We don't need to, or even necessarily want to, store all CFs in the database. Instead, taking inspiration from the read-only proxies in the IO table backend, we can interact directly with the processed arrays of characterization factors. This has several advantages:
Very fast import of methods when setting up projects
Very fast import of methods provided in the lciafmt specification
Don't need to figure out how to store more complicated CFs with uncertainty of scenarios in SQLite
Tasks:
[ ] Update bw2io.convert_lcia_methods_data to product a lciafmt dataframe
[ ] Design the extra metadata fields for storing LCIA data (e.g. Context separator character, how bibliographic data is stored)
[ ] SQLite row schema for methods. Optional inclusion of dataframe in database.
We don't need to, or even necessarily want to, store all CFs in the database. Instead, taking inspiration from the read-only proxies in the IO table backend, we can interact directly with the processed arrays of characterization factors. This has several advantages:
Tasks:
bw2io.convert_lcia_methods_data
to product alciafmt
dataframeContext
separator character, how bibliographic data is stored)bw2io
importer forlciafmt
methods