LoginSign Up
← Back to all sparks
D

datasetjson

ANALYTICS
Velocity0.0

Read and write CDISC Dataset JSON files. Contribute to atorus-research/datasetjson development by creating an account on GitHub.

github.com/atorus-research/datasetjson ↗

datasetjson rebuilt its object model to track the CDISC Dataset-JSON 1.1 schema.

clinical-datacdiscjsonr-packagepharmaverse
◆Current state
datasetjson reads and writes CDISC Dataset-JSON, the JSON replacement for SAS transport files in clinical-trial submissions. The package went from a thin reader in 2023 to a redesigned interface in 0.3.0 that targets the 1.1.0 schema, uses yyjsonr as its JSON backend, and exposes column metadata as first-class arguments. Development is contributor-driven inside the Atorus and pharmaverse orbit.
◆Where it's heading
The package's roadmap is not its own — it tracks a CDISC standard that is still moving, and 0.3.0 is what happens when the standard revises: object model, read and write paths, and JSON backend all changed together. Performance was addressed in the same pass, which matters because submission datasets are large enough that a slow serialiser is a real constraint.
◆Prediction
The next significant release will most likely follow the next Dataset-JSON schema revision rather than an internal roadmap, given that 0.3.0 was driven entirely by the 1.1.0 update.

◆Recent moves

  1. 1y ago

    Dataset-JSON 1.1.0 support with a redesigned object model

    ⚡ SPARK

    The release that resets the package: a new schema version, redesigned core objects, and a different JSON backend, all at once. Everything the package did before 0.3.0 was built against a superseded standard.

    View source ↗
  2. 2y ago

    Reads and validates Dataset-JSON from URLs

    Remote reading plus several defaults corrected — no auto-populating optional attributes with NA, no inferring fileOID from the output path. Small changes that stop the package from inventing metadata the user did not supply.

    View source ↗
  3. 2y ago

    Initial CRAN release

    The package becomes installable from CRAN, built against the schema version current at the time. The 1.1.0 overhaul two releases later shows how short that shelf life was.

    View source ↗