|
From: Taj M. <tm...@li...> - 2003-07-21 14:47:23
|
Hi Brian, > That's one solution. But I think that when RQ.NET can compile itself, then > >the backends can be separated off into there own DLLs. > Right. >So perhaps we just create an object for each backend: > >CSharpBackend >CppBackend >AsmBackend >PascalBackend > True, but then we have to check every time what the user wanted, and then call the correct backend... It could work, though. >To bad RapidQ didn't have polymorphism or we could just have one 'father' >class that all other backends were derived from and then select the actual backend at runtime. > True. That would be nice. >Something else that may work is function pointers. Store all the functions for a backend in an array, and then select the array on runtime. > You can store functions in an array? That might work a little better, but still be messy... --Taj -- "My software never has bugs, it just develops random features." |