Closed FeelyChau closed 3 years ago
And I'm -1 on renaming datasource to the dataset, the data source actually represents a data schema for multiple datasets.
And I'm -1 on renaming datasource to the dataset, the data source actually represents a data schema for multiple datasets.
If you mean the datasource type will be changed, I think we should confirm it before 2.0 releasing. Otherwise, if the interface type is exactly equal to Dataset
, I think the name dataset
is better, model script should not care the dataflow
.
Dataset is the concept of datacook, the data source is Pipcook's.
Dataset is the concept of datacook, the data source is Pipcook's.
What's the different except the name? In my opinion, they are the same.
Good question @FeelyChau, and I searched some for it:
In summary, they are not the same thing, and Data Source is a place where we fetch data or dataset, and the Dataset is a set of data, which represents the data itself. Therefore I'm agreed with using "dataset" :)
Pipeline:
Runtime: