Skip to content

[DO NOT MERGE] Autoresearch experiment to reduce the number of tests without reducing coverage - #6971

Closed
ericapisani wants to merge 31 commits into
masterfrom
autoresearch/less-tests-common-20260730
Closed

[DO NOT MERGE] Autoresearch experiment to reduce the number of tests without reducing coverage#6971
ericapisani wants to merge 31 commits into
masterfrom
autoresearch/less-tests-common-20260730

Conversation

@ericapisani

Copy link
Copy Markdown
Member

Focused on the common test suite on this pass, doesn't touch any integration tests.

The full debug x flush x pickle x level x algo x http2 matrix ran identical
assertions 192 times. The compression-relevant dimensions (level x algo x
http2) stay fully crossed; debug/flush/pickle rotate through cases so every
value is still exercised.
…4 -> 22)

Keep every distinct truthy/falsy word plus one mixed-case variant per
result class; the exhaustive table re-tested the same .lower() path.
… only

test_proxy 42->24, test_socks_proxy 18->10. Proxy resolution is
protocol-independent; each distinct scenario is still exercised.
env_to_bool parsing is exhaustively tested in test_utils; keep one case
per precedence arm (incl. config=False with env unset).
…moved the exact duplicate escaped-regex should_propagate_trace parameter case; the unescaped regex asserts identical matching behavior.

Result: {"status":"keep","test_count":2314,"runtime_s":122.32,"covered_lines":9779,"covered_branches":2781,"coverage_pct":37.503,"failed":0,"skipped":171}
…ault value (2); the unset case already asserts the same default branch, while the non-default override remains covered.

Result: {"status":"keep","test_count":2313,"runtime_s":122.86,"covered_lines":9779,"covered_branches":2781,"coverage_pct":37.503,"failed":0,"skipped":171}
…x substring matching is unchanged by the URL scheme, and the retained HTTP localhost case asserts the same positive match.

Result: {"status":"keep","test_count":2312,"runtime_s":122.21,"covered_lines":9779,"covered_branches":2781,"coverage_pct":37.503,"failed":0,"skipped":171}
…tain coverage of negative and boundary invalid inputs through the same max_size <= 0 rejection.

Result: {"status":"keep","test_count":2311,"runtime_s":123.61,"covered_lines":9779,"covered_branches":2781,"coverage_pct":37.503,"failed":0,"skipped":171}
… all non-string, non-list values follow the same no-op return path while None, integer, and boolean representatives remain.

Result: {"status":"keep","test_count":2310,"runtime_s":121.46,"covered_lines":9779,"covered_branches":2781,"coverage_pct":37.503,"failed":0,"skipped":171}
…some$' fail against 'some-string' under the default end-anchor behavior, while positive, case-sensitive, and explicit-anchor cases remain.

Result: {"status":"keep","test_count":2309,"runtime_s":120.8,"covered_lines":9779,"covered_branches":2781,"coverage_pct":37.503,"failed":0,"skipped":171}
… skip the matcher loop and return False independently of item, while empty-string and non-empty-item representatives remain.

Result: {"status":"keep","test_count":2308,"runtime_s":121.5,"covered_lines":9779,"covered_branches":2781,"coverage_pct":37.503,"failed":0,"skipped":171}
…list return occurs before item inspection, while empty-string and ordinary-string representatives remain.

Result: {"status":"keep","test_count":2307,"runtime_s":120.78,"covered_lines":9779,"covered_branches":2781,"coverage_pct":37.503,"failed":0,"skipped":171}
…the retained empty-string case covers the same loop-skipping False result, while ordinary-string behavior remains covered with regex_list=None and nonempty matchers.

Result: {"status":"keep","test_count":2306,"runtime_s":121.11,"covered_lines":9779,"covered_branches":2781,"coverage_pct":37.503,"failed":0,"skipped":171}
…retain both inclusive numeric boundaries, while True and False retain boolean coercion behavior.

Result: {"status":"keep","test_count":2305,"runtime_s":123.16,"covered_lines":9779,"covered_branches":2781,"coverage_pct":37.503,"failed":0,"skipped":171}
…retained empty-string None-list case covers the same early False return before item inspection.

Result: {"status":"keep","test_count":2304,"runtime_s":121.11,"covered_lines":9779,"covered_branches":2781,"coverage_pct":37.503,"failed":0,"skipped":171}
… boolean acceptance, and numeric zero independently preserves the False-coerced boundary outcome.

Result: {"status":"keep","test_count":2303,"runtime_s":122.01,"covered_lines":9779,"covered_branches":2781,"coverage_pct":37.503,"failed":0,"skipped":171}
…t already creates an empty cache and asserts get() returns None, exercising the same miss behavior before validating a subsequent hit.

Result: {"status":"keep","test_count":2302,"runtime_s":123.18,"covered_lines":9779,"covered_branches":2781,"coverage_pct":37.503,"failed":0,"skipped":171}
…perset: it asserts the same empty-cache miss and first set/get, then additionally validates overwriting an existing entry.

Result: {"status":"keep","test_count":2301,"runtime_s":123.54,"covered_lines":9779,"covered_branches":2781,"coverage_pct":37.503,"failed":0,"skipped":171}
… exercises the identical existing-key update branch and resulting retrieval assertion; cache capacity is irrelevant before eviction.

Result: {"status":"keep","test_count":2300,"runtime_s":122.33,"covered_lines":9779,"covered_branches":2781,"coverage_pct":37.503,"failed":0,"skipped":171}
…ed current-scope test covers both the None and assigned-span outcomes and exercises the public no-argument API.

Result: {"status":"keep","test_count":2299,"runtime_s":123.34,"covered_lines":9779,"covered_branches":2781,"coverage_pct":37.503,"failed":0,"skipped":171}
@ericapisani
ericapisani deleted the autoresearch/less-tests-common-20260730 branch July 31, 2026 12:52
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant