Error codes are unsigned 32-bit values which are private to each
server session. Most methods will set the last error code value when
they fail; a few methods set it when they succeed. Function failure
is typically indicated by a return value such as FALSE, NULL,
INVALID_SERVER or HTTP_ERROR.
If the dwError parameter is specified with a value of zero,
this effectively clears the error code for the last method that
failed. Those methods which clear the last error code when they
succeed are noted on their reference page.
Applications can retrieve the value saved by this method by
calling the GetLastError method to determine the specific reason
for failure.