We have a very large schema (hundreds of global elements, thousands of global complex types). When we use the Liquid Studio Data Binding tool to generate C++ data bindings from the schema, the resulting project exceeds the DLL export symbol limit of 64k symbols (65535). In fact, it exceeds the limit so thoroughly that we've had to split the schema into 6 different schemas (with some duplicate complex types) just to break it down enough so that each project stays under the symbol export limit.
Is there a better option? Is there something I can configure in Liquid Studio so that it doesn't try to export so many symbols? I've used another tool on the exact same schema and the resulting code doesn't have this problem because it generates mostly abstract interfaces with factory classes to create the desired objects. Thus the only symbols that need to be exported are the factory create methods. Right now we're kind of stuck with the LiquidXML data bindings, so if we can avoid all of the manual work of splitting the schema up each time there's an update it would really be nice.
In the XML Data Binder Wizard you can select the root elements you are interested in, the generator will then ignore the rest. This can be very effective, however, depending on the structure of the schema you may find that much of the schema is still potentially accessible via a small number of root elements.
The resulting .xdbproj file can then be re-run from the command line e.g. in an automated build.
If this approach does not work for you, please post a link to the schema and the root elements you are interested in and I'll look into it in more detail.