This repository publishes the specification of the DataLab workspace capsule format at https://datalab-platform.com/capsule/.
A capsule (.dlcapsule) is a ZIP archive that holds a DataLab HDF5 workspace (workspace.h5) and an RO-Crate manifest (ro-crate-metadata.json). The manifest describes the processing provenance of the workspace data: which operations ran, with which parameters, on which input states, in which software environment, and which of them can be replayed and verified. Capsules can be exchanged between DataLab Desktop and DataLab-Web, and validated or inspected without either application.
| Resource | Identifier | Page |
|---|---|---|
| Profile, version 0.1 (experimental) | https://datalab-platform.com/capsule/0.1 |
0.1/index.html |
Terms (JSON-LD prefix dlc) |
https://datalab-platform.com/capsule/terms# |
terms.html |
These identifiers are written into every capsule manifest. They are permanent: a published version keeps its meaning, a change of meaning gives a new profile version (0.2, …) next to the existing ones, and terms are only ever added.
The profile extends RO-Crate 1.3 and Process Run Crate 0.6.
DataLab-Capsule is the Python library and command line that create, validate and inspect capsules (workspace-capsule validate and workspace-capsule inspect). It runs in CPython 3.9+ and in Pyodide, and needs neither DataLab, NumPy nor h5py to validate or inspect a capsule.
index.html: landing page.0.1/index.html: profile 0.1 (archive layout, manifest requirements, manifest and state fingerprints).terms.html: definition of everydlc:term, one anchor per term.
The site is served by GitHub Pages under the datalab-platform.com domain of the DataLab Platform organisation.
BSD 3-Clause, see LICENSE.