Repository navigation
Install tzdata and tzdata-legacy explicitly in the release image - #184
Conversation
The release stage's apt list names g++, libeigen3-dev and python3. tzdata reaches the image today only as a dependency of python3's standard library packages, and it is what gives the engine /usr/share/zoneinfo for named time zones. List it explicitly, so the zone files stay if that dependency changes. The image's packages do not change: the built inventories with and without this line are identical on amd64 and on arm64. The apt cleanup does not touch /usr/share/zoneinfo, and the hpo and default stages inherit the line. The test that pins the release stage to the original runtime stage undoes exactly this one line, as it does for the other approved changes. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Debian 13, the release stage's base, ships the backward-compatible zone names (US/Eastern, US/Central, Japan, GB and the like) in a separate package, tzdata-legacy, while tzdata keeps the canonical Area/Location names. A script that names a time zone the old way needs that package to resolve it. The release stage now installs tzdata-legacy after tzdata; the hpo and default stages inherit it. The pinned release-stage test undoes exactly the two lines, refusing unless tzdata follows python3 and tzdata-legacy follows tzdata. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
There was a problem hiding this comment.
Graphify reviewed this change.
No issues found by static checks — no coupling regressions or blocking issues in the code graph. That is not the same as safe to merge: see what was not checked.
Not checked: tests were not run; no formal proof of the changed code.
Graphify review — findings
Adds tzdata and tzdata-legacy to the runtime stage's apt install list in docker/Dockerfile, right after python3, so the image ships timezone data. The release-image test adds a _without_tzdata_package helper that strips exactly those two lines, and fails if either is missing, duplicated or out of place. It runs in the original-runtime-stage hash check, so the stage hash comparison still ignores only this approved addition.
No blocking issues surfaced.
Analysis details — impact, health, verification
Impact & health
Graphify review
Impact — 110 functions depend on the 105 functions this change touches.
Health — grade A; 1 existing hotspot(s) in the area this change touches (pre-existing, not introduced here):
_toolchain_argument_problems()— 4 callers, 4 callees (medium)
Verification — 110 functions in the blast radius were not formally verified this run (proofs are advisory here).
Gate & verification
graphify gate
PASS — objectively clean (no health regressions, tests not run — proofs not run this pass (advisory)). Grounded, not self-assessed.
Advisory (not blocking):
- verification_scope: 110 function(s) in the blast radius were not formally verified this run
Test selection
Test selection
34 of 34 test file(s) selected (100%) via static blast radius.
Escalated to a full run for safety (
noncode-change) — the selection is not trustworthy on its own (see below). CI should run the whole suite.
tests/test_facts_cards.py— full-run-safetytests/test_facts_descriptions.py— full-run-safetytests/test_facts_fanout.py— full-run-safetytests/test_facts_validate.py— full-run-safetytests/test_fingerprint_canonical.py— full-run-safetytests/test_parity_bundle.py— full-run-safetytests/test_parity_source_proof.py— full-run-safetytests/test_release_action_pins.py— full-run-safetytests/test_release_asset_fetch.py— full-run-safetytests/test_release_binding.py— full-run-safetytests/test_release_binding_workflows.py— full-run-safetytests/test_release_coordinator_receipts.py— full-run-safetytests/test_release_docs_runtime.py— full-run-safetytests/test_release_fanout_check.py— full-run-safetytests/test_release_guard_policy.py— full-run-safetytests/test_release_hosted_postmerge_receipts.py— full-run-safetytests/test_release_hpo_image.py— impact, changed-test, full-run-safetytests/test_release_main_push_inventory.py— full-run-safetytests/test_release_pair.py— full-run-safetytests/test_release_parity.py— full-run-safetytests/test_release_parity_image.py— test-dependency, full-run-safetytests/test_release_parity_runtime.py— full-run-safetytests/test_release_parity_wire.py— full-run-safetytests/test_release_receipt_collect_sweep.py— full-run-safetytests/test_release_receipt_legacy_previous.py— full-run-safetytests/test_release_receipt_shards.py— full-run-safetytests/test_release_receipts.py— full-run-safetytests/test_release_redeliver.py— full-run-safetytests/test_release_runtime_smoke.py— full-run-safetytests/test_release_sink_image.py— full-run-safetytests/test_release_sources.py— impact, test-dependency, full-run-safetytests/test_release_workflows.py— full-run-safetytests/test_repo_descriptions.py— full-run-safetytools/release-dry-run/test_wfrun.py— full-run-safety
non-code file(s) changed (
docker/Dockerfile) → running the full suite for safety (a code graph can't see config/fixture/data deps)
Selection is safe under the controlled-regression assumption; always-run tests + a periodic full run are the backstops. Advisory — it never changes the check verdict.
test_notice_and_harness_changes_cannot_hide_other_instructions hashes the release stage with its own chain of exact undo helpers, and it failed because that chain did not undo the tzdata and tzdata-legacy lines. It now does, and so does the check that a change outside the approved blocks still breaks the pin, so that check keeps failing for the change it introduces rather than for the two lines. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Summary
Two lines in the release stage's
apt-get installlist:tzdataandtzdata-legacy.tzdata(the zone files under/usr/share/zoneinfofor names such asAmerica/New_York) reaches the image today only as a dependency of python3's standard library packages. Listing it keeps the zone files if that dependency ever changes; the image's packages do not change with it.tzdata-legacyholds the backward-compatible zone names (US/Eastern,US/Central,Japan,GBand the like). Debian 13, the base of the release stage since the trixie move, ships them in this separate package, so the 1.11.0 image and main have no/usr/share/zoneinfo/US/Eastern, and a script that namesUS/Easternruns in UTC without any error:TZ=US/Easternreads 12:00 +0000 for 2026-07-01 12:00 UTC where EDT is 08:00 -0400. The bookworm-based 1.8.0 image had these names. Withtzdata-legacythey resolve again.tzdata-legacyis built from the same Debian source package and version astzdata(Source: tzdata,2026f-0+deb13u1), so the image's Debian source pairs, and the registered source set, do not change.What changes
docker/Dockerfiletzdataandtzdata-legacyadded to the release stage'sapt-get installlist, afterpython3. The cleanup line removes the apt lists, man, locale and info only, so/usr/share/zoneinfostays; thehpoanddefaultstages inherit both lines.tests/test_release_hpo_image.pytzdatadirectly followspython3andtzdata-legacydirectly followstzdata), as it does for the other approved release-stage changes.Checks
Each image was examined in a container:
dpkg -sfor both packages, the zone files,TZ=America/New_YorkandTZ=US/Easternfor 2026-07-01 12:00 UTC (andAmerica/New_Yorkfor 2026-01-15), the zoneinfo file count and the fulldpkg-queryinventory (binary, version, source, source version). The release stage was built with engine and codegen 1.11.0 and the tree's guard policy pin.America/New_YorkJuly / JanuaryUS/EasternJulyUS/Easternand file count)US/Easternand file count)tzdataonly)On each architecture the branch's inventory is main's plus the one binary package
tzdata-legacy(sourcetzdata, same version), and main's amd64 inventory equals the published 1.11.0 image's. Debian 13's package index (in adebian:trixie-slimcontainer) liststzdata-legacywithSource: tzdataandVersion: 2026f-0+deb13u1, the same version astzdata.Not run here: the repository's test suite (the hosted CI runs it on this pull request); a build of v1.8.0 from source (the published 1.8.0 image above is what runs it).
🤖 Generated with Claude Code