I am working on a distributed computing project. The client for this project uses Twisted (MIT licensed) to communicate with the server. The client will locally invoke a PARI/GP (GPL licensed) script to do all of the computations based on two numbers passed from the client. The PARI/GP script will then return a few numbers to the client, which it will process into information for the server before sending it off. What I want is to have the client itself be closed-source, since this processing it does for the server is to discriminate against fake results. I don't care either way whether the PARI/GP script source is distributed.
I have read this GPL question and it seems that my client and the PARI/GP script would be considered separate programs in an aggregate, since they communicate very basic data and do so rather remotely. This would allow for the client to be closed-source. Are they really separate? Would it change if I offered proprietary alternatives to the PARI/GP script fulfilling the same purpose?
Is it possible that PARI/GP itself could be considered a "System Library" as defined in section 1 of GPLv3.0? If so, would I be allowed to provide a download link to PARI/GP itself, or could I just include PARI/GP in the whole application?
Clarification: The PARI/GP script does all of the actual computation necessary for the results of the distributed computing project. The processing the client does is just to ensure that those computations were actually done, and then show this to the server. That is what makes me believe they are two separate programs.