Hi guys,

For those of you that have used .NET Remoting with LLBLGen, do you usually host in a IIS process or do you make your custom executable?

I start with an EXE as the host. Then closer to the release, I wrap the code in the exe into a .NET Service.

One thing that I do that is pretty cool (IMO) is I use a web service as a broker for the remote host. So when the service starts on server A, it registers its location with the web service. When the client starts, it checks the web service for any remote servers. It can pull the address of the first remote service that it finds or it pulls the one with the fewest amount of activity.

The reason I dont host in IIS is because typically I only use remoting for apps located on the same network, and it is very hard to raise events from a remote host running in IIS to a client subscriber.

One thing that I do that is pretty cool (IMO) is I use a web service as a broker for the remote host. So when the service starts on server A, it registers its location with the web service. When the client starts, it checks the web service for any remote servers. It can pull the address of the first remote service that it finds or it pulls the one with the fewest amount of activity.

That is a pretty kool idea!

The reason I dont host in IIS is because typically I only use remoting for apps located on the same network, and it is very hard to raise events from a remote host running in IIS to a client subscriber.

So for clients outside of the LAN you use webservices? if so, just out of curiousity how do you do them, return everything as a string and call writexml() on server, readxml() on client?

I like the idea of IIS as i get a bunch of builtin stuff for it, but i also like the idea of the service as i can customize it completely.

In my web services I use simple objects that get serialized. In the services layer, I take the data from the simple objects received from the web service and map them back to llblgen objects and use a data access adapter to save the data.

When the data is leaving the web service, I have the web service invoke the remote services layer and convert the fetched llblgen objects into simple objects that get serialized and passed to the client consuming the web service.

Yes its extra code, but web services just arent there yet for some things (like abstraction and interfaces.) Another thing to note is that consumers of my services are typically VERY scalled down when compared to the consumers of the remote services layer.

So in essence, I code it like a big onion that has many layers, each layer servicing different consumers and serving different purposes.