assert() in C++

Hi, all. I'm watching a CPPCon 2016 video by Patrice Roy, in which he uses the code:

Code:

---

assert(denom && "divide by zero");

---

1. This is the same assert() function that is provided by the C language in the assert.h header, right?
2. Would a C++ programmer normally use assert()? I presume that for a compile-time assertion you would use static_assert().
3. What is the point of adding a string literal there? Does the assert() macro do something special with it? Or is it just a way of marking the relevant line of source code? Isn't it the same as :

Code:

---

assert(denom); /* divide by zero error if this asserts */

---

TIA,
Richard

1. I'm not sure, but I'll say yes. It's a macro btw, not a function.
2. You may want a run-time assertion instead though.
3. assert() prints its expression if it fails. The programmer may want to have a visual queue on what went wrong. Maybe even explain to the user what went wrong.

Quote:

---

Originally Posted by Richardcavell

1. This is the same assert() function that is provided by the C language in the assert.h header, right?
2. Would a C++ programmer normally use assert()?

---

1. For C++ you write #include <cassert>
2. Yes. It is pretty convenient. If you don't want to add string literals to your assert statements another approach is to write descriptive names for the things you're checking and put them in a namespace:

Code:

---

assert(Test::default_Money_is_zero());
assert(Test::negative_Money_throws());
assert(Test::fractional_Money_rounds_properly());
assert(Test::currency_conversion_works());
assert(Test::Money_plus_Money_works());
assert(Test::Money_times_Money_throws());
...

---

Inside those functions you would check stuff using whatever logic you want. Print diagnostics and return false if something goes wrong. In your release code, you will typically compile with NDEBUG defined, which will compile out all of these calls, making the production code faster.

Quote:

---

Originally Posted by Richardcavell

Hi, all. I'm watching a CPPCon 2016 video by Patrice Roy, in which he uses the code:

Code:

---

assert(denom && "divide by zero");

---

3. What is the point of adding a string literal there? Does the assert() macro do something special with it? Or is it just a way of marking the relevant line of source code? Isn't it the same as :

Code:

---

assert(denom); /* divide by zero error if this asserts */

---

TIA,
Richard

---

Assert() evaluates the expression within the (), and returns 0 if the argument is false. Such as Assert(1==0) would return a 0, and Assert(1==1) would return a true, so the string literal is only used to test the expression, to determine whether true or false should be returned.

Quote:

---

Originally Posted by Terrance

Assert() evaluates the expression within the (), and returns 0 if the argument is false. Such as Assert(1==0) would return a 0, and Assert(1==1) would return a true, so the string literal is only used to test the expression, to determine whether true or false should be returned.

---

A standard-compliant assert() returns nothing. If its expression is true, nothing happens. If false, it prints an implementation-defined message, normally including the line and file where the error happened, and calls abort().

Unless that capitalization wasn't a typo, and you literally meant "Assert()", in which case you could be right, but I wouldn't know.

Quote:

---

Originally Posted by GReaper

Unless that capitalization wasn't a typo, and you literally meant "Assert()", in which case you could be right, but I wouldn't know.

---

I apologize, I meant assert (). The a automatically goes to uppercase when you type it in at the beginning of a sentence, and I put down Assert in all instances, though I meant assert ().