Printable View
i know that we discussed it before and i learned that we write main like this:
but when i started my course i was told "the way to write main is":Code:int main()
{
return 0;
}
i was skeptic regarding what they says until i wrote it in visual studio 2005Code:void main()
{
}
and it even didnt put out a warning
whyyyy??
>whyyyy??
Visual C++ is one of the compilers that supports void main as an extension. It's safe on that compiler, but by using it you've removed any possibility of your code being portable.
And for future reference, "it compiles for me" and "it runs for me" are not synonymous with "it's correct".
Maybe the warning level was too low.Quote:
Originally Posted by transgalactic2
EDIT:
Oh yes, more like: maybe you did not disable language extensions.
Warning 2 warning C4326: return type of 'main' should be 'int' instead of 'void'
But it really does require language extensions to be OFF, when they are ON (?) by default.
Good programming practice would have main() return an int, at least. Main should return status upon exit. A good read on this would be
http://www.eskimo.com/~scs/readings/...in.960823.html
I think part of the problem is that in Java, main() returns void, and people that switch back & forth between the languages a lot forget that main() returns an int in C/C++.
What's to remember? main() returns int, simple as that.
Yeah, but then if you try that in Java you get an error. :pQuote:
Originally Posted by hauzer
I have the opposite problem. I try to write int main() in Java all the time.
Not to mention the prevalence of poor compilers which implemented void main, back from the days of DOS (which was so lame at recognising an exit status, that no body bothered), and the piles of cheap pulp-fiction books which had titles such as "teach yourself code mangler C++ in 42 minutes".
The OS returns an int from main, therefore using void main discards that value. Also on the off-chance that another program is monitoring it, using void main could cause problems. Anyway, compilers should at least support int main, if it doesn't, get anew one.
The problem is clearly demonstrated in Salem's avatar.
Actually...Quote:
Originally Posted by P4R4N01D
The program returns something from main, to the runtime, to the OS.
Same here. The result is actually undefined. Your program can return whatever.Quote:
therefore using void main discards that value.
Well I think they should either make void main() illegal (a compile error) or define what it should do, such as alway returning 0.
void main() *is* illegal, unless you have non-standard language extensions...Quote:
Originally Posted by cpjust
QuantumPete
Well, int main returns 0 if you don't return something. So there is no sense for void main to do the same. It should be a compile error with the option to be enabled, since there might be some use of it, like in an embedded system where the value of main won't be checked.
But on platforms where the return value of main() isn't used, why doesn't the compiler just allow you to use int main() but throw away the return value?Quote:
Originally Posted by C_ntua
What do you mean?Quote:
Originally Posted by cpjust
I think what cpjust means is this:
On some embedded systems, main typically returns void because nothing is returned to the operating system. So why could the compiler not just accept int main and ignore whatever main returns?
A cookie to anyone who solves the problem!
In many embedded platforms, where the code starts to execute your code isn't necessarily called main() anyways...
And of course, there's nothing stopping some startup code that calls main from ignoring the return value. We do that all the time with functions like fread, fwrite, printf, sprintf, scanf, etc that are all returning an integer, but most applications do not use that return value.
--
Mats
Precisely, on freestanding implementations the rules we often quote concerning main do not apply at all as it is all implementation defined.Quote:
Originally Posted by matsp
If there is no name mangling, the support library that calls main won't know whether main returns a value or not.Quote:
Originally Posted by Elysia
Of course, on really tiny architectures, returning a value may not be done in a register. Say the system only really has ONE 8 bit general purpose register, so if you return a 16-bit integer value, it has to be passed like a struct is passed back on 32-bit compilers when you return a struct: The compiler provides space for the return value on the stack, and passes the address to that space to the as a hidden parameter to the function - in that case, it would take up extra code in both the startup code and the main function to provide the return value, and if it's completely ignored, the embedded user may not want to "spend" memory on returning a meaningless value.
But void main() is not exactly common.
In our embedded OS, "main" isn't called "main", but it still returns a value.
--
Mats
I think what cpjust means is this:
On some embedded systems, main typically returns void because nothing is returned to the operating system. So why could the compiler not just accept int main and ignore whatever main returns?
A cookie to anyone who solves the problem!
Deja vu?Quote:
Originally Posted by Elysia
Basically, on systems that only support void main(), they could see int main() and change it to void main() when you compile it. I don't deal with embedded devices, so I don't really care one way or another, but that would seem one way to make main() consistent across multiple platforms.
The answer then, is that it could.Quote:
Originally Posted by Elysia