In general, this needs a more formal specification.
The column labels are fixed by the labels used by Brightway.
-
Version
-
License
-
Author
-
Dependencies
-
URI
-
List of strategies (plus libraries with strategies? Notebooks/python files?)
In metadata or following Futura recipe YAML?
- Brightway library versions
- Other optional libraries
Brightway can offer storage services for easy download through support from Départ de Sentier, but having something like Zenodo would be preferable.
Question: Do we need a "lite" storage to see if data is already present (to avoid unnecessary downloads)?
Question: Do we need to invent a unique identifier for LCI databases? Or make up a URI scheme? Is Zenodo download link a good URI?
Question: Single centralized library repo versus decentralized approach with search-engine-based discovery? Interaction with existing efforts, e.g. GLAD, OpenLCA NEXUS?
Question: How to handle links to proprietary dependencies?
Question: Security of external strategies. Maybe have one "blessed" repo with external strategies?
Question: Can we issue DOIs for data? Free versus paid?
Question: Do we need "known good result" tests?
Add functionality to import this data, and to download dependent databases
Need functionality to see if and when datapackages conflict
Headers are a mix of fixed values and scenario labels.
Proposal: Fix these values and number of columns as currently used.
Question: Is flow type reqiured and guaranteed to be correct?
Scenario labels don't seem to have a naming convention. That's probably fine? But the metadata needs to have more information.
- Version
- License
- Author
- Dependencies
- URI
I don't think we need custom strategies?
Same questions apply. In general, these files are larger, but not necessarily large. Chris will clarify if his institution can provide hosting - it would be good to have mirrors at other institutions.