>>58
And what if I unlink the file directly from the filesystem without closing its handle, using handle would subsequently fail anyway because the underlying file doesn't exist anymore.
Are you saying UNIX-like operating systems would handle it much more gracefully? Other alternatives to this would be to keep the handle and the file's contents alive(cached or the actual contents locked until all handles to it have been closed) while unlinking it from the filesystem.
Windows also offers an API for delayed removal/renaming of files in case they're in use ( even if the implementation for this is a bit useless as the 'delay' is usually the next reboot ).
Regardless of what you desire, implementation of any of this behaviour isn't hard, but given your attitude, I'd believe you don't really know anything about how the NT kernel works internally, so arguing with you is pointless, as you don't see that this is a non-issue and if there would be a need for this, it wouldn't be hard to implement by any seasoned Windows driver developer.