RE: [avr-gcc-list] Typdef chick 'n egg situation.. way out ?

[Schwichtenberg, Knut]

Larry,

I'm not sure if I missed anything in the discussion.... I had the
problem in a linked list and my project compiled with avr-gcc 3.4.
flawlessly.
Here is my solution:

struct tmLokCmdTab;

typedef struct tmLokCmdTab{
        struct tmLokCmdTab * ptrnext;
....
        };
}t_mLokCmdTab;
 
I think the topic is called type forwarding and is at least identical to
function protoypes in headers.

Cheers
Knut

> -----Original Message-----
> From: 
> address@hidden
> g 
> [mailto:address@hidden
ongnu.org] On Behalf Of larry barello
> Sent: Thursday, January 22, 2009 12:42 AM
> To: 'Vincent Trouilliez'; address@hidden
> Subject: RE: [avr-gcc-list] Typdef chick 'n egg situation.. way out ?
> 
> You can't make a "flexible" typdef.  If you did that the 
> compiler would have
> multiple definitions for various instances of the typedef and 
> get horribly
> confused.
> 
> I have worked around this in several ways:
> 
> 1. make the typedef array have a size of one.
>    Allocate some memory to contain the array and assign a 
> pointer of type to
> it.
> 2. Use a #define to create a unique typedef that maps to your 
> basic type but
> 
>    declares the proper size array. 
> 
> Both schemes have issues.  If there is a better way I would 
> like to know
> about it.
> I presume C++ handles all this with grace, but I don't really know c++
> 
> Finally, I bet there are better forums for this kind of 
> discussion as this
> isn't very particular to avr-gcc!
>