Compilation results#
Types on this page are in parcel_core::compile. They expose public fields; compilation constructs them together. SchemaRef is a shared Arrow schema, and Expr and LogicalPlan are DataFusion types.
Compilation#
Field |
Type |
Meaning |
|---|---|---|
|
|
Rules and column definitions used when applying the contract. |
|
|
Plan that computes quality results and statistics. |
|
|
Calculations and storage layout for writing data under the contract. |
The CompiledBytes trait provides to_bytes and from_bytes for saving and loading all three artifacts.
CompiledContract#
Field |
Type |
Meaning |
|---|---|---|
|
|
Contract name. |
|
|
Semantic version ( |
|
|
Hash of the contract definition. |
|
|
Hash of this compilation. |
|
|
Resolved owner identifier. |
|
|
Resolved data location. |
|
|
Resolved data-placement requirements. |
|
|
Schema of the input data. |
|
|
Schema of the exposed output. |
|
|
Caller-only expressions bound to DataFusion placeholders. |
|
|
Caller decision rules in CEL. |
|
|
Rule IDs paired with live row-filter expressions. |
|
|
Quality assertions and their stored flag definitions. |
|
|
Output column names paired with live expressions, in output order. |
|
|
Dataset requirements in CEL. |
|
|
Result-shaping rules. |
|
|
Exact function implementations required by the contract. |
|
|
Per-rule compilation explanations. |
|
|
Reference CEL expressions used to compare evaluators. |
|
|
Row calculations that can be stored at write time. |
|
|
Input schema including calculated custom row fields. |
|
|
Calculations producing custom row fields. |
|
|
Declared types of custom caller fields. |
|
|
Filters using stored calculated values; only valid for data written under this contract. |
|
|
Output expressions using stored calculated values. |
ValidationPlan#
Field |
Type |
Meaning |
|---|---|---|
|
|
DataFusion plan returning one validation-summary row. |
|
|
Statistics computed by the plan. |
|
|
Assertion IDs with failure counts in the result. |
|
|
Guarantee IDs evaluated from data alone. |
|
|
Guarantee IDs requiring write metadata or request time. |
WritePlan#
Field |
Type |
Meaning |
|---|---|---|
|
|
Custom row calculations performed first. |
|
|
Assertion results to store. |
|
|
Reusable row calculations to store. |
|
|
Suggested clustering, partition, and Bloom filter columns. |
|
|
Additional metadata keys required alongside statistics. |
Layout#
Field |
Type |
Meaning |
|---|---|---|
|
|
Stored flag columns used for clustering. |
|
|
Partition columns. |
|
|
Columns identified for Bloom filters. |
ReportEntry and Tier#
Field |
Type |
Meaning |
|---|---|---|
|
|
Rule ID. |
|
|
Operation name. |
|
|
Evaluation category. |
|
|
Explanation of the category. |
|
|
Canonical CEL expression, when present. |
|
|
The rule’s |
|
Meaning |
|---|---|
|
Evaluated once per request without reading rows. |
|
Filter can use statistics to skip files or row groups. |
|
Evaluated row by row during scanning. |
|
Calculated at write time and stored. |
|
Calculated when its output column is selected. |
|
Requires a physical operation. |
Supporting rule and statistic types are listed in Compiled rule types.