It's not uncommon, as I stated above. The diagram of a serial cable shows pins 4 & 6 and 7 & 8 connected together on one end of the cable with no connection on the other end. Note - this is the FIRST picture that came up when I searched for RS232 cable connections.
Also, I would argue in favor of the port expanders over shift registers and ring counters because every pin may be an input or an output. They are only a buck a piece, so the cost is reasonable and using them also opens up the possibility of testing from intermittent shorts and opens with the compare on change features.
Here is something I put together about a year ago for testing 30-Way ribbon cables... feel free to adopt this to your application where and if you see fit.
Maybe it is Possible to check the Impedance of each wire.
Give a short pulse to each wire and listen to the echo.
The source impedance of the measureing eqipment has to fit with the wire impedance.
The Time between the pulse and the echo will give You the lenght of the wire.
Ist there something shorten or open the signal ist not flat, or the end echo comes to early,
Because the Impedance of the Wire changed.
The reference can be built with a functionally cable and stored in an Table.
Therefor are very fast A/D Converter is to use to get any echo peak.
Bad Connections at the End of a wire ( e.g. bad crimp connections ) can not be detected with this method.
I've seen this Method in a ethernet cable Tester. Sorry but the source I don't remember.
Now we are talking nS and fractions of nS. Some thing beyond most computers.
I know of a machine that puts a 10 volt pulse on one circuit at a time and measures how many mV comes back on all the other circuits. They do not measure time but just signal strength. (8051 and ADCs)
On a cable any movement will effect the impedance. On a PCB the impedance is fixed and will not change. On the PCB we often test after the connectors are attached. If one pin is missing we can see the capacitance/impedance change.
Maybe you can tell I have designed test machines for sever companies that look for opens/shorts/wire length/ etc.
For cables, mostly the tests are for connected/not connected and shorted to a neighbor and wire in wrong place. Almost never have I seen a resistance test. (only where wires are parallel for high current)
There are lots of interesting suggestions here, but before we go off in the realms of Time Domain Reflectometry, it would be good to hear from the thread starter (Becky) regarding the exact requirements of this cable harness tester.
Or, maybe sufficient guidance has already been given to enable the design of a simple continuity tester.