Issue 168 #199
Issue 168 #199
Conversation
|
I took the liberty of resolving the conflict in |
|
|
||
| def memory(self): | ||
| pass | ||
|
|
||
| def network(self): | ||
| pass | ||
| """Build a network of the relationships between | ||
| data as specified by the container's metadata. |
There was a problem hiding this comment.
I think the relationships are being built by the data itself (assumes that the data is DataFrame or Series? object and constructs based on series names).
There was a problem hiding this comment.
For the network I was hoping that it would be possible to generate some (if not all) of the relationships based solely on the Data's metadata, rather than having to descend to the underlying data and inspect it directly. I see this approach not working great where multiple index-column relations exist, e.g. atom -> atom0, atom1.
| API should live in and be controlled by the Container? | ||
| Here can serve as a pass-through to parent if it exists. | ||
| Should it need to be present in data? | ||
| """ |
There was a problem hiding this comment.
I think this is a really good idea. When a data object is attached to a container, does the data object have a reference to its container?
Opening this to throw some ideas around about possible implementations for data <-> container functionality.