TESTS

RPC Testing Description
Andrew Mauragis

Initially a testing program was developed to ensure familiarity with sockets 
(which essentially ended up being baby-telnet).  From there a server was
created that printed arbitrary data.  This allowed the testing of the client
side packet construction by inspection of the data output on the server.

From here, the server side decode logic was implemented to deconstruct the
the packet and issue the syscall.  Because the server was already functional,
returning the packets did not encounter many problems.

Once both sides were developed, forking code was added to the server to allow
multiple connections.  At this point, testing was done from a functional
standpoint, where sample client programs were written that would excercise
a variety of different system calls.  

After each system call was individually tested, the program was tested with
another students copy, after verifying that he made the read argument swap.
Initial testing failed, however it was discovered that because he was using a
Mac, there were differences in data type sizes.  The tests in the previous
section were repeated successfully.

To test the code end-to-end, the server was started on guardian and the client
run on a local computer.  A image file was transfered to guardian, and opened
successfully.  A separate image file was transmitted back from guardian
correctly as well.  This proved that both the rclient1 and rclient2 programs
appeared to work correctly with arbitrary binary data.

This test was then repeated, but with both tests run simultaneously.  The
results were success.

The client programs were then invoked with bad arguments, and the client
correctly reported errors.