Dataset

A Dataset is a linked set of IPFS DAG entries that are used to store results of tasks, input, or task status.

There are two ways to reference a dataset in a task.

Explicit IPFS reference

Linking directly to the IPFS CID of a dataset will ensure computes uses the correct dataset. This reference is immutable and cannot change over time.

{
  "dataset": {
    "/": "zdpuB1cyssBgbsexpu2LcvCHe2k6FTTqwQAXCj7LsaGErPqvK"
  }
}

You may use the computes-cli to generate a new dataset from a unique value

computes-cli dataset create --generate-uuid
> zdpuB1cyssBgbsexpu2LcvCHe2k6FTTqwQAXCj7LsaGErPqvK

Initial Seed Value

Sometimes using the explicit IPFS reference is inconvenient. In this case, you may use a seed value which will be used to calculate the IPFS reference.

{
  "dataset": {
    "init": "some-value"
  }
}

The results should be the same as running the computes-cli command

computes-cli dataset create "some-value"
> zdpuArwKMPCexEap3fWXGS3DE2zmF5iRNG6vXaYhNAuCq6ecm