creating a frontend
- To: mathgroup at smc.vnet.net
- Subject: [mg16466] creating a frontend
- From: Chad Cunningham <ccunning at math.ohio-state.edu>
- Date: Sat, 13 Mar 1999 02:22:03 -0500
- Sender: owner-wri-mathgroup at wolfram.com
I'm working on a web based testing program where I need to be able to compare two mathematical functions and determine if they are equal. I've come up with several possible ways to do this, and am starting with the easiest, which is to feed the functions to an external math program to compare. Mathematica is my tool of choice, and with MathLink this is more than possible. The issue now comes down to performance. All of the online testing is done in PHP, a very fast programming language embedded into the web server. To keep things speedy, I don't want to call an external cgi, either a C program I write or mathscript. So, I've written a C program to do what I need, and I'm going to embed it in PHP. The problem being, for a standalone app, this is fine. But, for a web based app, I have to worry about similtaneous connections and heavy usage. So, what I'm wondering is... - Can I make it so that all of the copies of my program connect to one kernel? I don't know if the mathematica kernel can handle this, which is basically equivalent to connecting multiple front ends to the kernel. - How much overhead is there in firing up the kernel and making a connection? What worries me is that each hit would basially require starting a kernel, making a connection, sending and recieving data, closing the connection, and shutting down the kernel. With a lot of hits, I think this would be a resource hog. Ideally I would like to keep a single kernel running in the background and have everything connect to that, but I don't know if the kernel can handle that or if I have to create a new process for each hit.