Hi,
I just discussed with Michael how the calling interface for functions should be. So far, we have settled on a model where each function returns an integer error code, for which some constants are defined. There is one OK state and a number of error states. The caller must check the returned result and decide whether or not to abort its own operation on unexpected return states.
So far, we have decided not to use exceptions, one reason being that we are a bit unsure about proper cleanup.
But it remains the question if exceptions would be a better way to go, if they are used together with some return state signaling.
Please let's discuss this topic here, it is an important one before we can start developing actual code.
Thanks,
Rainer


