Skip to content

Releases

Pinned historical inputs for dataset 0.7.0

Choose a tag to compare

@Mappboy Mappboy released this 17 Oct 05:59
2d9e174

Pinned public raw inputs for the reviewed dataset 0.7.0 recipe at commit 2d9e174.

The deterministic archive contains exactly the 11 required data/raw files from data/historical-inputs-manifest.json, including historical ACARA profiles. All pinned file sizes and SHA-256 hashes have been verified. It excludes databases, finance GeoPackages, review logs and unrelated working files. It is a provisioning bundle, not the dataset release.

Archive: historical-inputs-0.7.0.tar.gz
Size: 83,426,059 bytes
SHA-256: 3a91b8bc30d2f6e8d8610bdf0744e1dceb719b77654a80c0fef48ad3f95b1dcf

APEMAP Dataset Release v0.7.0

Choose a tag to compare

@github-actions github-actions released this 17 Oct 06:14
2d9e174

Dataset 0.7.0, built from merged commit 2d9e17433015d45243a84419dcf3b2a6b0fe1a7e with package 0.15.0 and the fresh reviewed recipe.

  • Recognize only normalized “Home educated” as a resolved non-institution assertion. Canonical institution_id is intentionally NULL, institution_resolution is home_educated, and attendance context retains recorded text with identity_basis = "non_institution" and no attended school ID. School-specific analyses and maps retain institutional scope.
  • Jordon Steele-John's accepted assertion retains source URL, retrieval instant, attendance confidence and both evidence references. It has no institution, recorded school identity or map point and generates no school-resolution work.
  • Pins 501 append-only review events, retained evidence and the existing immutable historical inputs. Earlier recipes, releases and artifacts are preserved. Dataset 0.7.0 is a MINOR increment above retained candidate 0.6.1 for the supported nullable contract on 0.x; previous published dataset is 0.3.3. Package and dataset versions are independent.

Verification completed 17 October 2026:

  • Implementation PR #100 and recipe PR #101 merged after all CI checks passed. The original CI failures were fixed by explicit reviewers in tests, verified with Git identity disabled (47 tests).
  • Two fresh local builds matched recipe metadata and artifact hashes. The publication workflow independently built twice, compared deterministic outputs, strictly verified and published the exact merged revision.
  • Downloaded the published archive, verified its distribution checksum against GitHub's asset digest, safely extracted it and passed 238/238 strict checks across 54 files. Canonical CSV/Parquet retain Jordon's NULL references; exported unmatched-school work excludes home education.
  • Inspected the diff from published 0.3.3: 4 added payloads, no removed payloads, 41 modified and 9 unchanged. Member/service/education row counts remain 580/1,754/352; school snapshots remain 171,822. Current reviewed work adds a net 7 institutions; the old synthetic Home educated institution is removed.
  • Local Windows and Actions UTC builds use equivalent timestamp offsets in text exports and consequently different derived readiness fingerprints. Each environment's replay is byte-deterministic; all canonical Parquet files match across environments, and normalized payload values match. The checksum below identifies the authoritative published archive.

Archive: apemap-release-v0.7.0.tar.gz
Size: 33569205 bytes
SHA-256: efff402034ba846001c80b230593cea9b12a444ca79847505b15d3ea5c4dace6

Pinned historical input bundle was verified against every required manifest file before upload and restored by the workflow. No PyPI publication or frontend deployment is included.

Release dataset v0.3.3

Choose a tag to compare

@Mappboy Mappboy released this 06 Oct 04:42

This unpublished historical dataset draft incorporates the reviewed school and member decisions committed at 013a8df810ce22442801b4040b01297979a569e1.

The consumed ledger has 163 events and 103 effective decisions, SHA-256 9376bb877fd7ae66620fe547baca46538ced25df62102f5b43ef0592c790e600. It covers opening-day cohorts for parliaments 42–48, ACARA 2008–2025 profiles and 2024 finance analysis, without refreshing upstream sources or acquiring private Finance 2024 observations.

Bundle verification and archive extraction/read-back each pass all 217 checks across 50 payload files. Archive: apemap-release-v0.3.3.tar.gz, 32,607,470 bytes, SHA-256 af5432b39bd6a428b9476fce935b19d202c0c66cad036e5f0f87fc68d098cba8. Verify SHA256SUMS.dist before extracting; run uv run apemap release verify <extracted-directory> --strict-assertions after extraction.

Compared with the recorded 0.3.0 baseline, 66 education institution assignments change. All 580 members, 1,754 service records and 352 education assertion IDs remain; every parliament retains its opening-day membership counts. Parliament 48 verified/provisional school coverage rises from 116 to 144 members, and unresolved school records fall from 101 to 58. Annual profile and finance benchmark row counts are unchanged. This comparison includes the entire reviewed snapshot, not only the latest day's changes.

The historical manifest keeps the pinned input generation timestamp for deterministic replay. This draft was prepared on 6 October 2026; the source commit and review snapshot identify the new decisions. Original APH acquisition dates remain unknown/null. Preserve source attribution and the warning that current school profiles and financial benchmarks do not describe attendance-era resources.

This draft is for review and frontend preview. Public publication requires reviewed sources on main and a reviewed historical recipe in release.yml; the current Action rebuilds standard inputs with 2021 finance and cannot reproduce this candidate. A GitHub release does not update the cpoole.dev dataset pin automatically. Older 0.3.0/0.3.1 artifacts remain unchanged.

Pinned APH/AEC inputs — 2026-10-02

Choose a tag to compare

@Mappboy Mappboy released this 02 Oct 00:40

Pinned raw inputs for reproducible APEMAP offline builds. Contains exactly six APH/AEC cached files listed in inputs-manifest.json, preserving their existing bytes; no source refresh. Tracked ACARA/ABS/reference inputs stay in Git.

Archive SHA-256: aba83a54dfcabc2cd889c2756e3ffe50cce240adc5704f62a15955216378db56

Sources: Australian Parliamentary Handbook API (https://handbookapi.aph.gov.au/), and Australian Electoral Commission March 2025 Commonwealth electoral boundaries (https://www.aec.gov.au/Electorates/gis/gis_datadownload.htm). AEC boundaries are attributed to the Australian Electoral Commission under CC BY 4.0. APH biographies are attributed to the Parliament of Australia; see https://www.aph.gov.au/Help/Disclaimer_Privacy_Copyright.

Packaged 2026-10-02 from existing cached inputs; original retrieval times were not recorded. Historical recorded_date labels in the manifest are not verified retrieval timestamps. Use uv run apemap inputs restore online, or uv run apemap inputs restore --archive <bundle> offline, then uv run apemap inputs verify.