Subj : Re: help on silly code
To   : borland.public.cpp.borlandcpp
From : Leo Siefert
Date : Wed Sep 08 2004 10:17 am

On Wed, 8 Sep 2004 07:57:22 -0400, "Abilio Marques"
<abiliojr@cantv.net.nv> wrote:

>typedef struct {
>    int    sizeX;
>    int    sizeY;
>    double *data;
>} fSet;
>...
>fStet b = {2,2,
>                {3,4,
>                  5,6}};

{3,4,5,6} is not a double*.  It is an int[4].

>typedef struct {
>    int    sizeX;
>    int    sizeY;
>    char *data;
>} fSet;
>
>fStet b = {2,2,"hello"};

"hello" is a const char*. Strictly it should not be allowable unless
the struct contains a const char* member, but many compilers still
allow it a legacy code. This use is deprecated. BC5, however is a
pre-standard compiler and will accept this without comment.

Now, back to your problem.  How would you expect the compiler/the
other program code to know how to process the array you wish to
include in fSet? A char* is terminated by a '\0' byte, so "hello" is
actually a const char[6], and when some function is processing it, the
function can always determine the last character by looking for the
'\0'.

What do you intend to pass as your data? Always 4 doubles? If so you
can use:

double data[4];

otherwise you will need to include a member to indicate the number of
doubles in the array and you will need to new[] and delete[] the
actual array.

 - Leo

.