Open Source & Case StudiesICALENDAR · MERGED OPEN SOURCE
MERGED UPSTREAM

Reproducible CLDR Provenance for iCalendar Time-Zone Mappings

A directly merged iCalendar contribution that resolves the latest CLDR source revision, fetches generated mapping data from that exact commit, and records the SHA in the output to prevent provenance drift.

Role

  • Python data generation
  • Supply-chain provenance
  • Immutable source pinning
  • Security regression testing
  • Maintainer collaboration

Category

Reproducible Data Generation

Engineering discipline

MERGED UPSTREAM

Merged pull request:iCalendar PR #1581
Maintainer approvals:4 approvals
Full test suite:15,984 passed

Engineering objective

System challenge

The generated Windows-to-IANA time-zone mapping did not expose the exact CLDR revision behind its data. Fetching moving-branch content without pinning the source could allow the recorded version and generated mapping to diverge.

Constraints

Working within real limits

  • Resolve the newest revision that actually touched windowsZones.xml
  • Fetch data from the same immutable revision recorded in generated output
  • Keep generated provenance machine-readable and human-auditable
  • Defend version parsing and output generation against malformed input

Approach

How it was built

  1. 01

    Resolved the newest commit touching windowsZones.xml through GitHub’s commits API.

  2. 02

    Fetched the XML from that immutable SHA instead of a moving branch.

  3. 03

    Recorded the same SHA in the generated module’s version field and source URL.

  4. 04

    Added generator, generated-output, ASCII-date, and injection-safety regression coverage.

  5. 05

    Addressed successive maintainer review rounds and reran the full project suite.

Workflow

Implementation flow

Query latest CLDR file revision
Capture immutable commit SHA
Fetch windowsZones.xml at SHA
Generate 139 mappings
Embed identical provenance
Validate complete repository suite

Technical validation

Implementation evidence

Observable engineering evidence. Discipline: MERGED UPSTREAM

Merged pull request

iCalendar PR #1581

Maintainer approvals

4 approvals

Full test suite

15,984 passed

Generated mapping

139 entries

Public evidence

Review the upstream contribution

Outcome

What was delivered

  • Four maintainers approved the final revision before merge

  • The generator produced 139 entries with one consistent SHA across version and documentation

  • 15,984 tests passed with Ruff and Towncrier validation clean

  • The pinned two-request workflow prevents version/data mismatch by construction

Technologies

Stack and tools

PythoniCalendarUnicode CLDRGitHub APIpytestRuffTowncrierSource provenance

Next engineering system

MICROSOFT OPEN SOURCE

Microsoft Agent Governance: Fail-Closed by Design

Have an ambitious AI system to build?

Let’s build something serious.

I work across AI architecture, product engineering, private inference, automation, SaaS infrastructure, and production hardening.