Hi,

First of all, thanks for xmlrpc, we've recently finished a public demonstration of a client-server application where the clients are mobile devices (phones, PDA's etc).

The server serves a map and publicly authored content associated with points on the map.

The idea is that as you navigate the city you can leave (author) content as you go, and read/view things other people have authored.
For fuller details of our project see this: http://www.proboscis.org.uk/urbantapestries/index.html

ATM it is pretty much being held back by the lack of network coverage and the poor resolution of location awareness, but it was a research project not commercial software, and proved the concept to the satisfaction of our funders.

So to my point... 
We used XMLRPC to give us ready made RPC functionality across a broad range of client platforms and I'd already used it in another project so I knew what to expect.

The issue we have encountered is that not only is the XML quite big for our limted bandwidth use-case, but parsing it is also quite involved for a limited capability device.

I've seen discussions of zip'ing the xml to cut down on bandwidth, which will might possibly help (or just offload the bottleneck to the device again!), is there any similar action which could be taken to reduce the XML parsing effort, a simpler schema that could be used perhaps?

We deliberately restrict ourselves to only using String's for values throughout, but make much use of structs and arrays.

Have any other users encountered these (or similar) issues, and what, if any, solutions could you share with us?

Does anyone see a way in which we could be using xmlrpc more efficiently?

Thanks for your time, and a great little product.

d.