Known Limitations¶
bijux-gnss-nav owns scientific navigation behavior, not every condition under
which a full receiver run may succeed. The crate has broad proof, but each
proof family defends a specific claim: parsing, orbit/clock state, correction
models, SPP, PPP, RTK, integrity, or public-data comparability.
Limits Readers Should Know¶
| limitation | consequence | honest reading |
|---|---|---|
| Navigation proof is model-specific. | Passing a GPS broadcast orbit test does not prove Galileo, BeiDou, GLONASS, SP3, CLK, PPP, or RTK behavior. | Name the constellation, product type, correction, and estimator family in the claim. |
| Public-data fixtures are bounded. | Public references prove comparability for selected scenarios, not every station, epoch, geometry, or atmospheric state. | Treat station-truth and public-data tests as evidence, not universal certification. |
| Estimator families interact but remain distinct. | A correction or weighting change can alter SPP, PPP, RTK, integrity, and residual behavior differently. | Run the estimator family that consumes the changed model. |
| Parser correctness is not solution correctness. | Decoding a navigation message or product does not prove the final position solution is accurate. | Pair parser proof with orbit, clock, correction, or solution proof when the claim reaches those surfaces. |
| Local math helpers are domain-owned, not generic utilities. | Matrix, weighting, and model helpers can become attractive to other crates for the wrong reason. | Keep them private or nav-scoped unless they represent shared GNSS meaning. |
Claim Routing¶
flowchart TD
claim[navigation claim] --> parser{format or message decoding?}
claim --> model{orbit, clock, time, bias, atmosphere, or combination model?}
claim --> estimator{SPP, PPP, RTK, integrity, or residual behavior?}
parser --> parserProof[decoder and product tests]
model --> modelProof[model and correction tests]
estimator --> estimatorProof[solution and long-run tests]
parserProof --> scope[name supported constellation and fixture scope]
modelProof --> scope
estimatorProof --> scope
First Proof Route¶
Start with crates/bijux-gnss-nav/docs/TESTS.md, then select the proof by the
reader-facing claim:
- format and product parsing:
integration_precise_products.rs,integration_sp3_products.rs, or the constellation-specific navigation decode test - orbit, clock, time, and correction behavior:
integration_broadcast_orbit_accuracy.rs,integration_clk_reference_accuracy.rs, or the relevant ionosphere, troposphere, bias, or time-system test - positioning and integrity:
integration_position.rs,integration_position_protection_levels.rs, and the RAIM tests that match the claim - PPP and RTK:
integration_public_ppp_convergence.rs,integration_rtk_baseline_accuracy.rs, and the related ambiguity or differencing test - cross-run confidence:
long_run_stability.rsplus the focused family test above
If a document claims more than the selected proof can defend, narrow the claim or add the missing nav-owner evidence before publishing it.