René's URL Explorer Experiment

{"@context":"https://schema.org","@type":"DiscussionForumPosting","headline":"Implement Diagram.trace() for upstream restriction propagation","articleBody":"## Summary\n\nImplement `Diagram.trace(table_expr)` as the upstream mirror of `Diagram.cascade()` — walking the FK graph **upward** from a restricted table expression, propagating the key restriction through every ancestor with OR convergence.\n\n## Context\n\nDiscussion: #1232\nRelated: #242 (farfetch), PR #1407 (graph-driven cascade)\n\nPR #1407 delivered downstream graph operations (`Diagram.cascade()` and `Diagram.restrict()`). The upstream direction is missing — users cannot programmatically walk the FK graph upward to find which ancestor entities contributed to a given result.\n\n## Design\n\n`Diagram.trace(table_expr)` constructs a restricted upstream diagram:\n\n```python\ntrace = dj.Diagram.trace(MyChild \u0026 key)\ntrace.counts()          # entity counts per ancestor table\ntrace[Session]          # pre-restricted Session QueryExpression\ntrace[Session].fetch1(\"session_date\")\ntrace[ExtractTraces].to_arrays(\"trace\")\n```\n\n- **Direction**: upstream (child → ancestors)\n- **Convergence**: OR — an ancestor entity is included if reachable through *any* FK path\n- **`trace[TableClass]`**: returns the ancestor table pre-restricted through the FK join path\n- **`trace.counts()`**: entity counts per ancestor table\n- **`trace.heading()`**: all attributes across ancestor tables, grouped by table\n\n### Convergence rules\n\n| Method | Direction | Convergence | Question |\n|--------|-----------|------------|----------|\n| `cascade` | downstream | OR | What's affected if this is deleted? |\n| `restrict` | downstream | AND | What satisfies all these conditions? |\n| `trace` | upstream | OR | What contributed to this result? |\n\n### Restriction propagation (inverse of cascade)\n\nFor edge Parent→Child with `attr_map`, trace reverses the propagation: project child FK values back onto parent primary keys.\n\n## Implementation notes\n\n- Reuses existing Diagram infrastructure: dependency graph, attr_map handling, topological ordering, cross-schema discovery from #1407\n- `trace[TableClass]` via `__getitem__` — only table classes accepted\n- Returns standard `QueryExpression` objects — all existing query operators (join, aggr, proj, restrict) and fetch methods work unchanged","author":{"url":"https://github.com/dimitri-yatsenko","@type":"Person","name":"dimitri-yatsenko"},"datePublished":"2026-04-04T00:49:44.000Z","interactionStatistic":{"@type":"InteractionCounter","interactionType":"https://schema.org/CommentAction","userInteractionCount":2},"url":"https://github.com/1423/datajoint-python/issues/1423"}