DataManagerSnapshot

Overview

DataManagerSnapshot is a lightweight summary of a DataManager’s state, designed for golden trace tests. It captures three maps per data key:

• key_type — data key → type name string (e.g. "line", "points", "analog")
• key_entity_count — data key → number of entities/elements
• key_content_hash — data key → deterministic FNV-1a hex hash of content

The “media” key is always excluded since it is not command-managed.

Files

File	Purpose
src/DataManager/DataManagerSnapshot.hpp	Struct definition and snapshotDataManager() declaration
src/DataManager/DataManagerSnapshot.cpp	Implementation: entity counting, FNV-1a content hashing
src/DataManager/DataManagerSnapshot.test.cpp	Unit tests and golden trace tests

Entity Count Methods

Each data type uses a specific method to count entities:

Data Type	Method
PointData	getTotalEntryCount()
LineData	getTotalEntryCount()
MaskData	getTotalEntryCount()
AnalogTimeSeries	getNumSamples()
RaggedAnalogTimeSeries	getTotalValueCount()
DigitalEventSeries	size()
DigitalIntervalSeries	size()
TensorData	numRows()

Content Hashing

Content hashes use the FNV-1a 64-bit algorithm for determinism and speed. Each data type feeds its values into the hash in iteration order:

• Points/Lines/Masks: time index + coordinate values
• AnalogTimeSeries: time index + sample value
• RaggedAnalogTimeSeries: time index + value (via elements())
• DigitalEventSeries: time index (via view())
• DigitalIntervalSeries: start/end values (via view())
• TensorData: row count + column count (lightweight fingerprint)

Negative zero is normalized to positive zero for float inputs to ensure determinism.

Golden Trace Test Pattern

Golden trace tests verify command sequences produce expected DataManager states:

TEST_CASE("golden trace: CopyByTimeRange copies line entities") {
    auto dm = std::make_shared<DataManager>();
    // ... set up data ...

    auto const json = R"({
        "commands": [{
            "command_name": "CopyByTimeRange",
            "parameters": {
                "source_key": "src", "destination_key": "dst",
                "start_frame": 0, "end_frame": 100
            }
        }]
    })";

    auto seq = rfl::json::read<CommandSequenceDescriptor>(json).value();
    CommandContext ctx;
    ctx.data_manager = dm;
    auto result = executeSequence(seq, ctx);
    REQUIRE(result.result.success);

    auto snap = snapshotDataManager(*dm);
    CHECK(snap.key_entity_count.at("dst") == expected_count);
}

This pattern enables TDD for widget command coverage: capture a trace, replay it as a regression test, and verify the final state with snapshot comparison.

Test Coverage

The test file includes 15 test cases covering:

• Basic snapshot: empty DataManager, single data types, multiple data types
• Equality/Determinism: identical DMs produce equal snapshots, different data produces different hashes, empty objects have zero count
• Golden traces: CopyByTimeRange, MoveByTimeRange, AddInterval (with create_if_missing), multi-command sequences with variable substitution, replay determinism verification