SolverForge 0.11.x: Clone-Free Projected Scoring and Facade Exports

SolverForge 0.11.x is the previous projected-scoring runtime line. The line starts with clone-free projected scoring in 0.11.0 and ends at 0.11.1 before the 0.12.x runtime line.

Patch releases are folded into this line note. Use the latest 0.11.x patch only when you are intentionally staying on the 0.11 line, and keep generated-app targets explicit by checking the installed solverforge-cli output.

What Changed

Joined pairs can project retained scoring rows

Single-source projections still use named Projection<A> types with ProjectionSink and MAX_EMITS. Cross joins now also support retained scoring-only rows with the existing .project(...) verb:

type Streams = ConstraintFactory<Plan, HardSoftScore>;

Streams::new()
    .assignments()
    .join((
        Streams::new().capacities(),
        equal_bi(
            |assignment: &Assignment| assignment.bucket,
            |capacity: &Capacity| capacity.bucket,
        ),
    ))
    .project(|assignment: &Assignment, capacity: &Capacity| AssignmentCapacity {
        assignment_id: assignment.id,
        demand: assignment.demand,
        capacity: capacity.amount,
    })
    .penalize_hard_with(|row: &AssignmentCapacity| {
        HardSoftScore::of_hard((row.demand - row.capacity).max(0))
    })
    .named("Assignment capacity shortage");

The projected row is a scoring row, not a planning entity, fact, value range, or move target. It exists because the scoring rule needs a joined shape that is clearer than either source object alone.

Projected scoring paths are clone-free

Projected outputs, projected self-join keys, and grouped collector values no longer need to implement Clone. Retained projected state owns the row/key data that scoring needs instead of cloning it through hot paths.

This is a runtime contract improvement, not a modeling change. Existing single-source .project(NamedProjection) code keeps the same public shape.

Projected self-joins keep coordinate-stable order

Projected rows are retained by source ownership and emit index, not by sparse storage row IDs. Self-joins over projected rows therefore keep stable pair orientation even when storage rows are reused after incremental updates.

Constraint metadata borrows identity from the owner

Score analysis and lower-level scoring metadata preserve package-qualified ConstraintRef identity without cloning the reference into public reporting types. Package-qualified constraints still use ConstraintRef::full_name() keys such as package/name; package-less constraints use the short name.

Configuration controls are exported from solverforge

Since 0.11.1, application code can import the standard configuration controls directly from the facade crate:

use solverforge::{
    AcceptorConfig, ConstructionHeuristicType, ConstructionObligation,
    EnvironmentMode, ForagerConfig, HardRegressionPolicyConfig,
    MoveSelectorConfig, MoveThreadCount, PhaseConfig, RecreateHeuristicType,
    SolverConfig, SolverConfigOverride, UnionSelectionOrder,
};

Apps can keep using solverforge::SolverConfig::load(...), SolverConfig::from_toml_str(...), and SolverConfig::from_yaml_str(...) without depending on solverforge-config just to name phase, selector, acceptor, forager, or environment-mode types.

RecordingDirector is exported from the facade

RecordingDirector is available from solverforge beside Director and ScoreDirector:

use solverforge::{Director, RecordingDirector, ScoreDirector};

That matters for extension code that wraps a score director, records reversible trial moves, and rolls them back without depending on the lower-level scoring crate directly.

Runtime crate publishing is workspace-only

The release workflow publishes SolverForge runtime workspace crates in dependency order. The CLI remains its own repository and release line, so the runtime release process no longer treats a CLI package as part of the solverforge workspace publish sequence.

Install And Scaffold Status

For direct Cargo projects:

solverforge = { version = "0.11.1", features = ["serde", "console"] }

If you write custom incremental constraints that need lower-level identities, the companion workspace crates are also published at the same 0.11.x patch:

solverforge-core = "0.11.1"

For generated apps, confirm the installed CLI target:

solverforge --version

solverforge-cli 2.0.4 reports:

CLI version: 2.0.4
Scaffold runtime target: SolverForge crate target 0.11.1
Scaffold UI target: solverforge-ui 0.6.5
Scaffold maps target: solverforge-maps 2.1.4
Runtime source: crates.io: solverforge 0.11.1
UI source: crates.io: solverforge-ui 0.6.5
Maps source: crates.io: solverforge-maps 2.1.4

Generated apps created with solverforge-cli 2.0.4 start on solverforge 0.11.1. Move an older generated app to solverforge 0.11.x only when you are deliberately upgrading that app’s runtime target and validating the generated code against the newer core crate.

Patch History

Documentation Changes

At the time of this release, the docs tree was split so the 0.11.x source surface was easier to scan:

• Configuration documents the facade imports for SolverConfig, phase config, selector config, acceptor config, forager config, and overrides.
• Projected Scoring Rows covers single-source and joined-pair projection.
• Existence & Flattening separates if_exists, if_not_exists, and flatten_last from the main constraint-stream page.
• Scalar Move Selectors, List Move Selectors, and Composite Move Selectors split the previous monolithic move-selector reference.
• Crate & Runtime Map lists configuration controls and RecordingDirector as facade-level exports.
• The CLI command reference now links to scaffold, generator, and operations command pages instead of forcing every command into one long section.