Carbon Copy Countersign Evidence with airSlate SignNow

Integrate countersign copy

- [Jon] So I want to share something I do in my other job which is training, I'm gonna show you slides based on some work by Dave Abrahams. He showed me this, what I'm about to show you, on copy elision and return value optimization. And it totally opened my eyes about that, because I knew what those were, but I was always a little dubious. Like, how do I know the compiler can really do that? I was, "Yeah okay, people talk about it." And I just didn't believe it. So, he showed how it works which I'm about to show you in a minute, and I was astounded, it's like, "That makes so much sense, and now I get it!" And so I'm gonna show it to you, and I hope you appreciate it, but Dave wasn't that happy. I said, "These are just amazing!" And he said, "No they're not really right." And I say, "What, these are great?" And he says, "They're not so great." "What are you talking about?" I learned something from this that I will remember forever, and he said, "Well yeah but technically they're not correct because of this and this." So I said, "Well fine, I fixed them." So this is my fixed version, but all the credit goes to Dave, the real work is Dave's, really an inspiration. So copy elision is based on the fact that the compiler is allowed to follow the as if rule, which is the compiler is allowed to generate any code that has the same effect as the code you told it to write. (indistinct) all sorts of things, but there's an addition in the standard. The standard actually says that if the compiler is told to copy something, but the copy is not really necessary because the original isn't going to be used again, the compiler is allowed to elide that copy. And you can tell if they've done that if your copy constructor has side effects. And the standard specifically says that your side effects won't happen. The compiler is allowed to elide those. But now I'm gonna show you how and why this happens. The first thing I want to ask you though is this, this function F, how many parameters does it take? Okay so some people here are C++ programmers, and they're all saying zero. And some people are assembly language programmers and they're all saying one. The C++ programmers say none, but assembly language programmers say one, why? What is the return value? At the low level when we have a return value we have to tell the generated code where to put the return value. So that the function is passing that address. All right, so this is what's going on, our function G is gonna call our function F in order to populate a local. So here's our stack frame for G, no parameters passed in, but it has a local X, now we're gonna...