www.digitalmars.com/compiler.htm

Wow!

*    -Ab and -Aw are now always on (meaning that bool and
wchar_t are real types, not typedefs). Warning: This can break
existing code.
*    -Aa is now always on (meaning that operator new[] and
operator delete[] are separately overloadable) Warning: This can
break existing code.
*    parameter-declarations now allowed in function template
template-parameter-list.
*    typename can now appear in template-parameter instead of
class.
*    Explicit template function template-arguments implemented.
*    Partial ordering of function templates implemented.
*    elaborated-type-specifier (i.e. typename) implemented.
*    mutable implemented.
*    explicit implemented.



Walter wrote:

> www.digitalmars.com/compiler.htm

It still is not C++98 compliant with the templates, it's just a step in that direction.

Jan Knepper wrote in message <3BAE385F.A1AD58A0@smartsoft.cc>...
>Wow!
>
>*    -Ab and -Aw are now always on (meaning that bool and
>wchar_t are real types, not typedefs). Warning: This can break
>existing code.
>*    -Aa is now always on (meaning that operator new[] and
>operator delete[] are separately overloadable) Warning: This can
>break existing code.
>*    parameter-declarations now allowed in function template
>template-parameter-list.
>*    typename can now appear in template-parameter instead of
>class.
>*    Explicit template function template-arguments implemented.
>*    Partial ordering of function templates implemented.
>*    elaborated-type-specifier (i.e. typename) implemented.
>*    mutable implemented.
>*    explicit implemented.
>
>
>
>Walter wrote:
>
>> www.digitalmars.com/compiler.htm
>

On Sun, 23 Sep 2001 14:20:39 -0700, Walter wrote in c++.announce:
> It still is not C++98 compliant with the templates, it's just a step in that direction.

Does that mean that you are still working on improving the C++ compiler?


BTW, here are two code snippets which I think are valid C++ but don't compile with Digital Mars (just in case you want to fix some bugs/implement some more C++ features):


namespace my_ns
{
  class A
  {
  };
};

class B
{
  friend class my_ns::A;
  // Error: only classes and functions can be friends
};




class A
{
  static const int size = 10;
  // Error: initializer for static member must be outside of class def
  char buf[size];
};


bye, Christof

-- 
http://cmeerw.cjb.net                          Jabber: cmeerw@jabber.at mailto cmeerw at web.de                   ICQ: 93773535, Yahoo!: cmeerw

...and what have you contributed to the Net?

Yes, I still work all the time on improving the C and C++ compilers, as well as on D. I want to bring C up to C99 compliance and C++ up to C++98.

Thanks for the bug reports.

-Walter


Christof Meerwald wrote in message <9on6qs$1uog$1@digitaldaemon.com>...
>On Sun, 23 Sep 2001 14:20:39 -0700, Walter wrote in c++.announce:
>> It still is not C++98 compliant with the templates, it's just a step in
that
>> direction.
>
>Does that mean that you are still working on improving the C++ compiler?
>
>
>BTW, here are two code snippets which I think are valid C++ but don't compile with Digital Mars (just in case you want to fix some bugs/implement some more C++ features):
>
>
>namespace my_ns
>{
>  class A
>  {
>  };
>};
>
>class B
>{
>  friend class my_ns::A;
>  // Error: only classes and functions can be friends
>};
>
>
>
>
>class A
>{
>  static const int size = 10;
>  // Error: initializer for static member must be outside of class def
>  char buf[size];
>};
>
>
>bye, Christof
>
>--
>http://cmeerw.cjb.net                          Jabber: cmeerw@jabber.at mailto cmeerw at web.de                   ICQ: 93773535, Yahoo!: cmeerw
>
>...and what have you contributed to the Net?

On Mon, 24 Sep 2001 10:30:54 -0700, Walter wrote:
> Yes, I still work all the time on improving the C and C++ compilers, as well as on D. I want to bring C up to C99 compliance and C++ up to C++98.

That's great - I was a bit afraid that you were only working on D now.


> Thanks for the bug reports.

Here is another one:

namespace my_ns
{
  enum bla
  {
    BLABLA_1, BLABLA_2
  };
}

bla oops;
// this shouldn't work, but the compiler doesn't complain

my_ns::bla blabla;
// Error: 'bla' is not a member of namespace 'my_ns'


bye, Christof

-- 
http://cmeerw.cjb.net                          Jabber: cmeerw@jabber.at mailto cmeerw at web.de                   ICQ: 93773535, Yahoo!: cmeerw

...and what have you contributed to the Net?

Working on the C++ compiler is really admirable. Please KUTGW.
Thanks!!
-- Rajiv


Christof Meerwald <NOSPAM_seeMySig@fastrun.at> wrote in message news:9oo749$2gkn$1@digitaldaemon.com...
> On Mon, 24 Sep 2001 10:30:54 -0700, Walter wrote:
> > Yes, I still work all the time on improving the C and C++ compilers, as
well
> > as on D. I want to bring C up to C99 compliance and C++ up to C++98.
>
> That's great - I was a bit afraid that you were only working on D now.
>
>
> > Thanks for the bug reports.
>
> Here is another one:
>
> namespace my_ns
> {
>   enum bla
>   {
>     BLABLA_1, BLABLA_2
>   };
> }
>
> bla oops;
> // this shouldn't work, but the compiler doesn't complain
>
> my_ns::bla blabla;
> // Error: 'bla' is not a member of namespace 'my_ns'
>
>
> bye, Christof
>
> --
> http://cmeerw.cjb.net                          Jabber: cmeerw@jabber.at mailto cmeerw at web.de                   ICQ: 93773535, Yahoo!: cmeerw
>
> ...and what have you contributed to the Net?

On Mon, 24 Sep 2001 10:30:54 -0700, Walter wrote:
> Thanks for the bug reports.

Here are two more:

namespace my_ns
{
  class A
  { };

  template<class T> class B
    : A
  // Error: undefined identifier 'A'
  { };
}

my_ns::B<int> b;


and


struct C
{
  int operator->();
  // Error: illegal return type for operator->()
};

AFAIK, this restriction doesn't exist any more in the C++ standard.


bye, Christof

-- 
http://cmeerw.cjb.net                          Jabber: cmeerw@jabber.at mailto cmeerw at web.de                   ICQ: 93773535, Yahoo!: cmeerw

...and what have you contributed to the Net?

On Mon, 24 Sep 2001 10:30:54 -0700, Walter wrote:
> Thanks for the bug reports.

Here is another one:

enum A
{ A_VALUE };

bool operator ==(A a, A b)
// Error: at least one parameter must be a class or a class&
{ return false; }


But the C++ standard says (13.5.6):

| An operator function shall either be a non-static member function or be a | non-member function and have at least one parameter whose type is a class, | a reference to a class, an enumeration, or a reference to an enumeration.


bye, Christof

-- 
http://cmeerw.cjb.net                          Jabber: cmeerw@jabber.at mailto cmeerw at web.de                   ICQ: 93773535, Yahoo!: cmeerw

...and what have you contributed to the Net?

Here's a good library for testing the template code. http://www.boost.org/libs/python/doc/index.html

The author David Abrahams sits on the C++ standards committee, or used to.

From a Python development standpoint his library is the absolute best way to create Python extensions.  It is also a heavily template-laden body of code that breaks a number of compilers.

Mark


On Sun, 23 Sep 2001 14:20:39 -0700, "Walter" <walter@digitalmars.com> wrote:
> It still is not C++98 compliant with the templates, it's just a step in that direction.
> 
> Jan Knepper wrote in message <3BAE385F.A1AD58A0@smartsoft.cc>...
> >Wow!
> >
> >*    -Ab and -Aw are now always on (meaning that bool and
> >wchar_t are real types, not typedefs). Warning: This can break
> >existing code.
> >*    -Aa is now always on (meaning that operator new[] and
> >operator delete[] are separately overloadable) Warning: This can
> >break existing code.
> >*    parameter-declarations now allowed in function template
> >template-parameter-list.
> >*    typename can now appear in template-parameter instead of
> >class.
> >*    Explicit template function template-arguments implemented.
> >*    Partial ordering of function templates implemented.
> >*    elaborated-type-specifier (i.e. typename) implemented.
> >*    mutable implemented.
> >*    explicit implemented.
> >
> >
> >
> >Walter wrote:
> >
> >> www.digitalmars.com/compiler.htm
> >
> 
>