Known Limitations

bijux-gnss-signal provides deterministic signal definitions and reusable DSP computations. Its results describe the modeled signal layer; they do not include receiver scheduling, propagation, navigation estimation, file ingestion, or artifact persistence.

From Signal Proof To System Claim

flowchart LR
    spec["signal specification or reference"]
    model["catalog, code, replica, or DSP model"]
    signal_test["signal-layer evidence"]
    receiver["receiver acquisition or tracking evidence"]
    navigation["navigation evidence"]
    system["bounded system claim"]

    spec --> model
    model --> signal_test
    signal_test --> receiver
    receiver --> navigation
    navigation --> system

Each arrow adds assumptions. A reference-aligned code generator can be correct while a receiver uses the wrong component, search window, sample clock, or tracking policy. A receiver can track correctly while navigation rejects the observations. Cite only the stages actually proved.

Current Boundaries

limitation	observable consequence	safe interpretation
Registry coverage is not receiver support.	Metadata or a local replica may exist for a signal that is not enabled through every acquisition, tracking, observation, or navigation route.	Use the registry for signal facts and the receiver support matrix for operational support.
Some model lookups are intentionally optional.	A missing default component or unsupported signal definition may return None or a typed error rather than a replica.	Handle absence explicitly; do not substitute another component or infer support from constellation alone.
GLONASS frequency is satellite-channel dependent.	Carrier and replica calculations cannot resolve a GLONASS signal from PRN and band alone.	Supply a validated frequency channel and preserve it in downstream identity and evidence.
Spectrum and front-end helpers are bounded models.	Their output reflects the supplied sample rate, grid, taps, bandwidth, and idealized component description, not a measured RF chain.	State numerical configuration and compare with measured evidence before making hardware claims.
Reference fixtures cover selected identifiers and assumptions.	A passing vector may not cover every PRN, component, chip order, sample rate, or long-duration boundary.	Name the reference family and add a vector or property test for newly claimed coverage.
Observation validation is structural signal-layer validation.	It checks timing, uniqueness, finite values, variance, C/N0, Doppler model, and lock-state labels; it does not establish estimator accuracy.	Route solution and integrity claims to navigation evidence.
Error representation is not uniform across all helpers.	Core signal utilities use SignalError, while observation validation currently returns string reasons.	Do not expose display strings as a durable machine protocol; preserve typed context at higher boundaries.
The crate has no product persistence or scheduling policy.	Successful computation says nothing about sample-source reliability, run completeness, or artifact durability.	Verify those claims in receiver and infrastructure layers.

The typed utility failures are defined in the signal error model. Optional replica and component behavior is visible in the acquisition model.

Evidence By Claim

flowchart TD
    claim["signal-layer claim"]
    catalog{"identity or metadata?"}
    waveform{"code, replica, or timing?"}
    spectrum{"spectrum or front end?"}
    obs{"observation compatibility?"}
    registry["registry and reference evidence"]
    continuity["reference plus chunk and duration evidence"]
    numerical["analytic expectation plus numerical grid evidence"]
    validation["accepted and rejected observation evidence"]

    claim --> catalog
    catalog -- yes --> registry
    catalog -- no --> waveform
    waveform -- yes --> continuity
    waveform -- no --> spectrum
    spectrum -- yes --> numerical
    spectrum -- no --> obs
    obs --> validation

Use the signal test strategy to choose a proof family. Useful entrypoints include the signal component registry evidence, CBOC spectrum evidence, and IQ conversion evidence. Constellation-specific code claims need their corresponding independent reference, not just another implementation of the same generator.

When To Leave This Crate

Use the receiver limitation guide for lock, handoff, and runtime claims. Use the navigation test strategy for solution, correction, and integrity claims. Use the infrastructure artifact contract for durability and history.

If available evidence does not cover the requested physical condition, narrow the claim and state the missing measurement. Do not make the signal crate sound like a field receiver by omission.