fun with structures

[Kennedy]

I have two structures:

struct something {
        char id;
        char data[100];
};

struct somethingelse {
        char name;
        char data[100];
};

Without using a union, could I make these two structures as one? I had even thought to make it something like

struct something {
        char id_or_name;
        char data[100];
};

but that is so ugly. I haven't tried it (but will), but would this be valid (and the best way to do this):

#define name id
struct something {
        char id;
        char data[100];
};

struct something stuff;
struct something name;

stuff.id = 3;
name.name = 'j';

[bithub]

While using that define is legal, it will probably cause you problems elsewhere in the code (some variable or function will have "name" in it somewhere). It also hurts code readability.

What is the reasoning for combining the two structures? Is it so you can have one function which takes both types?

[Kennedy]

What is the reasoning for combining the two structures? Is it so you can have one function which takes both types?

Readability and maintainability. The trick is that I will only have one structure (it is not as simple as char name and data, it has MANY more parts to it). So, in the effort to abstract the data, the first byte of the structure becomes problematic for me.

One who is more knowledgeable than I suggested that I just make the first byte descriptor more descriptive and specific so that I'll not stomp any other names in the name space. To that end, I'll make my identifier long and mixed case which will make this puppy special (I don't use mixed case identifiers usually -- That takes too long).

[twomers]

Yeah. I don't get why you'd want that either... just come up with a common name for name/id... It compiles fine for me, anyway (VC '03), though intellisense (while not infallable) doesn't catch the name member of the name variable. Just calls it id.
Just call it tag and live with it!

(Posted afterI read bithub's response, and before Kennedy's second)

[MK27]

Readability and maintainability.

Hmmm. But in what sense is your conception here superior, "readability and maintainability"-wise, compared to the first example in the OP:

struct something {
        char id;
        char data[100];
};

struct somethingelse {
        char name;
        char data[100];
};

Especially since these have two distinct names -- or do you intent to #define aliases there too?

[Sebastiani]

What's wrong with a union?

struct something {
        union {
		char id;
		char name;
	};	
        char data[100];
};

int main(void) {
	struct something x;
	x.id = 'x';
	x.name = 'y';
	return 0;
}

[Kennedy]

What's wrong with a union?

I don't want to add in another . to my code. . . but then, I didn't realize that it would be acceptable to have no name, thus no additional . in the code. That's really cool and that is JUST what I was wanting.

[slingerland3g]

A union would be the way to go, but unless I mis understood the logic by what Sebastiani showed, if you redefine the x union your x.name will overwrite x.id data.

[Kennedy]

A union would be the way to go, but unless I mis understood the logic by what Sebastiani showed, if you redefine the x union your x.name will overwrite x.id data.

Exactly! That is the whole point. The "real" world use would be something like this:

struct data {
        union {
                uint8_t id;
                uint8_t name;
        };
        uint32_t some, other, stuff;
        uint16_t etc, etc1, etc2;
};

int main(void)
{
        struct data type0, type1;
        type0.id = 42;
        type1.name = 'a';
        <more>
}