3 Replies Latest reply on Sep 1, 2009 6:42 AM by jlakumb

    Paths and Default Servers


      Cross posted from the blog of the same name:


      PI SDK, as everyone knows, will use the default server if you request a tag without specifying a server, e.g., sinusoid rather than \\server1\sinusoid.  Off the top of one's head, then, you would want PI WS to support the same behavior for pi: and pe: paths.


      Or would you? PI WS is remote to the caller (well, usually).  For implementation reasons I won't go into here, there could be yet another hop involved.  Consequently, the user making the call might not know the default, or the default might have been changed without his knowledge.


      What's your opinion?  Allow paths like pi:sinusoid to use the configured default server of the remote machine, or insist that any pi or pe path name the server?

        • Re: Paths and Default Servers

          This is kind of like Relative paths and Full in URL's. I will think about it a bit and comment.

          • Re: Paths and Default Servers

            I am always explicit in server connections and don't use default servers (or default PI system, default database in AF).  My vote would go for "insist they name the server".


            Besides from a debugging point of view surely you want to know where your data is coming from rather than rely on the PISDK default server configuration on the PI WS/DS server.

            • Re: Paths and Default Servers

              There is a use case for partial tag names...  An application can easily be moved from the test PI Server to the production PI Server by simply changing the default server.  Many customers take advantage of this functionality in the PI Data Access products.  For this reason (and for consistency) we should support it with PI Web Services too.