An LCA Data Package Manager (LDPM)

Highlights

• Open-source LCA data is increasingly being published in smaller collections than whole background DBs
• The classic software-package-manager pattern might be usefully adapted to LCA data
• Prototyping of an LCA Data Package Manager (LDPM) was initiated to support knitting U.S. FLCAC data into a unified DB

Description

The U.S. Federal LCA Commons (FLCAC) is a federation of data packages (DPKG) independently developed and maintained by various U.S. federal agencies. This separation ensures clear ownership and attribution but also partially constrains the development of a single, combined database (DB). Data object links (i.e., a "Ref" in the olca-schema) across DPKGs are not well supported by the FLCAC's backend LCA Collaboration Server instance, and even if they were, independent DPKG releases pose a series of pair-wise version control dilemmas. These inter-DPKG links have to-date been managed via bridge processes and libraries, with the former requiring users to manually update Exchange.defaultProvider selections.

To overcome these challenges and begin developing an all-in-one FLCAC DB, ERG began prototyping an LCA Data Package Manager (LDPM), with design principles and manifest format inspired by Rust's Cargo package manager. The LDPM should be fast and deterministic: it must yield a reproducible output from a given manifest and set of additional configuration files. Users should be able to easily read, write, reuse, and extend manifests, and the resulting background DB easily rebuilt on any machine. Additionally, the latest LDPM prototype requires configuration files containing instructions for two main features: data object de-duplication and inter-DPKG object linking. The de-duplication routine amends previously necessary patterns on the FLCAC; later on, this could be condensed into a collection of tests. Linking, however, is a core feature, as it enables cross-DPKG Exchange.defaultProvider and Flow selections. Rather than trying to nest Ref links in a referent object, we index links by pairs of versioned DPKG identifiers (e.g., [DPKG_A-version, DPKG_B-version]) as top-level keys. Maintaining DPKGs and link sets separately offers certain version control and dependency resolution advantages.

Broadening our horizons further, the steady emergence and improvement of national LCI DBs writ large means that an LDPM-like tool could offer a common LCA-stack component and language (via the manifest format) to define and mint background DBs that encompass global supply chains. For example, an LDPM-assembled background DB could serve as the foundation upon which experts build a PEF or EPD template. Those experts, relevant regulators, and regulated parties alike could then each propose refinements to that template's background DB by way of changes to the LDPM manifest and/or configuration files. Much like the FLCAC, national LCI DBs and other open-source industrial ecology datasets can be collectively viewed and managed as a federation of DPKGs, all waiting to be reproducibly woven together by a dedicated "data loom"—the LDPM.