This is the mail archive of the gcc@gcc.gnu.org mailing list for the GCC project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

Re: Question about aliases


On 6 November 2012 09:23, Alex Markin wrote:
> Also, according to the issue 6.5 (7), we cat access to an object value
> with expression that has
>
>> a qualified version of a type compatible with the effective type of the object
>
> So, `const int *' can legally point to the `int *' but not in reverse
> order,

A `const int*' can point to the same object as an `int*' points to, it
can't point to the `int*' itself (as you wrote).

Accessing a non-const `int' through a const int*' would be a case
where the effective type of the object [int] is accessing through a
qualified version of the type [const int],

If the effective type is `const int' then an `int*' cannot point to
it, because `int' is not `const int' and is not a qualified version of
`const int'.

But that's to do with what a `const int*' and `int*' can point to.
That relates to the alias sets of `int' and `const int' which is not
the same as `int*' and `const int*'

> and that's why `const int *' and `int *' should be in different
> alias sets?

No, they are in different alias sets because a `const int**' and an
`int**' cannot point to the same object. See
http://c-faq.com/ansi/constmismatch.html


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]