Summation on HP 42S

[Albert Chan]

(09-28-2018 11:20 PM)ijabbott Wrote:  Question is, how do you know the optimizer is at fault? It could be at fault, but more often than not it is invalid assumptions of the programmer that is at fault. (No accusations intended here. I'm just stating the general case.)

I agreed that many times we don't know the answer.
What assumptions the optimizer were using ? Who knows ...

The fully optimized code might still work.
But, once in a while, it may hit a wrong assumption, causing undefined behavior.
If lucky, the rare cases only affect some rounding errors. Again, you don't know.

With so many unknowns, you are better off without too much optimization.
Example: I will never buy a self-driving car (optimized away the driver)