EnvironmentQuery

Remarks

Doc 37 § D14, built. Unreal's EQS generates candidate points, runs scored tests over them and takes the best; utility scoring generates candidate actions, runs scored considerations over them and takes the best. Those are the same machine, so this is a IScoredCandidateSet`1 and its scoring goes through CandidateScoring — the same weighted geometric mean, with the same zero rule, as UtilitySet.

⚠ A list asset, not a second node graph. Generators, then tests in order, and no wiring decisions anywhere — which is what Unreal's EQS editor also is, once you look past the fact that it is drawn on a graph canvas: a root with a fixed list of children.

⚠ Cheap tests first is the author's decision and the query honours the order. A filtering test rejects a point and the run stops reading it, so putting a trace above a distance check costs a raycast per point that a subtraction would have thrown away. The editor says so; the runtime does not reorder, because a runtime that reordered would make a query's cost unpredictable and its behaviour depend on a heuristic nobody can see.

Fields and properties (6)

• public Symbol Name

  What it is called.

• public ReadOnlySpan<IQueryGenerator> Generators

  What makes the candidates.

• public ReadOnlySpan<QueryTest> Tests

  What filters and scores them, in order.

• public int Estimate

  How many points the generators expect to make.

• public QueryOrigin LastOrigin

  Where the last Run was measured from.

• public int CandidateCount

Methods (7)

• public EnvironmentQuery(Symbol name, IQueryGenerator[] generators, params QueryTest[] tests)

  Creates a query.

• public bool Run(in AgentContext context, in QueryOrigin origin, QueryResults results)

  Runs it.

• public QueryPoint CandidateAt(int index)

  One of them.

• public Symbol CandidateName(int index)

  What one candidate is called, for a table, an overlay and a diagnostic.

• public int FactorsOf(int index)

  How many factors score one candidate.

• public Symbol FactorName(int index, int factor)

  What one of a candidate's factors is called.

• public float ScoreOf(in AgentContext context, int index, Span<float> detail = default(Span<float>))

  Scores one candidate, and optionally reports every factor it read.

Used by (12)

• EnvironmentQueryLibraryVixen.Ai
• QueryBindingVixen.Ai.Nodes
• QueryContentCompilerVixen.Ai
• QueryContentTestsVixen.Ai.Tests
• QueryPreviewTestsVixen.Ai.Diagnostics.Tests
• QueryRunnerVixen.Ai.Nodes
• QueryScorerExitCriteriaTestsVixen.Ai.Tests
• QueryTestPurposeTestsVixen.Ai.Tests
• RunQueryNodeTestsVixen.Ai.Nodes.Tests
• WorldQueryTestsVixen.Ai.Nodes.Tests
• QueryDocumentVixen.Editor.AssetEditors
• QueryViewVixen.Editor.AssetEditors