Rough idea: build the bipartite graph of who contributed to which project
Naive implementation: go from each origin (or last visit), and collect every author it references. But there is a lot of duplicate work.
Slightly less naive implementation: reverse-traversal in swh-graph, by tagging every revision/release/snapshot with the set of authors who contributed to their parent, starting from oldest revisions.
This might overflow memory though? If not, could we get away with a database?
| rDGRPH Compressed graph representation | |||
| D8972 | rDGRPH43567e66a132 origin_contributors: Rename 'person' to 'contributor' in outputs | ||
| D8971 | rDGRPHb7a991441696 origin_contributors: Write table mapping origin ID to origin URL (base64… | ||
| D8970 | rDGRPH09365b488aa1 origin_contributors: Use origin IDs instead of SWHIDs | ||
| D8912 | rDGRPHdfd4c1dc3b22 ListOriginContributors: Ignore null author/committer in revisions/releases | ||
| D8910 | rDGRPH559d4068bfe1 Regenerate the test dataset to include a release with no author | ||
| D8908 | rDGRPHf3235e318485 Add ListOriginContributors | ||
| D8883 | rDGRPHab2703efcb9a Add Luigi task TopoSort and add a simple test | ||
| D8883 | rDGRPH58f44785816b Improve comments | ||
| D8883 | rDGRPH922894410b6e Add a sample of two ancestor with each node | ||
| D8883 | rDGRPH7bee5d47a6eb revert multithreading, it's actually twice as slow as singlethread | ||
| D8883 | rDGRPH30dad16a2365 tentative multithread DFS | ||
| D8883 | rDGRPHed6636c26be8 Implement a naive topological sort | ||
| Status | Assigned | Task | ||
|---|---|---|---|---|
| Migrated | gitlab-migration | T4695 Provide a collaboration graph / dataset | ||
| Migrated | gitlab-migration | T4729 collaboration graph: drop pseudo-SWHIDs and add mapping ori<->url |
TODO: deanonymized dataset should be just a <contributor_id,contributor_base64,contributor_escaped> table, rather than repeating the origin<->contributor mapping