https://gcc.gnu.org/bugzilla/show_bug.cgi?id=121499
--- Comment #6 from Robert Dubner <rdubner at gcc dot gnu.org> --- Andrew, thank you. It's a personality quirk, I suppose. I don't normally concern myself much about compiler optimization for performance. My view is that unless code is in an innermost loop, it doesn't matter all that much how optimized it is for performance. And if the code is in an innermost loop, I tend to be careful about how I construct it. Oh, I know the compiler can do a better job than I can. But if the code is critical, then I do a careful job of constructing it, and my ego is healthy enough for me to believe that even if a compiler can do a better job, it can't be a *lot* better. But now that I know about the problem, I will take a careful look at my violations of strict aliasing, and see what I can do about them. There probably isn't any good reason for me to force the compiler into shoving one foot into a bucket in order to compile my code. Thanks again.
