I've got one team of students in my Software Development class trying to finish the second in our series of games on earthquake preparedness.
After spending a huge amount of time manually testing the first game, I tried hard to make it possible to write automated tests for the second one. Among other things, all controls are with the keyboard (as opposed to the mouse) and movement is on a discrete grid.
At some point before this semester started, we blinked and the tests stopped working. I thought a good first task for the students would be to repair and complete the tests, so that (among other things) we could push a button and watch the game 100% itself.
After a week or two, there was a revolt. Trying to set up an automated testing framework that interacts with Unity, MoreMountains' TopDown Engine, Pixel Crushers' Dialogue System, and everything that a rotating body of students have written over a couple of years proved to be overwhelmingly difficult. In a discussion of my priorities as the "customer" (and the good/fast/cheap trilemma), one student said, "It feels like the tests are the product." They all agreed that automating this would be more work than testing manually.
I relented, but slightly: we'll still have tests, but they'll be scripts for humans to work through rather than automated. A slight upside is that non-programmers working on the project can run these tests.
In the "real" world, a manager with a spreadsheet would guess at how much testing would be required over the lifetime of the software, then guess at how many more hours of programmer time would be required to finish the testing suite, then compare the programmer's salaries to the salaries of the human manual testers. If the latter was cheaper, she'd fire the programmers writing the test suite and use the cheaper human testers.
@/Pat
>"In the "real" world, a manager with a spreadsheet... "
Actually, that's how businesses are supposed to do it. But in the real "real" world, the manager would contact her company's biggest competitor, Evilco, and cut a secret deal with that CEO to give her a C-suite position in exchange for sabotaging the release of her current company's game.
Then she'd go back to her current boss at Badco, and tell him that they can save a ton of money and release the game sooner if they cut back on testing and fire more people. Then when the game releases early, she collects her bonus and moves to her new position at Evilco.
When Badco's game fails because it wasn't properly tested, and Evilco monopolizes the market, she blackmails Evilco's CEO into letting her run a new subsidiary they acquired with the profits they made from having a monopoly in that game market.
(But I guess we probably shouldn't teach our kids to be so cynical, huh? They'll learn that on their own.)
