Thank you for your email. Flexible arrays are indeed legal. http://gcc.gnu.org/onlinedocs/gcc-3.4.3/gcc/Zero-Length.html#Zero-Length (and it really is not a hack or a bad trick :), it has been used for ages and I would be hard-pressed to come up with an equally elegant solution to represent variable size arrays ) The trick of using the max array size is not workable in embedded applications where memory is at a premium. And then there is the static intialization problem. e.g typedef struct { } a_t; I thought C++ is a superset of C and C code should compile in a C++ compiler (except for strict type checking) ..but guess not. N. --- Eljay Love-Jensen <eljay@xxxxxxxxx> wrote: > Hi Nitin, > > >Static initializations of a flexible array works > with gcc but not with > g++ . is this by design? > > Yes, this is by design. C++ does not have flexible > arrays (or what I've > heard called "stretchy arrays" or "stretchy > buffers"). C++ has STL > std::vector. > > The stretchy buffer is a bad trick -- it's not > portable, and may cause > certain optimizations to fail is a bad way. > > (I'm not sure if it is even legit C code. Maybe it > is with C99. I dunno.) > > A suggestion is to do the reverse: specify the > structure with the array > given the maximum length, and allow allocations of > less-than-maximum when > used off the heap. (There are caveats with this > approach as well.) > > struct flex_array_1000 > { > int A; > int Data[1000]; > }; > > Alternatively, you can use template structs to > provide more exacting > "hard-coded" arrays. > > template <int Count> > struct flex_array > { > int A; > int Data[Count]; > }; > > ...and you can have the struct derived from a common > base class if you need > some sort of polymorphism. > > HTH, > --Eljay > > __________________________________ Do you Yahoo!? Take Yahoo! Mail with you! Get it on your mobile phone. http://mobile.yahoo.com/maildemo
