Random thoughs about data source integration in envd (and envd-server)

[gaocegege]

Option 1: The build.envd way

def volume():
    data = s3.dataset("bucket").create_if_not_existed()
    # Or use existed dataset created in the dashboard.
    data = dataset(name="")
    return data

def build():
    runtime.mount("/data", data)

Option 2: The CLI way

envd data create --name imagenet ./imagenet_folder/ 

Then use it in build.envd:

def volume():
    data = dataset(name="")
    return data

def build():
    runtime.mount("/data", data)