Contributing

The registry is maintained as open source under the untded organization, on the mandate of ISO/TC 154’s plenary resolutions (see about). Contributions follow a lightweight data-maintenance process — no toolchain needed to propose a change.

Raise a request

Two GitHub issue templates, on the dataset repository:

• Correction request — you believe a cell is wrong (a name, a definition, a representation, a bridge). Give the tag and what the source prints. Open a correction request.
• New edition — you want a later edition of the directory (or a related publication) digitized. Say which and why. Open an edition request.

What happens next

1. The maintainers re-verify the claim against the source publication — every correction must be grounded in the printed text, never in another digital edition (see the provenance note).
2. Corrections that verify become a change to the dataset’s extraction or data through a pull request; the verification suite (regeneration, character conservation, UID reconciliation) must stay green.
3. The site and the machine-readable artifacts regenerate from the dataset — nothing is edited by hand downstream.

Code

The website and the tooling accept pull requests. Small, focused changes with tests are the fastest to land; the repository READMEs describe the layout. Everything ships through review — no direct commits to main.