-
Notifications
You must be signed in to change notification settings - Fork 0
79 lines (76 loc) · 3.76 KB
/
Copy pathci.yml
File metadata and controls
79 lines (76 loc) · 3.76 KB
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
name: CI
on:
push:
pull_request:
jobs:
test:
strategy:
fail-fast: false
matrix:
os: [macos-latest, ubuntu-latest]
python: ["3.11", "3.13"]
runs-on: ${{ matrix.os }}
steps:
- uses: actions/checkout@v4
- uses: actions/setup-python@v5
with:
python-version: ${{ matrix.python }}
cache: pip
- run: python -m pip install -e ".[semantic,dev]" uv
- run: coverage run --branch -m unittest discover -s tests -q
# All first-party modules hold the clean-run 77% baseline. The four
# hardened helpers hold 90. The facade family and its largest gaps have
# separate floors, so one module cannot hide another module's loss.
- run: coverage report --fail-under=77
- run: coverage report --include="archive_intent.py,archive_store.py,platform_lock.py,schema_migrations.py" --fail-under=90
- run: coverage report --include="session_search.py,ss_*.py" --fail-under=75
- run: coverage report --include="ss_adapters.py" --fail-under=74
- run: coverage report --include="ss_retrieval.py" --fail-under=55
- run: coverage report --include="ss_cli.py" --fail-under=74
- run: python tests/run_public_card_quality_eval.py --expected evidence/public-card-quality-v0.1.0.json
- run: python tests/run_public_retrieval_eval.py --mode all --expected evidence/public-retrieval-v0.1.0.json
- run: python tests/benchmark_public.py --sessions 1000 --semantic --expected evidence/public-benchmark-v0.1.0.json
# The published baseline is a macOS hardware measurement. Running its
# model-build limit across unlike hosted runners creates false failures.
if: matrix.os == 'macos-latest' && matrix.python == '3.11'
- run: python tests/run_dashboard_latency_gate.py
- run: uv lock --check
- run: python scripts/check_public_privacy.py
- run: ss demo
package:
strategy:
fail-fast: false
matrix:
os: [macos-latest, ubuntu-latest]
python: ["3.11", "3.13"]
runs-on: ${{ matrix.os }}
steps:
- uses: actions/checkout@v4
with:
fetch-depth: 0
- uses: actions/setup-python@v5
with:
python-version: ${{ matrix.python }}
- run: python -m pip install build "pexpect>=4.9,<5" "wcwidth>=0.2,<1"
- run: python -m build
- run: git worktree add /tmp/session-search-v020 72e26f0
- run: python -m build --outdir /tmp/session-search-v020-dist /tmp/session-search-v020
- run: python -m venv /tmp/session-search-install
# Installed on purpose without the semantic extra. The journeys below
# are bounded at 90 seconds and must not reach the network, so they run
# with semantic retrieval unavailable. The test job above covers
# semantic and hybrid retrieval against the published evidence.
- run: /tmp/session-search-install/bin/pip install dist/*.whl
- run: /tmp/session-search-install/bin/ss capabilities
- run: /tmp/session-search-install/bin/ss demo
- run: /tmp/session-search-install/bin/python ${{ github.workspace }}/tests/test_adapter_pipeline.py -q
working-directory: /tmp
- run: /tmp/session-search-install/bin/python ${{ github.workspace }}/tests/run_installed_acceptance.py --ss /tmp/session-search-install/bin/ss
working-directory: /tmp
- run: python ${{ github.workspace }}/tests/run_task_usability_acceptance.py --ss /tmp/session-search-install/bin/ss
working-directory: /tmp
timeout-minutes: 2
- run: python ${{ github.workspace }}/tests/run_upgrade_acceptance.py --previous-wheel /tmp/session-search-v020-dist/*.whl --candidate-wheel ${{ github.workspace }}/dist/*.whl
if: runner.os == 'macOS'
working-directory: /tmp
timeout-minutes: 2