[Spice-devel] Full featured (qxl compatible) spice web client released

[Gianni Pirozzi]

Hello,
I'm in the process of understanding if spice-web-client would be a suitable replacement for spice-html5 in our company web application.
The first obstacle I'm encountering is that I can't easily control the complete websocket URL.

Our http server behaves as a websocket server for certain values of the requested URL path, and copies data back and forth to a TCP connection. We do not use a separate websocket proxy, it is all written in the Go programming language.
This is trivial to do with the Go standard library, I can provide the code.

This setup is necessary because the browser doesn't (and shouldn't) know where the actual spice server is. Our URLs look like this:
wss://somehost/spice?vm=VMID&sid=SID
The server checks permissions for the session, finds out the spice host and port for the VM, makes the TCP connection and pair it with the websocket.
Another advantage is that the browser only needs access to the standard http(s) port.

I noticed that spice-web-client has its own criterion for generating websocket URLs. What protocol is that, and what is its function?
Is it possible to run spice-web-client in a similar way as spice-html5's SpiceMainConn passing a full URL to it?
Can I expect that spice-web-client will always work regardless of the content of the websocket URL, so that I can construct it for my own purposes?

Any clarification would be helpful,
thanks,
GP