I've been told on several occasions that it's a real NO NO to ever have a simple
return;
without at least setting a 0 value to it.

Even if the calling program never reads the return value, perhaps it's still a wise thing to do?

Fair enough, but what's the correct way to terminate a program rather than a function?

I've seen lots of scripts with return; in the program body, but maybe we should be using exit; instead? Does it matter in the slightest? Not that I've seen exit documented.

I also find it curious that program/endprogram statements are in fact not necessary to compile a script - ie ecompile doesn't generate any errors.

Or even that you can put a ; at the end of the if() line or not. Compiler doesn't care.

Sorry to bring these up, it's been in the back of my mind for years but I never got around to checking up on them before.

This is just my take on it:

It's like, in C, when you write: It lets you know that you actually DO mean to pass ( or in this case, return ) a blank value. That just makes it easier to read "at a glance" by yourself AND others.

Another reason is that it provides a certain level of consistency between scripts that a return DOES matter, and scripts that a return DOESN'T matter in. Some scripts return a value to the core, and depending on the value they return, the core will perform a different set of actions, or no action at all. In this case, you NEED to know what your passing. You can pass return; to the core, and hope it's interpreted as 0 every time- but then why not just pass return 0; and be sure? This isn't nessessary in all scripts; in fact, it's completely irrelevant in most scripts.

So, in short, it's not that it's wrong or evil so much as it's "frowned upon as improper". It doesn't matter that often, and more-or-less parallells the idea of using void main() instead of int main( ... ) in a program.

OldnGrey wrote:I've been told on several occasions that it's a real NO NO to ever have a simple
return;
without at least setting a 0 value to it.

Even if the calling program never reads the return value, perhaps it's still a wise thing to do?

I would wonder if there's some specific eScript reason that it's better to slap a 0 at the end of that return;...

Other than that, though, my only thing is to keep it consistent; if a function truly returns no value, and -NEVER- returns a value, I don't see the harm. In practice, I usually don't end up with many plain return;s without some value, because at minimum, I end up returning a success/error value (thinking HRESULT?)

But for very simple functions that do something, and where any 'error' condition would be truly irrelevant, I do a return;, or simply let the function fall through (which I assume has the same result).

Essentially, I look at a function 'falling through' or exiting with 'return;' as a void function. Certainly void functions are not 'huge no nos' - though again, you might often want to return some sort of status code from the function.

(Just for the record, too; for those not aware, When I say 'void function', I do not at all mean the same thing SMJ is referring to. SMJ's "SomeFunction(void)" is a function that has no parameters and uses the 'void' term simply to make that explicit, where a void function is one that returns no value.)

I'd be interested in knowing if there was some eScript-specific reason for this suggestion, though.

OldnGrey wrote:Fair enough, but what's the correct way to terminate a program rather than a function?

Again, I'd be interested in knowing if there's a 'best practices' way to do this that is based on some objective situation in eScript. Personally, I look at 'Programs' as simply a function that happens to be something of an 'entry point' to the script - sort of like it's an "int main()"; it's just another function, but it has a special relationship with the environment because it can be called from outside of the unit of script code in question.

But in this case, the return value is usually not important. I should add, "to me"; because I have eliminated "run_script_to_completion()" calls from my shard entirely; For me, most 'programs' might as well be void functions, and I treat them as such. Obviously, system hooks and such are a different story.

Something contains... NOTHING. It would be an uninitialized variable. If you use it as a conditional it would be equivalent to 0.
Something in this case is obviously 0.

The second approach is done in the distro because it keeps the values safer to use. It has a more predicable outcome if something catches the return value from a function and later tries to use it by copying it to another function or using it with some sort of operator.