Thread: [Dev-C++] templates in C++

SourceForge Headquarters 1320 Columbia Street Suite 310 San Diego, CA 92101 +1 (858) 422-6466

Hi,
Does anybody know where I can find a detailed description on templates
syntax in C++ ?

    Thanks.

                Erwan

well, this is my problem:

I would like to make a template class that owns a set of elements that have
the type of one of the class parameters. And it doesn't compile...
example:

#include <set>
template <class T>
class A
{
    public:
        A();
        B();
        std::set<T> _list;
};

the compiler's output tells that the type T is not defined... Well, of
course, but it will be defined when the template will be instanciated. So it
should be possible...
Does anybody know a solution to my problem?

-------Original Message-------=0D
 =0D
From: erwan ancel=0D
Date: =CA=F5=F1=E9=E1=EA=DE, 12 =C9=E1=ED=EF=F5=DC=F1=E9=EF=F2 2003 12:28=
:13 =EC=EC=0D
To: dev...@li...=0D
Subject: [Dev-C++] templates in C++ : precisions=0D
 =0D
well, this is my problem:=0D
 =0D
I would like to make a template class that owns a set of elements that ha=
ve=0D
the type of one of the class parameters. And it doesn't compile...=0D
example:=0D
 =0D
#include <set>=0D
template <class T>=0D
class A=0D
{=0D
public:=0D
A();=0D
B();=0D
std::set<T> _list;=0D
};=0D
 =0D
the compiler's output tells that the type T is not defined... Well, of=0D
course, but it will be defined when the template will be instanciated. So=
 it=0D
should be possible...=0D
Does anybody know a solution to my problem?=0D
 =0D
 =0D
=3D=3D> If you remove that B() call, the code should compile.=0D
 =0D
 =0D
 =0D
Ioannis Vranos=0D
 =0D
* Programming pages: http://www.noicys.freeurl.com=0D
* Alternative URL: http://run.to/noicys=20

This is only a function declaration, not a call. Anyway, somebody on the
gcc-help mailing list gave me a solution that seems to work: replace
"template <class T>" by "template <typename T>". But I don't know why ...

----- Original Message -----
From: "Ioannis Vranos" <no...@ho...>
To: <dev...@li...>
Sent: Sunday, January 12, 2003 12:38 PM
Subject: Re: [Dev-C++] templates in C++ : precisions

-------Original Message-------

From: erwan ancel
Date: =CA=F5=F1=E9=E1=EA=DE, 12 =C9=E1=ED=EF=F5=DC=F1=E9=EF=F2 2003 12:28=
:13 =EC=EC
To: dev...@li...
Subject: [Dev-C++] templates in C++ : precisions

well, this is my problem:

I would like to make a template class that owns a set of elements that ha=
ve
the type of one of the class parameters. And it doesn't compile...
example:

#include <set>
template <class T>
class A
{
public:
A();
B();
std::set<T> _list;
};

the compiler's output tells that the type T is not defined... Well, of
course, but it will be defined when the template will be instanciated. So=
 it
should be possible...
Does anybody know a solution to my problem?

=3D=3D> If you remove that B() call, the code should compile.

Ioannis Vranos

* Programming pages: http://www.noicys.freeurl.com
* Alternative URL: http://run.to/noicys

-------------------------------------------------------
This SF.NET email is sponsored by:
SourceForge Enterprise Edition + IBM + LinuxWorld =3Domething 2 See!
http://www.vasoftware.com
_______________________________________________
Dev-cpp-users mailing list
Dev...@li...
TO UNSUBSCRIBE: http://www.noicys.cjb.net/devcpp/ub.htm
https://lists.sourceforge.net/lists/listinfo/dev-cpp-users

-------Original Message-------=0D
 =0D
From: erwan ancel=0D
Date: =CA=F5=F1=E9=E1=EA=DE, 12 =C9=E1=ED=EF=F5=DC=F1=E9=EF=F2 2003 3:51:=
27 =EC=EC=0D
To: dev...@li...=0D
Subject: Re: [Dev-C++] templates in C++ : precisions=0D
 =0D
This is only a function declaration, not a call.=0D
=0D
=3D=3D> Then you declared a function without a return type.=0D
=0D
Correct it to void B();=0D
=0D
=0D
 Anyway, somebody on the=0D
gcc-help mailing list gave me a solution that seems to work: replace=0D
"template <class T>" by "template <typename T>". But I don't know why ...=
=0D
=0D
=0D
=3D=3D> The typename has exactly the same effect. They are both equivalen=
ts for
the definition of template arguments.=0D
=0D
=0D
Ioannis Vranos=0D
=0D
* Programming pages: http://www.noicys.freeurl.com=0D
* Alternative URL: http://run.to/noicys=20
>>The typename has exactly the same effect. They are both equivalents =
for the definition of template arguments.

With gcc, they don't. I did some testing: typename compiles, and class =
doesn't. Perhaps gcc tries to compile something more when it knows the =
template parameter is a class...

This is non-sense (at least). Did you correct B() as i suggested?=0D
=0D
 =0D
Ioannis Vranos=0D
=0D
* Programming pages: http://www.noicys.freeurl.com=0D
* Alternative URL: http://run.to/noicys =0D
-------Original Message-------=0D
 =0D
From: erwan ancel=0D
Date: =CA=F5=F1=E9=E1=EA=DE, 12 =C9=E1=ED=EF=F5=DC=F1=E9=EF=F2 2003 5:04:=
32 =EC=EC=0D
To: Dev-C++ Mailing List=0D
Subject: Re: [Dev-C++] templates in C++ : precisions=0D
 =0D
>>The typename has exactly the same effect. They are both equivalents for
the definition of template arguments.=0D
 =0D
With gcc, they don't. I did some testing: typename compiles, and class do=
esn
t. Perhaps gcc tries to compile something more when it knows the template
parameter is a class...=0D
 =0D
 =0D
 =0D
 =0D
=20
???
You're right. The change from class to typename made my code compile, =
but I continued with coding, and now, when I change my code from =
typename to class, it does compile... I do not know what makes it =
compile now...
Anyway, I will stay with the class keyword. I found it more beautiful :)

nb: there is no B(), this was just a simple sample to illustrate my =
problem... I thought ....

  ----- Original Message -----=20
  From: Ioannis Vranos=20
  To: Dev-C++ Mailing List=20
  Sent: Sunday, January 12, 2003 5:05 PM
  Subject: Re: [Dev-C++] templates in C++ : precisions

        This is non-sense (at least). Did you correct B() as i =
suggested?

        Ioannis Vranos

        * Programming pages: http://www.noicys.freeurl.com
        * Alternative URL: http://run.to/noicys=20

        -------Original Message-------

        From: erwan ancel
        Date: =CA=F5=F1=E9=E1=EA=DE, 12 =C9=E1=ED=EF=F5=DC=F1=E9=EF=F2 =
2003 5:04:32 =EC=EC
        To: Dev-C++ Mailing List
        Subject: Re: [Dev-C++] templates in C++ : precisions

        >>The typename has exactly the same effect. They are both =
equivalents for the definition of template arguments.

        With gcc, they don't. I did some testing: typename compiles, and =
class doesn't. Perhaps gcc tries to compile something more when it knows =
the template parameter is a class...

      =20
              =20
      =20

Hi, do you put all your template code in a header file and then include it
in yout main program ?

//file.h

template <class T>
class A{
    A();
    void B();
    std::set<T> _list;
};

template <class T>
A<T>::A(){
...
}

template <class T>
void A<T>::B(){
...
}

// end file.h

// main.cc
#include "file.h"

int main(void)
{
    A<int> a1;
    a1.B();

    return 0;
}

best regards, Esteban

Yes. Alternatively you can use the keyword export for the definition but
that is not supported yet by all compilers.=0D
 =0D
 =0D
=0D
 =0D
Ioannis Vranos=0D
 =0D
* Programming pages: http://www.noicys.freeurl.com=0D
* Alternative URL: http://run.to/noicys =0D
 =0D
 =0D
-------Original Message-------=0D
 =0D
From: Esteban Valverde=0D
Date: 13 January 2003 15:13:12=0D
To: dev-cpp=0D
Subject: Re: [Dev-C++] templates in C++=0D
 =0D
Hi, do you put all your template code in a header file and then include i=
t=0D
in yout main program ?=0D
 =0D
//file.h=0D
 =0D
template <class T>=0D
class A{=0D
A();=0D
void B();=0D
std::set<T> _list;=0D
};=0D
 =0D
template <class T>=0D
A<T>::A(){=0D
=2E..=0D
}=0D
 =0D
template <class T>=0D
void A<T>::B(){=0D
=2E..=0D
}=0D
 =0D
// end file.h=0D
 =0D
// main.cc=0D
#include "file.h"=0D
 =0D
int main(void)=0D
{=0D
A<int> a1;=0D
a1.B();=0D
 =0D
return 0;=0D
}=0D
 =0D
best regards, Esteban=0D
 =0D
 =0D
-------------------------------------------------------=0D
This SF.NET email is sponsored by: FREE SSL Guide from Thawte=0D
are you planning your Web Server Security? Click here to get a FREE=0D
Thawte SSL guide and find the answers to all your SSL security issues.=0D
http://ads.sourceforge.net/cgi-bin/redirect.pl?thaw0026en=0D
_______________________________________________=0D
Dev-cpp-users mailing list=0D
Dev...@li...=0D
TO UNSUBSCRIBE: http://www.noicys.cjb.net/devcpp/ub.htm=0D
https://lists.sourceforge.net/lists/listinfo/dev-cpp-users=20

Well, this export directive is a very good idea, but which compiler supports
it ? At least, gcc doesn't.

----- Original Message -----
From: "Ioannis Vranos" <no...@ho...>
To: <dev...@li...>
Sent: Monday, January 13, 2003 2:26 PM
Subject: Re: [Dev-C++] templates in C++

Yes. Alternatively you can use the keyword export for the definition but
that is not supported yet by all compilers.

Ioannis Vranos

* Programming pages: http://www.noicys.freeurl.com
* Alternative URL: http://run.to/noicys

-------Original Message-------

From: Esteban Valverde
Date: 13 January 2003 15:13:12
To: dev-cpp
Subject: Re: [Dev-C++] templates in C++

Hi, do you put all your template code in a header file and then include it
in yout main program ?

//file.h

template <class T>
class A{
A();
void B();
std::set<T> _list;
};

template <class T>
A<T>::A(){
...
}

template <class T>
void A<T>::B(){
...
}

// end file.h

// main.cc
#include "file.h"

int main(void)
{
A<int> a1;
a1.B();

return 0;
}

best regards, Esteban

-------------------------------------------------------
This SF.NET email is sponsored by: FREE SSL Guide from Thawte
are you planning your Web Server Security? Click here to get a FREE
Thawte SSL guide and find the answers to all your SSL security issues.
http://ads.sourceforge.net/cgi-bin/redirect.pl?thaw0026en
_______________________________________________
Dev-cpp-users mailing list
Dev...@li...
TO UNSUBSCRIBE: http://www.noicys.cjb.net/devcpp/ub.htm
https://lists.sourceforge.net/lists/listinfo/dev-cpp-users

-------------------------------------------------------
This SF.NET email is sponsored by: FREE  SSL Guide from Thawte
are you planning your Web Server Security? Click here to get a FREE
Thawte SSL guide and find the answers to all your  SSL security issues.
http://ads.sourceforge.net/cgi-bin/redirect.pl?thaw0026en
_______________________________________________
Dev-cpp-users mailing list
Dev...@li...
TO UNSUBSCRIBE: http://www.noicys.cjb.net/devcpp/ub.htm
https://lists.sourceforge.net/lists/listinfo/dev-cpp-users

The export keyword is part of the ANSI/ISO C++ standard. That means that
sooner or later it will be supported by all C++ compilers. The standard a=
lso
guarantees that templates work if you place the same definition of them i=
n
every file you use them (One Definition Rule). That is, the same that you
can do with inline functions. However that also means that if you change =
a
definition, you must recompile all the files having the (new) definition =
in
them.=0D
 =0D
If i recall correctly the Intel C++ compiler supports it (but i am not 10=
0%
sure). Perhaps the latest versions of GCC also support it. One thing is
certain, that soon all compilers will be supporting export.=0D
 =0D
 =0D
Ioannis Vranos=0D
=0D
* Programming pages: http://www.noicys.freeurl.com=0D
* Alternative URL: http://run.to/noicys =0D
 =0D
-------Original Message-------=0D
 =0D
From: erwan ancel=0D
Date: 13 January 2003 15:46:40=0D
To: dev...@li...=0D
Subject: Re: [Dev-C++] templates in C++=0D
 =0D
Well, this export directive is a very good idea, but which compiler suppo=
rts=0D
it ? At least, gcc doesn't.
Ahem. Look how soon Microsoft Visual C++ gets updated to be compatible
with the C++ standard.

Also, while the major players may strive for standard adherence, a lot of
smaller compilers - for smaller operating systems, embedded development
etc - will have a very tough time. For some platforms, there might never
be a compiler with the latest C++ standard.

/Per W

On Mon, 13 Jan 2003, Ioannis Vranos wrote:

> The export keyword is part of the ANSI/ISO C++ standard. That means that
> sooner or later it will be supported by all C++ compilers. The standard also
> guarantees that templates work if you place the same definition of them in
> every file you use them (One Definition Rule). That is, the same that you
> can do with inline functions. However that also means that if you change a
> definition, you must recompile all the files having the (new) definition in
> them.
>
> If i recall correctly the Intel C++ compiler supports it (but i am not 100%
> sure). Perhaps the latest versions of GCC also support it. One thing is
> certain, that soon all compilers will be supporting export.
>
>
> Ioannis Vranos
>
> * Programming pages: http://www.noicys.freeurl.com
> * Alternative URL: http://run.to/noicys
>
> -------Original Message-------
>
> From: erwan ancel
> Date: 13 January 2003 15:46:40
> To: dev...@li...
> Subject: Re: [Dev-C++] templates in C++
>
> Well, this export directive is a very good idea, but which compiler supports
> it ? At least, gcc doesn't.

Well to be accurate, MS has decided to provide new releases of the VS ver=
y
often. The upcoming 14 April 2003 the Visual Studio 2003 will be released
and VC++ will be even more ANSI C++ compliant.=0D
=0D
=0D
For smaller platforms things may be a bit slower, but is there nowadays a=
ny
C compiler that does not support C90? Eventually all compilers will suppo=
rt
C++98.=0D
=0D
=0D
Ioannis Vranos=0D
=0D
* Programming pages: http://www.noicys.freeurl.com=0D
* Alternative URL: http://run.to/noicys =0D
=0D
-------Original Message-------=0D
 =0D
From: Per Westermark=0D
Date: 13 January 2003 16:14:39=0D
To: Ioannis Vranos=0D
Cc: dev...@li...=0D
Subject: Re: [Dev-C++] templates in C++=0D
 =0D
Ahem. Look how soon Microsoft Visual C++ gets updated to be compatible=0D
with the C++ standard.=0D
=0D
Also, while the major players may strive for standard adherence, a lot of=
=0D
smaller compilers - for smaller operating systems, embedded development=0D
etc - will have a very tough time. For some platforms, there might never=0D
be a compiler with the latest C++ standard.=0D
=0D
/Per W
A lot of compilers for the more unusual targets are developed by one or
two persons. For a C compiler, processing the language is trivial. The
only real work is to generate code - especially for some more excentric
processors. Ever tried to generate code for a processor without real
stack, and 8 hardcoded return addresses?

While it's possible to use Lex and Yacc and write a working C compiler as
an examination work at the university, trying to handle the current C++
standard is evilishly hard. Operator overloading, polymorphism, templates,
multiple inherritance etc does take quite a lot of time to get correct.
For a small company supplying the only C++ compiler for a specific target,
there really isn't much incentives to spend man-months or man-years to go
for the last standard, or to handle all of the fine-prints. In the end, if
they don't make more money from selling the product then they have to
spend maintaining it, there really isn't any reason to continue.

/Per W

On Mon, 13 Jan 2003, Ioannis Vranos wrote:

> Well to be accurate, MS has decided to provide new releases of the VS very
> often. The upcoming 14 April 2003 the Visual Studio 2003 will be released
> and VC++ will be even more ANSI C++ compliant.
>
>
> For smaller platforms things may be a bit slower, but is there nowadays any
> C compiler that does not support C90? Eventually all compilers will support
> C++98.
>
>
> Ioannis Vranos
>
> * Programming pages: http://www.noicys.freeurl.com
> * Alternative URL: http://run.to/noicys
>
> -------Original Message-------
>
> From: Per Westermark
> Date: 13 January 2003 16:14:39
> To: Ioannis Vranos
> Cc: dev...@li...
> Subject: Re: [Dev-C++] templates in C++
>
> Ahem. Look how soon Microsoft Visual C++ gets updated to be compatible
> with the C++ standard.
>
> Also, while the major players may strive for standard adherence, a lot of
> smaller compilers - for smaller operating systems, embedded development
> etc - will have a very tough time. For some platforms, there might never
> be a compiler with the latest C++ standard.
>
> /Per W

I haven't played with embedded devices so far, but i assume that a
mainstream embedded system like Pocket PC which has its own Visual Studio
(and the upcoming VS will have integrated the mobile .NET so as to make a=
pps
for the Pocket PC) will have more than 1 companies making compilers for i=
t.
At least for Pocket PC you can expect C++98 compliance at satisfactory
levels. For less mainstream devices, naturally it will take much time to =
see
100% compliance.=0D
 =0D
 =0D
Ioannis Vranos=0D
 =0D
* Programming pages: http://www.noicys.freeurl.com=0D
* Alternative URL: http://run.to/noicys =0D
 =0D
-------Original Message-------=0D
 =0D
From: Per Westermark=0D
Date: 13 January 2003 16:37:41=0D
To: Ioannis Vranos=0D
Cc: Dev-C++ Mailing List=0D
Subject: Re: [Dev-C++] templates in C++=0D
 =0D
A lot of compilers for the more unusual targets are developed by one or=0D
two persons. For a C compiler, processing the language is trivial. The=0D
only real work is to generate code - especially for some more excentric=0D
processors. Ever tried to generate code for a processor without real=0D
stack, and 8 hardcoded return addresses?=0D
 =0D
While it's possible to use Lex and Yacc and write a working C compiler as=
=0D
an examination work at the university, trying to handle the current C++=0D
standard is evilishly hard. Operator overloading, polymorphism, templates=
,=0D
multiple inherritance etc does take quite a lot of time to get correct.=0D
For a small company supplying the only C++ compiler for a specific target=
,=0D
there really isn't much incentives to spend man-months or man-years to go=
=0D
for the last standard, or to handle all of the fine-prints. In the end, i=
f=0D
they don't make more money from selling the product then they have to=0D
spend maintaining it, there really isn't any reason to continue.=0D
 =0D
/Per W=0D
 =0D
 =0D
On Mon, 13 Jan 2003, Ioannis Vranos wrote:=0D
 =0D
> Well to be accurate, MS has decided to provide new releases of the VS v=
ery=0D
> often. The upcoming 14 April 2003 the Visual Studio 2003 will be releas=
ed=0D
> and VC++ will be even more ANSI C++ compliant.=0D
>=0D
>=0D
> For smaller platforms things may be a bit slower, but is there nowadays
any=0D
> C compiler that does not support C90? Eventually all compilers will
support=0D
> C++98.=0D
>=0D
>=0D
> Ioannis Vranos=0D
>=0D
> * Programming pages: http://www.noicys.freeurl.com=0D
> * Alternative URL: http://run.to/noicys=0D
>=0D
> -------Original Message-------=0D
>=0D
> From: Per Westermark=0D
> Date: 13 January 2003 16:14:39=0D
> To: Ioannis Vranos=0D
> Cc: dev...@li...=0D
> Subject: Re: [Dev-C++] templates in C++=0D
>=0D
> Ahem. Look how soon Microsoft Visual C++ gets updated to be compatible=0D
> with the C++ standard.=0D
>=0D
> Also, while the major players may strive for standard adherence, a lot =
of=0D
> smaller compilers - for smaller operating systems, embedded development=
=0D
> etc - will have a very tough time. For some platforms, there might neve=
r=0D
> be a compiler with the latest C++ standard.=0D
>=0D
> /Per W =0D
=20

Thread: [Dev-C++] templates in C++

Open Source C & C++ IDE for Windows

dev-cpp-users