I've been spending some quality time with the memory management code recently, albiet in an older gSoap version, but I can't answer your question definitively for all cases.
The standard approach is to use soap_malloc() and the soap_new_* methods to allocate data. This leverages the built-in garbage collection system. I suppose one could implement his own system to track allocations. In you example, calling soap_end() after each could cause problems depending on the content of ns1_FunctionResponse1. If that's an atomic class or structure than the applications code will have complete control over the allocation. If that contains data that needed to be allocated inside the call, the invoking soap_end() may cause problems. It looks like multiple calls into the generated methods may cause allocated pointers to be retained in the linked lists, but I don't know for sure. If so then you could do a soap_end() after the pair. Give it a whirl! --- In gsoap@yahoogroups.com, "partridge.brian" <[EMAIL PROTECTED]> wrote: > > So just to confirm, in the following case: > > // global soap struct maintaining a connection > > __ns1__Function1(ns1__Function1, ns1__FunctionResponse1); > // Create an object that maintains a reference to the response > MyClass *pObj = new MyClass(ns1__FunctionResponse1); > __ns1__Function2(ns1__Function2, ns1__FunctionResponse2); > return pObj; > > In this case I should call soap_end after each functions 1 and 2 and > it wont invalidate anything that I need, right? How should I cleanup > response 2 before it goes out of scope at the return, soap_delete? > What about response 1? Since pObj maintains a reference to response 1 > and is returned I think I should use soap_delete within ~MyClass. > > Thanks for the help. >
