Did you think of optimizing the process of parsing and serializing and XML response?
I am working currently on a project, where we receive a huge XML response, and there is a lot of old code, I thought to be refactored. So I tried to use RestKit against the current parser and serializer.
In the images from below, you can see the result of the experiment realized in Allocations instrument. Result of our solution also includes CRUD with the DB, while the RestKit version only receive response and objects mapping.
Our current solution works based on the libxml2 library and streaming parsing method. It means it is parsing while downloading and it is extremely fast and low memory consuming.
Will be taking a look at streaming support and revising XML during the 1.0 development cycle.