Technical planning discussion

After the community call today, @beanow and @decentralion spent an hour or two discussing SourceCred implementation work.

  • Discussed having an intermediate “readable/manual” graph format that the initiatives (and artifacts, and spotlight) plugin compile down to
    • Discussed having a directory structure of manual graphs, where type information is encoded in a file header or (better yet?) a type.json or config.json file that is in the root of every manual graph directory
    • We kind of need an instance system for this to make sense
  • Discussed why we have the optimized (sorted, etc) graph json format that we have
  • Dandelion volunteered to champion a first-pass instance system in the coming weeks (should include the frontend, the scores file, the weights, and automatic grain distributions)
  • Spent a while bikeshedding :stuck_out_tongue: about config file formats – did a deep dive into yaml vs toml. Turns out there are a lot of people badmouthing toml, e.g. 1, 2
  • For prioritization: having the manually edited graph format would be nice (and allow people to e.g. write ad-hoc scripts for adding new contributions to the graph), however we should prioritize building a better UI for exploring the whole graph first