ulyses said:
I'm trying to put pointer to flexible array of structures in other
structure. I want to have pointer to array of pixels in screen
structure. Here is mine code, but I think it isn't quite all right:
struct pixel
{
int x;
int y;
int color;
};
struct display
{
pixel *screen;
};
There's a problem here. You've defined a type called "struct pixel";
there is no type called "pixel". This should be
struct pixel *screen;
<OT>
In C++ the type "struct pixel" can be referred to as "pixel", but if
you were programming in C++ you'd have posted to comp.lang.c++,
wouldn't you?
Now I would like to do something like that (to have screen made of 10
pixels):
struct display scr;
scr.screen = (struct pixel*)malloc(10*(sizeof(struct pixel)));
scr.screen[0].x = 10;
and it is working,
So far, so good, except that the cast on the malloc() is unnecessary
and can mask errors such as forgetting to "#include <stdlib.h>" or
compiling your C code with a C++ compiler. If you get an error
message without the cast, you need to fix the error, not hide it
with a cast that says to the compiler, "Shut up, I know what I'm doing".
The recommended style is
str.screen = malloc(10 * *scr.screen);
By not referring to the type "struct pixel" in the malloc() call, you
don't need to change the call if the type of scr.screen changes later
on.
but what is worrying also
scr.screen[11].x = 10;
or
scr.screen[100].x = 10;
is working. I don't now what is wrong. I'm not sure but I think I'm
making mistake when defining structures. How should I define flexible
array of structures in structure? And how should I point it?
No, "scr.screen[11].x = 10;" is not "working". It's clobbering memory
beyond the end of your array, which invokes undefined behavior. That
means anything can happen, including having it appear to "work".
C doesn't allow you to index beyond the end of an array object, but it
doesn't do anything to prevent you from doing so. If you index past
the end of an array anyway, the compiler isn't obliged to do anything
about it. It's entirely your responsibility to stay within the bounds
of your object. This allows better performance when you don't make
mistakes, at the expense of arbitrarily bad behavior when you do make
mistakes.
You should probably have a second member in your "struct display" that
keeps track of how many pixels you've allocated. For example:
struct pixel
{
int x;
int y;
int color;
};
struct display
{
struct pixel *screen;
size_t pixel_count;
};
struct display scr;
scr.screen = malloc(10 * sizeof *scr.screen);
if (scr.screen == NULL) {
scr.pixel_count = 0; /* malloc failed */
}
else {
scr.pixel_count = 10;
}
The allocation and initialization should probably be encapsulated in a
function that takes the number of pixels as an argument. You might
want more elaborate error handling as well.
What you have so far is a decent starting point, but there are a few
things you should think about as you develop it further.
I find the use of the terms "screen" and "display" confusing. They
tend to be nearly synonymous in my mind, but you're not using them
consistently as far as I can tell. "display" is the name of the
structure, "screen" is the name of the member of that structure that
points to the array of pixels, but "scr", an abbreviation of "screen",
is also the name of an object of type 'struct display". Pick a
consistent naming scheme now, before your program becomes too
complicated to go back and fix.
Also, a screen/display is more than a linear array of pixels. You're
going to want to keep track of the horizontal and vertical dimensions,
i.e., you'll need a dynamic 2-dimensional array. Section 6 of the
comp.lang.c FAQ has some good tips on this.