Optimized to death

f0dder: I didn't say that HLLs like C are "evil". Nonetheless, they are maybe even more useful than asm alone. I just tried to explain that "beating asm at optimization" is not the goal of HLLs, but rather a dream. This doesn't imply that you should ALWAYS (even when you make a "debug" and not a "release" version) optimize as much as you can. When debugging, it's not only a waste of time to optimize, but also unclear. When releasing it's a different story.

And besides, algorithms should be decided first. In asm source code or other notes/documents accompanying it, I strongly suggest you write some pseudo-code or HLL code in comments or notes, so the one reading it might get an appreciation for it. Also I encourage pseudo-code since C cannot cover all of asm functions, tricks, etc.

My point was that people tend to forget that HLLs are HLLs, and they start to bash asm and claim it's useless, nonetheless it may be harder for most people, but it can never be beaten by a HLL. Of course plain asm is beaten, but when you program in asm, the optimizing compiler between your ears starts to think differently, low-level, understanding stuff and making decisions on the specific platform, and not abstracted. That's what makes asm a "low-level language". And that's what makes it different than other "high-level languages".

The "primary" goal of HLLs is not "optimization" -- that's asm's primary goal. Also, I know most people say that "it runs fine, so it's good enough", but still it's nice to apply optimizations (i.e not necessarily to "death") wherever you can. Somehow my conscience is hurt by unoptimized code on purpose (like the assistant in Office, etc..). So you see, Microsoft did not have time to optimize the program a bit (even if not in asm), but they did waste their time with pics and other things. Saying "it's a waste of time" is really false, because that's not what they mean when they say optimization is useless.