Home › Learn › Shift left › What is missing
Shift left · page 9 of 13
The gap report: what a repo is missing
A gap report for data engineering repos: missing tests, checks, lockfiles and CI setup, built on repo-review and OpenSSF Scorecard plus per-function gaps.
In plain words
Point it at a repo; it lists the gaps
Findings say what is wrong in the code that exists. The gap report says what should exist but doesn't: a test for a function, a check in CI, a lockfile, a coverage threshold. It is the first thing to run on an inherited repo.
SETUP missing pre-commit configuration -> add .pre-commit-config.yaml (template provided) missing SQL lint for 14 .sql files and 9 SQL notebook cells missing coverage threshold in CI missing dependency lockfile present tests folder, CI workflow, bundle validated before deploy SECURITY PRACTICES pinned dependencies: no branch protection: yes static analysis in CI: no TESTS 17 pure functions with no test (listed) 3 modules with no test file 29 impure functions that could be pure (listed, with a hint each)
Illustrative report, not measured on a real repo.
Built from
Off the shelf, plus data-specific checks
| Part | Tool | Verdict | Why |
|---|---|---|---|
| Framework for 'is this repo set up right?' checks | repo-review (BSD licence) | Pursue | Checks are small plugins; we write the data-repo ones |
| Python project checks | sp-repo-review (Scientific Python's checks) | Borrow | pyproject, pre-commit and Ruff checks reused as they are |
| Security practices score | OpenSSF Scorecard | Pursue | Pinned dependencies, branch protection, static analysis in CI, and more |
| Per-function gaps | the purity, coverage and complexity reports | Build | Joins the three into one list |
| Data-repo checks | our plugins | Build | SQL lint present, bundle validated, notebooks thin, contract files present, plan check wired |
The report ends with one list in fix-first order: missing setup that blocks everything else, then complex untested functions, then pure untested ones.
FAQ
Questions people ask
Is there a tool that reports what a Python repo is missing?
repo-review is a framework for repo checks, and Scientific Python's sp-repo-review ships many checks on it. OpenSSF Scorecard scores security practices. Data-specific checks are added as plugins.
What should a data engineering repo have before code review?
A tests folder with unit tests for the pure logic, pre-commit with Python and SQL lint, a lockfile, CI that runs tests with a coverage threshold on changed lines, secret scanning, and bundle validation before deploy.
Next
Where to go next
Next page: The shift-left bundle: design, tools and status
↑ All of shift left · The repo on GitHub · Data project review