The DataRaft cheatsheet
Keep a report reproducible: retain its metric definition, input release IDs, quality evidence and report release. A saved number alone is not provenance.
| Your task | Call |
|---|---|
| Name an asset and its requirements | dr_product() + dr_add_contract() |
| Describe columns and keys | dr_contract() |
| Import a reviewable contract file | dataraft.adapters::dr_contract_from_odcs() |
| Define transformations | dataraft.core::dr_recipe() + dr_step_*() |
| Bind asset, recipe, source and target | dataraft.core::dr_workflow() + dr_add_*() |
| Check a delivery without writing | dr_run(write = FALSE, ) |
| Inspect checks and rejected rows | dr_quality_report(), dataraft.core::dr_quality_rows() |
| Recover a failed piped run | dr_last_failure() |
| Collect checked output | dr_collect() |
| Execute the configured writer | dr_run() |
| Choose a destination at execution | dr_publish(..., to = target) |
| Freeze a versioned report | dataraft.metrics::dr_report_release() |
Start with dr_run(write = FALSE, ). If the delivery passes, bind a target and use dr_run().
dataraft.lake::dr_ingest() is the lake-specific ingestion API. dataraft.core::dr_write_target() and
dataraft.core::dr_publish_metadata() are extension protocols, not additional beginner verbs.
| Quality action | Effect |
|---|---|
block |
Above threshold, no publication |
warn |
Above threshold, retain a warning and publish |
quarantine |
Remove every failing row; validate the remainder before writing |
threshold = .02 tolerates a two-percent failure fraction for block/warn.
Quarantine always removes failures, including missing predicate values.
Inspect dataraft.core::dr_quarantine_rows(result) locally; saving sensitive rows is explicit.
Evaluation errors always block. Contracts still check keys, types and required
columns after quarantine. A fully rejected delivery needs allow_empty = TRUE.
Vocabulary: product = asset specification; recipe = transformation model; trial = validation-only run; delivery = input batch; release = versioned snapshot; measurement = evaluated metric. These labels describe roles, not a new storage standard.
Attach reusable preparation with dataraft.core::dr_add_recipe(product, recipe).
Direct dplyr verbs on a product are a compact alternative. Sources, preparation,
checks and destinations all belong to the product definition.
Code examples are documented source examples; they are not executed in this website build.