The machine also had a single cycle button and we got up to the offending instruction, and single cycled through it and saw it happen. I pulled out the wiring diagram of the computer, figured out what had to be the problem, and replaced a transistor on a circuit board to fix it. My boss was amazed because he thought I was only a programmer.In the last twenty years, I've found more problems in h/w than in software, which I suppose says good things about the compilers. The regression test suites have gotten very good, I think, and apparently are very effective.For hardware errors, I've seen one where a return with four consecutive 1 bits in the return address would fail. I've seen another where running a memory test of my own authoring would power down the processor -- but only if run on all four CPUs.As you say, however, the biggest "problem" with optimizers is that what they do with incorrect programs is undefined. Fortunately, the compilers and optimizers have gotten much better in that respect as well, with compilers now doing considerably more analysis than 'lint' and similar programs used to do. They're good at finding uninitialized variables, resource leaks, &c, although they tend to generate a lot of spurious warnings.rj
Best Of |
Favorites & Replies |
Start a New Board |
My Fool |
BATS data provided in real-time. NYSE, NASDAQ and NYSEMKT data delayed 15 minutes.
Real-Time prices provided by BATS. Market data provided by Interactive Data.
Company fundamental data provided by Morningstar. Earnings Estimates, Analyst Ra