I am looking at bypassing the web server and using the API for C to let (possibly) hundreds or thousands of distributed clients exchange info P2P using a tamino server (or a few) or store data onto it. XML is perfect for the kind of evolution I need to do.

Is there a specific port that is used? what kind of encryption capabilities are possible? What limitations exist in terms of the simultaneous number of clients?

Is there a link to the docs and something I can use to evaluate the whole technology?

I’m not entirely certain of the full capabilities of the C API.

However, you can download a Tamino 4.4 starter kit (it’s big!) from here http://www.xmlstarterkit.com/down/download_XMLServer.htm.

This will give you a time limited Tamino 4.4 to experiment with. It also comes with the documentation. There should be documentation for the C API. You would also need to look at the X-Machine HTTP programming documentation to see the sort of stuff you would need to drive Tamino.

If you’re really looking at thousands of distributed clients you don’t want to use the C API. A program using the C API has to run on a system that has XTS and several other Tamino components installed.

Just last week at the Natural Conference I gave a presentation on the APIs for accessing Tamino that has some additional discussion. You can view the paper or the slides online.

From what they advertize it enables you to contact a server and bypass the web. That’s all I am interested in to transfer textual data accross and make sure the server will handle atomicity etc…

How heavy would be the whole installation on the client side?

Also is there a limit on the amount of simultaneous connections? Initially maybe there can be only 50-100 clients at one time.

I’m not sure why you’re worried about bypassing the web server. The reason most Tamino access is through a web server is to provide distributed access, so if you are designing a distributed application you want to go through the web server. I don’t know exactly what you’re thinking of when you mention “atomicity”, but have you looked at the sessioning commands available to the APIs that go through the web server?

The C API is designed for what we’d traditionally call a batch application, and what you’ve described doesn’t appear to fit in that category.