volatile const

[Rahul]

Hi Everyone,

The following code compiles without any error,

int main()
{
volatile const int a = 10;
return(0);
};

But i expected a compilation error as a is being told as a constant
and at the same time indicating that its value could change outside
the program's control...

Why isn't this an error?

Thanks in advance ! ! !

[Andrey Tarasevich]

Re: volatile const

Rahul wrote:

The following code compiles without any error,
>
int main()
{
volatile const int a = 10;
return(0);
};
>
But i expected a compilation error as a is being told as a constant
and at the same time indicating that its value could change outside
the program's control...
>
Why isn't this an error?

Why should it be an error? This is exactly what 'volatile const' is
supposed to mean in the first place: the program is not allowed to
change the object, but its value can change "by itself".

--
Best regards,
Andrey Tarasevich

[Richard Bos]

Re: volatile const

Rahul <sam_cit@yahoo. co.inwrote:

int main()
{
volatile const int a = 10;
return(0);
};
>
But i expected a compilation error as a is being told as a constant

No, it's being declared as something _you_ won't change.

and at the same time indicating that its value could change outside
the program's control...

Quite. It is something your code won't change, but something else will.
One example of this could be a hardware counter, or an input port.

Richard

[Joachim Schmitz]

Re: volatile const

Rahul wrote:

Hi Everyone,
>
The following code compiles without any error,
>
int main()
{
volatile const int a = 10;
return(0);
};
>
But i expected a compilation error as a is being told as a constant
and at the same time indicating that its value could change outside
the program's control...
>
Why isn't this an error?

Because for some stupid reason in C const doesn't mean 'constant' but 'read
only'

Bye,Jojo

[Richard Tobin]

Re: volatile const

In article <ftkmd2$v35$1@a ioe.org>,
Andrey Tarasevich <andreytarasevi ch@hotmail.comw rote:

> volatile const int a = 10;

>Why should it be an error? This is exactly what 'volatile const' is
>supposed to mean in the first place: the program is not allowed to
>change the object, but its value can change "by itself".

The standard gives this example:

extern const volatile int real_time_clock ;

You can't change the time, but it changes anyway.

-- Richard
--
:wq

[raof01]

Re: volatile const

Hello Joachim,

Because for some stupid reason in C const doesn't mean 'constant' but
'read only'
>

Yes, absolutely. Furthermore, "const" means read only JUST for you. On the
other hand, "volatile" doesn't mean you can modify it but you HAVE TO reload
it from memory every time you use it. "volatile" is more like an indicator
for compiler to protect from optimizing.

raof01
"Thoughts are but dreams till their effects be tried. -- William Shakespeare"

[Chris Thomasson]

Re: volatile const

"raof01" <raof01@gmail.c omwrote in message
news:7c6e25954a 388ca6bb36481ec c0@news.cn99.co m...

Hello Joachim,
>

>Because for some stupid reason in C const doesn't mean 'constant' but
>'read only'
>>

>
Yes, absolutely. Furthermore, "const" means read only JUST for you. On the
other hand, "volatile" doesn't mean you can modify it but you HAVE TO
reload it from memory every time you use it. "volatile" is more like an
indicator for compiler to protect from optimizing.

Its not clear at what level a compiler will stop optimizing; its all
implementation defined. It seems like link-time optimizations can mess
around with volatiles if they can obtain certain levels of information... I
think smart-a$@ compilers think they can reorder something, and ultimately
end up generating phantom behavior under the random nature of various
race-conditions. This is especially relevant when the 'volatile' keyword is
used in conjunction with multi-threaded algorithms...