[karhu]

The Stack

This stack covers refactoring the internal logic of how releases are resolved during read, download & build, in particular also in the context of parent releases / parent datasets. On top of that it refactors the public TimeNet APIs and CLI interfaces around these actions, to streamline the behavior across them.

This PR

Every test runs with its own empty TIMENET_HOME.

Details

Some tests wrote into the real home folder of the developer. A build test in test_build_plan.py ran in-process without a cache folder, and the remote and S3 tests wrote control databases into ~/.cache/timenet. A TIMENET_* variable in the shell, or a .env file in the working directory, also changed the result of a test.

• A conftest.py at the root of the repository gives each test its own empty TIMENET_HOME. It deletes every other TIMENET_* variable, and the settings do not read a .env file during the tests.
• The connector checks run pytest from the root of the repository, so the root conftest.py isolates each connector test too.
• The tests that cleared TIMENET_ISOLATION themselves no longer do, because the fixture clears every TIMENET_* variable.
• Size: about 90 changed lines.

Design decisions

• The fixture sets only TIMENET_HOME, because at this point in the stack the store, the storage and the cache all derive from it.

Breaking changes

• None.

Stack

• docs: update the docs for the build coordinator #683
• feat(hints)!: give every error one hint for the SDK and one for the CLI #698
• feat(cache)!: record where each release in the local registry comes from #674
• feat(client)!: list and search every source #673
• feat(client)!: replace get() with get_manifest() and get_card() #672
• feat(build)!: build into the local registry, publish with --publish-to #671
• feat!: add timenet.build() and publish() #670
• feat(cli)!: take --local-registry, --remote-registry and --auto-build #669
• feat(config)!: make the default list of sources from three variables #668
• 👉 test: isolate every test from the real home and guard it #667
• feat(client)!: download releases into the local registry #662
• feat(client)!: take an ordered list of sources #661
• feat(client)!: read releases through the coordinator #660
• feat(coordinator)!: plan and run reads through the list of sources #659
• feat(coordinator)!: plan parents through an ordered list of sources #658
• feat(sources)!: resolve a release through an ordered list of sources #657
• feat(build)!: run every build through the coordinator #655
• feat(build)!: move the build CLI and build() onto the coordinator #654
• feat(build)!: build one prepared release in the built-in backend #653
• feat(coordinator)!: plan and run builds through build_release #652
• refactor(coordinator)!: name the plan step type Step #651
• feat(registry)!: copy a release between registries #649
• feat(registry)!: publish a local release with put_release #648
• refactor(build)!: let the connector choose the values backend #646

Changes later in the stack:

• The fixture sets TIMENET_HOME. The PR on defaults switches it to the new variables.

Merge message

test: isolate every test from the real home and guard it

A conftest.py at the root of the repository gives each test its own
empty TIMENET_HOME and deletes every other TIMENET_* variable. So no
test writes into the home folder of the developer or depends on the
shell.

- The settings read no .env file during the tests.
- The timenet tests run as if no connector package were installed. A
  test that needs a connector installs a fake one.
- Guard tests make sure that the shell variables, a .env file and the
  settings paths of a child process cannot reach the real home.

[claude]

LGTM. This is a test-only change that moves the per-test TimeNet isolation into autouse fixtures.

What was reviewed:

• The root conftest.py fixture is loaded for every package, because the only pytest config is the root pyproject.toml. It patches env_file safely: settings() builds a new TimeNetSettings on each call, so it reads the patched model_config every time.
• Each removed TIMENET_ISOLATION delenv/setenv call relied on the default "on", so removing them changes nothing. The "setting" branch in test_build_cli.py still sets "off".
• The .env case of the dropped test_token_defaults_to_none is now covered by the new .env guard test. The child-process guard runs with cwd=tmp_path, and the ruled-out .env concern only affects child builds outside these tests.

Extended reasoning...

The change touches only test code: a new root conftest.py, a new autouse fixture in packages/timenet/tests/conftest.py, a new guard test file, and per-test isolation lines removed from about ten test files, roughly 90 lines in all. It touches no production code and no security-sensitive surface. The removed lines either duplicate what the new fixtures do or set the default value. No CODEOWNERS file covers these paths. The only review threads are the author's own comments on a file outside this diff, so no third-party objection is open.

[claude]

No new blocking issues. 3 optional suggestions (nits or notes on pre-existing code) were found and not posted. Nothing in this review needs a push before merging.

[claude]

No new blocking issues. 2 optional suggestions (nits or notes on pre-existing code) were found and not posted. Nothing in this review needs a push before merging.