[C++] strange problem with std::ostringstream

[Eric Boutin]

Hi ! I have a strange problem with a std::ostringstr eam..

code :
#include <sstream>

/*...*/

std::ostringstr eam ss();
ss << "\"\"" << libpath << "\"\" \"\"" << argfilename << "\"\" \"\""
<< outfilename << "\"\""; //line 75

std::string callstring = ss.str(); //line 77

/* ... */

I don't have a undeclared identifier error when declaring std::ostringstr eam
ss();
however, I have thoses stranges errors :

c:\documents and settings\eric\m es documents\progr amming\stdcpp dynamic
code\loadlib.hp p(75) : error C2296: '<<' : illegal, left operand has type
'class std::basic_ostr ingstream<char, struct std::char_trait s<char>,class
std::allocator< char> > (__cdec
l *)(void)'
c:\documents and settings\eric\m es documents\progr amming\stdcpp dynamic
code\loadlib.hp p(75) : error C2297: '<<' : illegal, right operand has type
'char [3]'
c:\documents and settings\eric\m es documents\progr amming\stdcpp dynamic
code\loadlib.hp p(77) : error C2228: left of '.str' must have
class/struct/union type


I use vc++ 6.0

that's a very strange error and I really do not have a clue how to solve
it.. I mean.. ss is not a undeclared identifier.. and it's really defined
as a std::ostringstr eam.. but I have thoses errors...


any help appreciated.. really !!

thanks !

-Eric Boutin

[Rob Williscroft]

Re: [C++] strange problem with std::ostringstr eam

Eric Boutin wrote in news:t_czb.1244 12$Ac1.2065931@ weber.videotron .net:

This:
[color=blue]
> std::ostringstr eam ss();
>[/color]

is a function declaration

std::ostringstr eam ss( void );

change it to

std::ostringstr eam ss;

Rob.
--

http://www.victim-prime.dsl.pipex.com/

[Eric Boutin]

Re: [C++] strange problem with std::ostringstr eam

This is a normal behavior or a behavior from a crappy compiler ?

"Rob Williscroft" <rtw@freenet.RE MOVE.co.uk> a écrit dans le message de
news:Xns9446275 BB2EFEukcoREMOV Efreenetrtw@195 .129.110.130...[color=blue]
> Eric Boutin wrote in news:t_czb.1244 12$Ac1.2065931@ weber.videotron .net:
>
> This:
>[color=green]
> > std::ostringstr eam ss();
> >[/color]
>
> is a function declaration
>
> std::ostringstr eam ss( void );
>
> change it to
>
> std::ostringstr eam ss;
>
> Rob.
> --
> http://www.victim-prime.dsl.pipex.com/[/color]

[Rob Williscroft]

Re: [C++] strange problem with std::ostringstr eam

Eric Boutin wrote in news:eWjzb.1797 0$xP3.571876@wa gner.videotron. net:
[color=blue]
> This is a normal behavior or a behavior from a crappy compiler ?
>[/color]

Its normal standard conforming behaviour.

[snip]. <<-- hint

Rob.
--

http://www.victim-prime.dsl.pipex.com/

[Dan W.]

Re: [C++] strange problem with std::ostringstr eam

On 03 Dec 2003 11:54:53 GMT, Rob Williscroft
<rtw@freenet.RE MOVE.co.uk> wrote:
[color=blue]
>Eric Boutin wrote in news:eWjzb.1797 0$xP3.571876@wa gner.videotron. net:
>[color=green]
>> This is a normal behavior or a behavior from a crappy compiler ?
>>[/color]
>
>Its normal standard conforming behaviour.
>
>[snip]. <<-- hint
>
>Rob.[/color]

I've always been a bit confused by declarations of variables
explicitly calling the default constructor. And now I wonder how would
the compiler not interpret

std::ostringstr eam ss();

as a forward declaration of a function returning an ostringstream.
When exactly is it necessary to add "()" to class instantiation?

Would that only be when instantiating a derived type in the argument
list of a function call like

class bar {};
void foo(bar);
class dbar : bar {};
...........
foo( dbar temp_db() );
?

dan

[Rob Williscroft]

Re: [C++] strange problem with std::ostringstr eam

Dan W. wrote in news:8jqrsvcckf qbdj2jgbc08c7e3 tu5ebg593@4ax.c om:
[color=blue]
> On 03 Dec 2003 11:54:53 GMT, Rob Williscroft
> <rtw@freenet.RE MOVE.co.uk> wrote:
>[color=green]
>>Eric Boutin wrote in news:eWjzb.1797 0$xP3.571876@wa gner.videotron. net:
>>[/color][/color]
[color=blue]
> I've always been a bit confused by declarations of variables
> explicitly calling the default constructor. And now I wonder how would
> the compiler not interpret
>
> std::ostringstr eam ss();
>
> as a forward declaration of a function returning an ostringstream.
> When exactly is it necessary to add "()" to class instantiation?[/color]

Well never (almost). You could write:

std::ostringstr eam ss = std::ostringstr eam();

Unfortunatly it won't work in this case as the streams lack copy-ctor's
but *fortunatly* it unnesassery as the compiler will default construct
ss for you anyway.
[color=blue]
>
> Would that only be when instantiating a derived type in the argument
> list of a function call like
>[/color]

The one and only time it's truly usefull to explicitly call a default
constructor is when initialising a POD (int, double, struct without
constructors etc).

struct X
{
int a; /* a is POD no default constuctor */
X() : a() {}
};

Though a( 0 ) would be just as good above. This is truly usfull within
templates.

template < typename T > struct Y
{
T b;
Y() : b() {}
};

Now Y::b is properly initialized even if T is a type (say int)
without a default constructor. Also unless for some reason we
know that T will always be initializable from 0, b( 0 ) won't
work.
[color=blue]
> class bar {};
> void foo(bar);
> class dbar : bar {};
> ..........
> foo( dbar temp_db() );[/color]

Did you mean

foo( bar() );

or maybe

foo( dbar() ); // watch the slicing !!


struct Z { /*whatever*/ };

int foo( Z const &z = Z() )
{
return 0;
}

Not from all of the above the only place I write variable_name() is
in the initialization list of stucts X and Y, otherwise its type().


Rob.
--

http://www.victim-prime.dsl.pipex.com/

[Dan W.]

Re: [C++] strange problem with std::ostringstr eam

Thank you!

I guess I had foggy memories coming up, like in nameless temporary
construction, such as

class foggy_memory {};
....
throw( foggy_memory() );

Such off sightings often throw me off for a sec... ;-)