Gerald Pfeifer wrote: > It seems there are some code paths (in the error case?) which may > leave this unitialized. At least the compiler thinks so, and I > could not convince myself otherwise reading this code for a while. >
It's a false positive, probably coming from the compiler not picking up that it is set when the exception variable is false and that when exception is true it is set earlier in the function. > - RpcPktHdr *response; > + RpcPktHdr *response = 0; > > If you're going to change the code at all, then at least keep to the coding convention used in the rest of the function and use NULL here. -- Rob Shearman