Hey, dkf! Glad to see this bug taken care of! :)
It looks like that test case expects an error to happen with the badurl, but since there's a -command, the socket gets created asynchronously, which means the error will happen in the command callback, not during the socket creation.
So I think that this is a good thing indeed and working as expected. You'd just have to confirm that the error...
2013-01-21 17:02:26 PST in Tcl