From Moo & Koenig, "Accelerated C++:

   1: // source file for the `median' function
   2: #include <algorithm>    // to get the declaration of `sort'
   3: #include <stdexcept>    // to get the declaration of `domain_error'
   4: #include <vector>       // to get the declaration of `vector'
   5:
   6: using std::domain_error; using std::sort;   using std::vector;
   7:
   8: #include "median.h"
   9:
  10: // compute the median of a `vector<double>'
  11: // note that calling this function copies the entire argument `vector'
  12: double median(vector<double> vec)
  13: {
  14:  typedef vector<double>::size_type vec_sz;
  15:
  16:  vec_sz size = vec.size();
  17:  if (size == 0)
  18:         throw domain_error("median of an empty vector");
  19:
  20:  sort(vec.begin(), vec.end());
  21:
  22:  vec_sz mid = size/2;
  23:
  24:  return size % 2 == 0 ? (vec[mid] + vec[mid-1]) / 2 : vec[mid];
  25: }
  26:
  27:

        throw domain_error("median of an empty vector");
                                                      ^
median.cpp(18) : Error: cannot find constructor for class matching
std::domain_error::std::domain_error(char *)
--- errorlevel 1

C:\dm\stlport\stlport\stdexcept, lines 88-94
/*...
class _STLP_CLASS_DECLSPEC domain_error : public logic_error {
public:
  domain_error(const string& __arg) : logic_error(__arg) {}
# ifdef _STLP_OWN_IOSTREAMS
  ~domain_error() _STLP_NOTHROW_INHERENTLY;
# endif
};
...*/

In article <b62vcc$1oqn$1@digitaldaemon.com>, Jim Jennings says...
>
>From Moo & Koenig, "Accelerated C++:

good choice of writers!

>        throw domain_error("median of an empty vector");

>median.cpp(18) : Error: cannot find constructor for class matching
>std::domain_error::std::domain_error(char *)

Add this to your code:

#include <string>

Why? Probably a problem with typedefed, namespace qualified, forward template references involved in conversions during a throw operation ;-)

Richard

"Richard Grant" <fractal@clark.net> wrote in message news:b65csj$gpq$1@digitaldaemon.com...
> In article <b62vcc$1oqn$1@digitaldaemon.com>, Jim Jennings says... Add this to your code:
>
> #include <string>
>
> Why? Probably a problem with typedefed, namespace qualified, forward
template
> references involved in conversions during a throw operation ;-)
>
I'll take your word for it. It works now. Thanks again, Richard. Jim J.

"Jim Jennings" <jwjenn@mindspring.com> wrote in message news:b65hbp$jqa$1@digitaldaemon.com...
>
> "Richard Grant" <fractal@clark.net> wrote in message news:b65csj$gpq$1@digitaldaemon.com...
> > In article <b62vcc$1oqn$1@digitaldaemon.com>, Jim Jennings says... Add this to your code:
> >
> > #include <string>
> >
> > Why? Probably a problem with typedefed, namespace qualified, forward
> template
> > references involved in conversions during a throw operation ;-)
> >
> I'll take your word for it. It works now. Thanks again, Richard. Jim J.
>
>
Oops! I missed the smile.

"Richard Grant" <fractal@clark.net> wrote in message news:b65csj$gpq$1@digitaldaemon.com...
> In article <b62vcc$1oqn$1@digitaldaemon.com>, Jim Jennings says...
> >
> >From Moo & Koenig, "Accelerated C++:
>
> good choice of writers!
>
> >        throw domain_error("median of an empty vector");
>
> >median.cpp(18) : Error: cannot find constructor for class matching
> >std::domain_error::std::domain_error(char *)
>
> Add this to your code:
>
> #include <string>
>
> Why? Probably a problem with typedefed, namespace qualified, forward
template
> references involved in conversions during a throw operation ;-)
>
> Richard
>
>
BTW, I added it to Fomitchev's code where it belongs. ;o) Jim

In article <b65qcc$pnm$1@digitaldaemon.com>, Jim Jennings says...

>BTW, I added it to Fomitchev's code where it belongs. ;o) Jim

Its debatable whether <stdexcept> should include the definition for std::string instead of declarations for std::string. In any case, it might be better to get into the habit of including <string> with <stdexcept> rather than adjusting the <stdexcept> lib. This will make sure nothing breaks if/when you upgrade stlport.

Richard

"Jim Jennings" <jwjenn@mindspring.com> wrote in message news:b65qcc$pnm$1@digitaldaemon.com...
>
> "Richard Grant" <fractal@clark.net> wrote in message news:b65csj$gpq$1@digitaldaemon.com...
> > In article <b62vcc$1oqn$1@digitaldaemon.com>, Jim Jennings says...
> > >
> > >From Moo & Koenig, "Accelerated C++:
> >
> > good choice of writers!
> >
> > >        throw domain_error("median of an empty vector");
> >
> > >median.cpp(18) : Error: cannot find constructor for class matching
> > >std::domain_error::std::domain_error(char *)
> >
> > Add this to your code:
> >
> > #include <string>
> >
> > Why? Probably a problem with typedefed, namespace qualified, forward
> template
> > references involved in conversions during a throw operation ;-)
> >
> > Richard
> >
> >
> BTW, I added it to Fomitchev's code where it belongs. ;o) Jim

Or is it Stepanov's code, and who's who?

"Richard Grant" <fractal@clark.net> wrote in message news:b6758n$1kbo$1@digitaldaemon.com...
> In article <b65qcc$pnm$1@digitaldaemon.com>, Jim Jennings says...
>
> >BTW, I added it to Fomitchev's code where it belongs. ;o) Jim
>
> Its debatable whether <stdexcept> should include the definition for
std::string
> instead of declarations for std::string. In any case, it might be better
to get
> into the habit of including <string> with <stdexcept> rather than
adjusting the
> <stdexcept> lib. This will make sure nothing breaks if/when you upgrade
stlport.
>
> Richard
>
OK, that makes sense. But I still think stlport is missing the #include
<string>.
Jim J.