"new int[10]" and "new int[10]()"

[peter.xiau]

I found that
std::vector<int > v(10) ;
will automatically initiallize every elem to 0 in the vector,
I check the source code (VS.NET2003),
and I found a line of code like this
*T = new T[n]() ;

So I tested the three lines of code
1. int *p = new int[10] ;
2. int *p = new int[10]() ;
3. std::vector<int > v[10] ;

In Code::Blocks(us ing gcc), the result is:
1. Elements do NOT be initiallized.
2. Elements do NOT be initaillized.
3. Elements are initiallized.

In VS.NET 2003, the result is:
1. Elements do NOT be initiallized.
2. Elements are initiallized.
3. Elements are initiallized.

The result confusing me,
In the C++ standard,
Does "new int[10]()" need to initiallize the elements?
Or the behavior is depend on compilers?

[Alf P. Steinbach]

* pe********@gmai l.com:

In the C++ standard,
Does "new int[10]()" need to initiallize the elements?

Yes. One learns something new every day. I thought the syntax was a
language-extension, but when I looked it up it turns out it's standard.

Syntax in §5.3.4/1.

Semantics in §5.3.4/15.

--
A: Because it messes up the order in which people normally read text.
Q: Why is it such a bad thing?
A: Top-posting.
Q: What is the most annoying thing on usenet and in e-mail?

[Simon Biber]

Alf P. Steinbach wrote:

Yes. One learns something new every day. I thought the syntax was a
language-extension, but when I looked it up it turns out it's standard.

Syntax in §5.3.4/1.

Semantics in §5.3.4/15.

If you're going to use non-ASCII characters in your posts, please
include a valid Content-type header. I assume that you meant U+00A7
SECTION SIGN, which means your post was probably ISO-8859-1. In that
case, it should have had the header
Content-Type: text/plain; charset="iso-8859-1"

--
Simon.

[Old Wolf]

Alf P. Steinbach wrote:

* pe********@gmai l.com:

In the C++ standard,
Does "new int[10]()" need to initiallize the elements?

Yes. One learns something new every day. I thought the syntax was
a language-extension, but when I looked it up it turns out it's
standard.

I thought it was an addition in the 2003 standard..

[Andrey Tarasevich]

pe********@gmai l.com wrote:

...
The result confusing me,
In the C++ standard,
Does "new int[10]()" need to initiallize the elements?
Yes. The elements have to be default-initialized, which is
zero-initialization for type 'int'
Or the behavior is depend on compilers?

The standard requires it. But many older compilers "forget" to perform
the initialization when the '()' initializer is used. This used to be a
popular bug among several compilers.

--
Best regards,
Andrey Tarasevich

[Sumit Rajan]

"Andrey Tarasevich" <an************ **@hotmail.com> wrote in message
news:11******** *****@news.supe rnews.com...

Yes. The elements have to be default-initialized, which is

Not that I can see the difference between a "value initialize" and "default
initialize" in this case, but doesn't the Standard (5.3.4/15) say "value
initialize" in this case?

Regards,
Sumit.
--
Sumit Rajan <su****@msdc.hc ltech.com>

[Andrey Tarasevich]

Sumit Rajan wrote:

...

Yes. The elements have to be default-initialized, which is

Not that I can see the difference between a "value initialize" and "default
initialize" in this case, but doesn't the Standard (5.3.4/15) say "value
initialize" in this case?
...

You are right. Value-initialization is one of the changes made by TC1. I
was just thinking in terms of the original C++98 standard.

--
Best regards,
Andrey Tarasevich