must the imperative programming paradigm comply with the forward declaration rule?

Question

In a exam in my course of Language Programming was this question:

Is the next program valid in a imperative programming?

   int a = 0;
   z = a + 2;
   int z = 3;

was not specified or grammar, or syntax, or anything extra. My answer was YES because eventually there may be a language, a syntax and a grammar design for wich is valid use z without declare it and 're-declarate' z later. My grade was 0.0 but I still find it unfair. could help to clarify the issue?. Thank.

Answer 1

You were marked unfairly.

Paste the following code snippet into the debugger console of your browser:

(function() {
    var a = 0;
    z = a + 2;
    alert(z);
    var z = 3;
    alert(z);
})()

Answer 2

Considering your course is titled "Language Programming", I imagine "imperative programming" has to be interpreted strictly. Quoting the current tag-wiki for imperative, we have

Imperative programming is a paradigm of expressing the logic of a computer program or computation by explicitly describing its control flow in terms of statements that change a program state.

Arguably the most reasonable interpretation of the given C-like code is that int z = 3; is a variable declaration (and assignment), and the line z = a + 2; is thus usage before declaration. Assuming this is a legal program, this does not follow the imperative paradigm since its control flow was not explicitly described i.e., it wasn't sequential.

---

That said, you are right that since there was no specified grammar or syntax, the program could literally do anything, including what you specified, calculate π, or display a pink elephant on the monitor. Obviously, we need to draw a line somewhere. Given that your course is titled "Language Programming," I think that critically evaluating how reasonable your specification for a possible language is a good way to go.

We have plenty of languages where simple use of a previously unknown variable implicitly declares it, so that's quite reasonable. However, the "re-declaration" part of it, I find quite unusual and confusing and believe that it would be a bad language feature. I'm unaware of any major programming language which allows it¹.

Therefore, I believe that your answer of "YES" in that narrow circumstance should at best have been a side note in an answer where the main answer was "NO".

---

1. Similar, but unrelated cases, are where different variables have the same name but different scopes, and where the same variable can be assigned objects of different type.