Dataset

Synthetic ingestion source, deliberately defective

Name: Synthetic ingestion source, deliberately defective
Creator: OpenDeco
License: Apache-2.0

This record documents 8 exposures from Invented, defective on purpose. Access is public. Source-defined outcomes, limitations and provenance are documented below.

CautionSynthetic fixture, defective on purpose

Every value in this file was invented to make each gate fail at least once. The defects are deliberate and the file must never enter a benchmark or a published release.

RestrictedThis normalised version cannot become benchmark-eligible

Twenty-nine findings were raised across the fifteen gates, twenty of them at fail severity.

Overview

The file declares nine records and supplies eight. That single mismatch is one of the defects, and the source metadata consistency gate catches it.

Each row's notes field states which defect it carries.

Record identifier: synthetic_ingestion_failing_fixture
Source format: opendeco_synthetic_tabular_exposure_source 1.0.0
Rows supplied: 8
Records declared: 9
Findings raised: 29

Source and citation

As with the clean fixture, the file carries no source URL, no accession reference, no citation and no retrieval date. Its label field states that the defects are deliberate.

Source owner: OpenDeco
Source URL: Does not exist for this sourceNo public record is available for this field.
Parser: opendeco_tabular_exposure_parser@1.0.0
Citation: Does not exist for this sourceAn invented file is not a citable source.

Population and exposures

Eight rows reach the gates. Their profile values include an impossible ambient pressure, a negative duration and a discontinuity, so the numbers describe defects and not exposures.

Records evaluated: 8
Records declared: 9
Environment: Open water

Outcomes

Two findings come from the outcome code validity gate. The codes in those rows sit outside the vocabulary the file declares, which is what the gate exists to catch.

Does not exist for this source

No outcome class is assigned to this file, because outcome parsing fails on it by design.

Variables and units

The source vocabulary matches the clean fixture. One row declares a unit the parser cannot resolve, and the unit ambiguity gate raises a fail finding on it instead of guessing.

Source vocabulary declared in the fixture, and the canonical unit the parser converts it to under the canonical data contract.
Source field	Source unit	Canonical field	Canonical unit
depth	msw	start_ambient_pressure_pa	Pa
bottom_time	min	segment_duration_s	s
descent_time	min	end_time_s	s
ascent_rate	msw/min	pressure_change_rate_pa_per_s	Pa/s
gas	percent	oxygen_fraction	fraction of 1
temperature	degrees Celsius	water_temperature_k	K
surface_interval	min	preceding_surface_interval_s	s
outcome	vocabulary code	outcome_type	controlled term

Missingness

One finding comes from the missing required fields gate. Missingness on this file is a defect under test and carries no information about how a real source behaves.

Missing required field findings: 1

Access and licence

Data classification: PUBLIC
Access type: Public
Licence: Apache-2.0
Redistribution: Allowed

Model-development history

Does not exist for this source

No model was developed or calibrated on this file, and the leakage registry holds no relationship pointing at it.

OpenDeco normalization

The parser still produces a normalised version from a defective source, and that version is hashed like any other. Normalisation and acceptance are separate steps, so a file can have a stable content address and still be refused by the gates.

Parser version: 1.0.0
Quality policy: opendeco_default_quality_policy@1.0.0
Gate suite version: 1.0.0

Quality flags

Zero gates pass and fifteen fail. Twenty-nine findings are spread across three severities: three at warning, six at quarantine and twenty at fail.

Quality-gate outcomes for the defective ingestion fixture, gate suite 1.0.0.
Gate	Result	Records	Findings	Warning	Quarantine	Fail
schema_validity	Fail	8	4	0	0	4
impossible_times	Fail	8	1	0	0	1
negative_durations	Fail	8	2	0	0	2
impossible_gas_fractions	Fail	8	1	0	0	1
gas_fraction_sum	Fail	8	2	0	0	2
profile_discontinuity	Fail	8	1	0	1	0
impossible_ambient_pressure	Fail	8	4	0	0	4
duplicate_exposure_ids	Fail	8	2	0	0	2
duplicate_subject_exposures	Fail	8	2	0	2	0
outcome_code_validity	Fail	8	2	0	0	2
missing_required_fields	Fail	8	1	0	0	1
repetitive_ordering	Fail	1	2	1	1	0
source_metadata_consistency	Fail	8	3	2	1	0
unit_ambiguity	Fail	8	1	0	0	1
profile_semantics	Fail	8	1	0	1	0

Content addressing

Normalised version, SHA-256: 8698f0ae1865109c6c31261ae70ab1388f3bfc8aa1ae71e0a28109391958e3ccThe canonical serialisation of the normalised records the parser produced from this file.

Record state

Acquisition state: Quality gates pending
Classification: PUBLIC
Benchmark eligible: No

Elsewhere