Setting up a workflow for a pLCA database generator tool

[avdoever]

In our last meeting (find the notes here) we defined an objective to work on: creating a tool that combines compatible prospective scenarios into harmonized and ready-to-use databases for LCA practitioners. All ideas on how to progress toward that objective are welcome here.

Some tasks are:

• Defining and scoping the scenario types and choosing the data format we want to share/use in the tool
• Creating a metadata template
• Set up an automatic review of data and metadata
• Select the right tools/code for the database generator

What other tasks do you see? What are the first steps to be taken in the coming months?

[avdoever]

It may be good to consider end-user friendliness. Github is probably the best place for building the tool, but not necessarily most user-friendly for the general LCA practitioner. Not the first thing to work on, but maybe good to keep in the back of our heads.

[romainsacchi]

A server hosting a CKAN system might be more apt to do all that.
I'll come back with a demo if I find some time.

[romainsacchi]

Here is a suggestion of a workflow, using datapackages for custom premise scenarios (this is not about finished database exchanges).

There could be a main repository, where each sub-repo is a custom scenario for premise (e.g., see the bread scenario example).

On the user side, life's easier. Because we are using datapackages, we are only concerned by getting the datapackage's URL. With that URL comes all the needed files (scenario data, inventories, configuration file, etc.).

On the scenario generator side: the generator will try to push his datapackage to the main repo (containing all the needed files). Upon push, the datapackage will have the structure of its datafiles tested, but also github will run a battery of tests to make sure the inventories, scenario data, and the rest are as they should be. After passing the tests, this triggers a Github release of the repo, which archives the repo at Zenodo, and gets us a DOI.