So there is a C++ application to be tested via a component test and there is a 'test-flag' passed to the test executable to indicate that it shall be executed in 'test-mode'.

That is because some external dependencies aren't available and shall be left out or simulated in the test run.

I was arguing with a colleague, that we are effectively testing a different software by doing so.

My colleague on the other hand stuck to the point, that it's still the same software being tested, because in his mind software equals software build. For changing the build flags would result in 'different software'.

I was trying to convince him, that software is a more abstract term than its manifestation in an exectuable file, and that by his definition there could be two branches within the executable doing completely different things based on the 'test-flag'.

Actually I'm still triggered by his stupidly narrowminded definition of software.

</rant>