tringa
is a tool for querying test output across multiple CI builds on GitHub.
It is in early development and not ready for use.
Use uv:
uv tool install git+https://github.com/dandavison/tringa
And log in with gh.
gh auth login
tringa --help
Some commands print to the terminal, some bring up a TUI, and some bring up a SQL REPL for interactive queries.
By default the database is duckdb
and persists across invocations.
The REPL can be a traditional SQL REPL, or a Python session using the DuckDB Python API.
tringa repo show
tringa pr show
tringa pr tui
The DB has one table, named test
.
tringa pr repl
D select artifact, name from test
where passed = false and skipped = false and repo = 'temporalio/cli';
┌───────────────────────────────────────────┬─────────────────────────────────────────────────────────┐
│ artifact │ name │
│ varchar │ varchar │
├───────────────────────────────────────────┼─────────────────────────────────────────────────────────┤
│ junit-xml--10631569269--1--ubuntu-latest │ TestSharedServerSuite/TestWorkflow_Update_Execute │
│ junit-xml--10631569269--1--ubuntu-latest │ TestSharedServerSuite/TestWorkflow_Update_Start │
│ junit-xml--10631569269--1--ubuntu-latest │ TestSharedServerSuite │
│ junit-xml--10884926916--1--windows-latest │ TestServer_StartDev_ConcurrentStarts │
│ junit-xml--10885937402--1--ubuntu-arm │ TestSharedServerSuite/TestActivity_Complete │
│ junit-xml--10885937402--1--ubuntu-arm │ TestSharedServerSuite/TestWorkflow_Reset_ReapplyExclude │
│ junit-xml--10885937402--1--ubuntu-arm │ TestSharedServerSuite │
└───────────────────────────────────────────┴─────────────────────────────────────────────────────────┘
D SELECT name, type FROM pragma_table_info('test');
┌─────────────────┬───────────┐
│ name │ type │
│ varchar │ varchar │
├─────────────────┼───────────┤
│ artifact │ VARCHAR │
│ repo │ VARCHAR │
│ branch │ VARCHAR │
│ run_id │ VARCHAR │
│ sha │ VARCHAR │
│ file │ VARCHAR │
│ suite │ VARCHAR │
│ suite_time │ TIMESTAMP │
│ suite_duration │ FLOAT │
│ name │ VARCHAR │
│ classname │ VARCHAR │
│ duration │ FLOAT │
│ passed │ BOOLEAN │
│ skipped │ BOOLEAN │
│ flaky │ BOOLEAN │
│ message │ VARCHAR │
│ text │ VARCHAR │
├─────────────────┴───────────┤
│ 17 rows 2 columns │
└─────────────────────────────┘
Suppose you want to query for tests that have failed on main
in the last 7 days:
tringa --json --since-days 7 repo sql \
'select classname, name, count(*) as failure_count from test where passed = false and skipped = false group by classname, name order by failure_count desc' \
--branch main --workflow-id 80591745 https://github.com/temporalio/temporal
To experiment with alternative SQL queries interactively, change that to repo repl
:
tringa --since-days 7 repo repl \
--branch main --workflow-id 80591745 https://github.com/temporalio/temporal
For tringa
to find output from a CI workflow run, at least one job in the run must upload an artifact containing a directory of junit-xml format files (named uniquely for that job).
For example, the following fragment of GitHub Actions workflow yaml creates a directory containing junit-xml output from two different test suite runs, and uploads the directory as an artifact.
You must ensure that the artifact name is unique within the repository (so you'll probably want to use ${{github.run_id}}
at least)
- run: mkdir junit-xml
- run: my-test-command --test-suite-variant=something --junit-xml=junit-xml/${{ matrix.python }}-${{ matrix.os }}-something.xml
- run: my-test-command --test-suite-variant=something-else --junit-xml=junit-xml/${{ matrix.python }}-${{ matrix.os }}-something-else.xml
- name: "Upload junit-xml artifacts"
uses: actions/upload-artifact@v4
if: always()
with:
name: junit-xml--${{github.run_id}}--${{github.run_attempt}}--${{ matrix.python }}--${{ matrix.os }}
path: junit-xml
retention-days: 30