We are experimenting with multi-touch client/host instead of the regular Windows mouse interface. Getting WM_TOUCH type messages into the client with the multiple touch points is easy but the approach for how to support these in the server (Spice/QEMU) and the guest OS raises some questions. It would seem that there are several options. 1) Add multi-touch messages to the existing Inputs channel 2) Add a multi-touch channel 3) Extend the vdagent protocol And of course, there are the requisite changes to QEMU or the vdagentd running in the guest. I have a vague memory from a prior post that creating channels is complex and not for the faint of heart. Also at question, if you are going to introduce new messages, and thus an extension of the protocol, is it better to limit them to a separate channel that other clients may not know of or is it better to add new messages to an existing channel and then use a capability flag to determine if the new message types can be consumed? Or does it matter if you have to introduce a new protocol version? And do you avoid new protocol versions like the plague. It doesn't really matter much for our experiment, but if the experiment every became real, what would the best way for this to work in the existing Spice clients/servers? Thanks, Brian _______________________________________________ Spice-devel mailing list Spice-devel@xxxxxxxxxxxxxxxxxxxxx http://lists.freedesktop.org/mailman/listinfo/spice-devel