guruofquality
commented
10 years ago A simpler version of this would work with the existing notion of static types (set once at activation time), but allow for dynamically determined input port types.... So, the output port type can be placed into the input port by the topology so the user can query the input type at runtime. We will probably have an option to let the user explicitiy set the input port to automatic mode -- absorb the type from the output port of the connection.
Incoming buffer chunks are dynamic from run to run based on the dtype associated with them. The output port type is defined by the dtype that the user posts to the output.