m/market labResearch, with a record.
Source on GitHub ↗

Shared test inputs and resource ownership

These modules contain helpers, not collected tests. Test modules may import them; helpers do not import test modules. Keep expectations in the tests that assert them.

File Responsibility
agents.py Synthetic saved Agent definitions and run settings
clock.py Shared deterministic time with optional closed-market skips, a stand-in for the App's clock during assembly, and bounded condition waits
scheduling.py Runtime settings and a manual in-process clock
trading.py Trading source events and positioned request envelopes
tools.py Detached market/account values, tool envelopes and schema checks
async_reads.py Owned real-process fixtures with explicit cleanup registration
prepared_market.py Small prepared source and temporary store setup
book.py Indexed order-book source and prepared reader setup
financial.py Trusted native cursor/financial probe assembly
carrier.py Scripted model replies and reusable Agent experiment setup
models.py Agent-local model provider and process fixtures
execution.py Account/market scenario; direct callers must close the owned fixture
desktop.py Qt window setup and shutdown; Qt imports happen only during setup
failures.py Isolated expected-crash capture, diagnostic checks and private-output assertions

Setup mixins use the caller's addCleanup registration. ExecutionFixture also supports direct ownership with close() and cleans partially initialized resources on failure. No helper silently takes ownership of another TestCase's cleanup stack.

Native frame builders and the independent price-level reference live in itch_fixtures.py, shared by native and platform suites.

Maintained source: tests/platform/support/README.md