By Winsock? What about MFC!? That's extreme Windows.
Regardless, although I only update that project infrequently, several major changes are in motion (some not yet pushed) to rework/stabilise their code. Removal of MFC is a priority, networking code rewrite also, but porting it to other operating systems -- highly doubtful (I can imagine it would only end badly for most users).
In addition: why the strong love for PDO? Saw someone else pressing its use? PDO isn't a must. It's great and all, but it's not a must - probably it's strongest argument for use is its emphasis on parameterization, but other drivers can handle that fine (even ODBC, in a limited form [i.e. no named parameters], which is still perfectly acceptable for use). The second strongest argument would be its use of exceptions, however exceptions are slow. For typical panel use, there's absolutely nothing wrong with raw odbc_*(), or mssql_*() functions. For larger applications sure, PDO is advisable.
Not that any of that matters to you, or this thread. I just wish you'd try to at least start reading up on things before you start spouting nonsense... it may sound like magical flying unicorns to the typical poster here, but to the seasoned programmer all you're saying is BS. Troll elsewhere!