From: Arno P. <ar...@pu...> - 2011-09-18 08:47:10
|
the reason we opted for not including return types is to reduce the number of the dependent classes. If you ignore the return value of a method, there is no need to generate an #include. As you pointed out, the moment you use the type for something, an #include will be generated. As you also pointed out, this is a bit of a problem for classes annotated with @XMLVMSkeletonOnly since we don't know what types the injected code requires. As a 'solution' we usually add a protected dummyMethod() whose parameter types are the ones we need in the injected code. It might be a good idea to add return types to the dependency analysis and see by how much that increases the number of classes. If the increase is not dramatic, perhaps we should do this instead of the dummyMethod() workaround. I also know that this is an issue with the C# backend for WP7. Arno On 9/18/11 1:15 AM, Markus Heberling wrote: > Hi, > > for classes that appear in method parameters, an include is generated in the c-file. This is not done for classes that occur as method return types. This is no problem for normal classes, since the "missing" class will be added as an dependency as soon as it is instantiated with new or something else. But for classes annotated with @XMLVMSkeletonOnly those includes are missing. You can see that for example in the CGContext class. CGPoint is used as a return type there, but its missing as an include in the generated file. > > I think those return type dependencies should be added to the include list. > > Markus > ------------------------------------------------------------------------------ > BlackBerry® DevCon Americas, Oct. 18-20, San Francisco, CA > http://p.sf.net/sfu/rim-devcon-copy2 > _______________________________________________ > Xmlvm-developers mailing list > Xml...@li... > https://lists.sourceforge.net/lists/listinfo/xmlvm-developers |