Skip to content

merge.yaml now requires credential-gated lpsn_api output (breaks credential-less merges) #600

Description

@realmarcin

Follow-up from PR #598, which wires lpsn (GSS base) and lpsn_api (JSON-API enrichment) into merge.yaml.

Problem

The lpsn_api merge source lists data/transformed/lpsn_api/{nodes,edges}.tsv as required inputs. Those files can only be produced by poetry run kg transform -s lpsn_api, which raises RuntimeError without LPSN_USERNAME/LPSN_PASSWORD (and network access to the DSMZ API). So any environment that runs kg merge -y merge.yaml without LPSN credentials — a fresh checkout, CI, or a contributor without a DSMZ account — now fails on missing input files. Before this PR, LPSN was absent from the merge, so a credential-less merge succeeded.

The lpsn base source (GSS/CSV, no creds) is comparatively safe; lpsn_api is the credential-gated one.

Options

  1. Document that merge.yaml now presupposes an authenticated lpsn_api transform run (README / merge docs), and keep it as-is (acceptable if the canonical build always has LPSN creds).
  2. Make the lpsn_api source optional — e.g. a merge.nolpsn.yaml variant, or skip-if-missing semantics, so credential-less merges still succeed.
  3. Split the base lpsn (always merged) from lpsn_api (opt-in).

Licensing note

LPSN data is CC BY-SA 4.0. Wiring it into the merge was an explicit maintainer decision; this issue is only about the operational credential coupling, not the licensing choice.

Metadata

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels

    Type

    No type

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions