Summation on HP 42S

[Thomas Okken]

(09-29-2018 11:58 AM)ijabbott Wrote:  Mostly for statistical reasons. There are far more programs than there are compilers, so the probability of a optimization-induced bug being in the program is higher than the probability of the bug being in the optimizer itself.

You're assuming your conclusion there. Circular reasoning.

Until I see some evidence to the contrary, I feel comfortable assuming that all optimizer-induced faults are optimizer bugs. I am honestly curious to see examples of incorrect code that somehow manages to work as intended when not optimized, and fails when optimized. Maybe it exists, but it's not something I've come across.

(09-29-2018 11:58 AM)ijabbott Wrote:  But do you then go and report those bugs / check if they are already reported, or do you just let them lie?

Irrelevant to the topic at hand. A bug is a bug, whether I report it or not.