Prevent hard-coding of server name or IP about pityoulish HOT 3 CLOSED

Running two servers in parallel makes sense. Preferable when there are two server implementations. For the socket exercise, the second server could be in a different programming language. Or one with blocking IO and the other with asynchronous NIO and connection keep-alive.

Some ideas on how to integrate this into the classroom exercises:

1. Run two servers on different ports of the instructor's machine, or have two instructor machines in the network. Let students test every step against both servers.
2. Run a second server in the public internet, for example hosted in a cloud service. After students have fixed the client, it should run without recompilation against the second server, too. For me, this would link the exercise to another topic I'm covering in my lectures.
3. Provide a server implementation that students can run on their workstation. Let students test every step against the local server as well as against the server on the instructor's machine. This will tempt students to look at the local server implementation and/or to import that into their IDE, see issue #25. An advantage is that students would have access to the log file of their local server, and that they can select a different log level there.

from pityoulish.

As a stop-gap measure, I'm improving the inline comments to explicitly mention that host and port should be taken from the command line. I wonder how many are going to ignore that comment anyway ;-)

from pityoulish.

There's no point in pursuing this further. If students choose to ignore instructions, I can't help it.

from pityoulish.