Was FORTRAN buggy 4325
Was FORTRAN buggy 4327
That's my point. These can be impossible to fix because the new circuitry cannot be modeled into the code even though a human can fix it by stringing a wire from here...
But there are now. Why should we not use them because they weren't handy when you were working on TOPS-10?
I'm not saying that test-driven development made sense THEN. But times have changed a lot, and test-driven development makes a lot of sense NOW.
And really, the code has to be tested anyway - what's the advantage of testing it manually rather than automatically?
Was FORTRAN buggy 4326
I don't believe that there are good emulators. Sure, they're better than before. But they are not good where good means they work with a 99.9% success. Doesn't anybody believe in Murphy's Law anymore? I'm...
Yes; how long ago was this?
Right now the computer I'm writing this on has more processing power, disk, and memory than all the servers in my college's server room put together, and the laptop I'm using to connect to it has even more. I have more cycles on these two computers than I can possibly use, and that doesn't include the three computers I have sitting off or idle. Human time, not computer time, is the limiting factor now.
When you have even 3 people working on a system, and cycles are cheap, writing test suites is not frivolous, because it helps you find the bugs that much faster and it helps you make sure the bugs don't come back when you change other things.