const keyword

[Ravi]

I have my TC++ help file which has the following in it:

const my_age = 39;
___________________________
WARNING: A const variable can be indirectly modified by a
pointer, as in
the following:

*(int *)&my_age = 35;

When the const modifier is used with a pointer parameter in
a function's
parameter list, the function cannot modify the variable
that the pointer
points to. For example,

int printf (const char *format, ...);

Here the printf function is prevented from modifying the
format string.
_________________

I can't figure out what they are trying to say.
Could someone help?

TIA.

--
main(){char s[40]="sbwjAeftqbnnfe/dpn!ps!CSbwjACjhgppu/dpn";
int i;for(i=0;i<39;putchar(s[i++]-1));return 0;}

[Josh Sebastian]

On Sun, 12 Oct 2003 23:01:12 +0530, Ravi wrote:

I have my TC++ help file which has the following in it:

const my_age = 39;
___________________________
WARNING: A const variable can be indirectly modified by a
pointer, as in
the following:

*(int *)&my_age = 35;
They're saying that you can cast away const. This might or might not work
all the time (it'll compile, but it might fail in strange ways at
runtime). If your compiler documentation says you can do it, though, it
should be safe enough with that compiler.
When the const modifier is used with a pointer parameter in
a function's
parameter list, the function cannot modify the variable
that the pointer
points to. For example,

int printf (const char *format, ...);

Here the printf function is prevented from modifying the
format string.

They're saying that the string (the object format points to) is what's
const, not format itself.

int const* p; // *p is const
int* const p; // p is const

Josh

[osmium]

Josh Sebastian writes:

On Sun, 12 Oct 2003 23:01:12 +0530, Ravi wrote:

I have my TC++ help file which has the following in it:

const my_age = 39;
___________________________
WARNING: A const variable can be indirectly modified by a
pointer, as in
the following:

*(int *)&my_age = 35;

They're saying that you can cast away const. This might or might not work
all the time (it'll compile, but it might fail in strange ways at
runtime). If your compiler documentation says you can do it, though, it
should be safe enough with that compiler.

With my compiler: no errors, no warnings, no work. The const was not
changed.

Borland TC++W 3.1

[Ravi]

On Sun, 12 Oct 2003 13:34:08 -0700, "osmium"
<r1********@comcast.net> wrote:

With my compiler: no errors, no warnings, no work. The const was not
changed.

Borland TC++W 3.1

Same here. But then why did they write that in the docs?

--
main(){char s[40]="sbwjAeftqbnnfe/dpn!ps!CSbwjACjhgppu/dpn";
int i;for(i=0;i<39;putchar(s[i++]-1));return 0;}

[Rolf Magnus]

osmium wrote:

Josh Sebastian writes:

On Sun, 12 Oct 2003 23:01:12 +0530, Ravi wrote:

> I have my TC++ help file which has the following in it:
>
> const my_age = 39;
> ___________________________
> WARNING: A const variable can be indirectly modified by a
> pointer, as in
> the following:
>
> *(int *)&my_age = 35;

They're saying that you can cast away const. This might or might not
work all the time (it'll compile, but it might fail in strange ways
at runtime). If your compiler documentation says you can do it,
though, it should be safe enough with that compiler.

With my compiler: no errors, no warnings, no work. The const was not
changed.

That's the reason why you should avoid C style casts. You tell the
compiler to be silent and do whatever it takes to convert to the
specified type. You basically say "shut up, I know this doesn't make
sense, but I want it anyway coz I know what I'm doing". Of course you
_must_ actually know what you're doing.

[Micah Cowan]

Rolf Magnus <ra******@t-online.de> writes:

osmium wrote:

Josh Sebastian writes:

On Sun, 12 Oct 2003 23:01:12 +0530, Ravi wrote:

> I have my TC++ help file which has the following in it:
>
> const my_age = 39;
> ___________________________
> WARNING: A const variable can be indirectly modified by a
> pointer, as in
> the following:
>
> *(int *)&my_age = 35;

They're saying that you can cast away const. This might or might not
work all the time (it'll compile, but it might fail in strange ways
at runtime). If your compiler documentation says you can do it,
though, it should be safe enough with that compiler.
With my compiler: no errors, no warnings, no work. The const was not
changed.

Perhaps the documentation is outdated; or perhaps they merely
wanted to point out that it *might* work when executed on other
platforms (of course, it might induce nasal daemons to fly out
your nose, too ;-) )
That's the reason why you should avoid C style casts. You tell the
compiler to be silent and do whatever it takes to convert to the
specified type. You basically say "shut up, I know this doesn't make
sense, but I want it anyway coz I know what I'm doing". Of course you
_must_ actually know what you're doing.

How would that be any less true using const_cast<> ?

-Micah