IMO, we need an abstract base class that defines the interface for messages containing signals and features that will be passed through signal processing, feature extraction, and decoding nodes. AxisArray will be an implementation of this class.
Then our nodes (mostly in ezmsg-sigproc) can be refactored to use class methods as much as possible.
Then we can add new classes. e.g., one where data is sparse, or where data is a handle to GPU memory.
IMO, we need an abstract base class that defines the interface for messages containing signals and features that will be passed through signal processing, feature extraction, and decoding nodes.
AxisArray
will be an implementation of this class.Then our nodes (mostly in ezmsg-sigproc) can be refactored to use class methods as much as possible.
Then we can add new classes. e.g., one where
data
is sparse, or wheredata
is a handle to GPU memory.