feat(context): add gen_test_case_guid for globally unique test case ids - #1166
Open
max-trunk wants to merge 4 commits into
Open
feat(context): add gen_test_case_guid for globally unique test case ids#1166max-trunk wants to merge 4 commits into
max-trunk wants to merge 4 commits into
Conversation
A test case id is only unique within a (collection, repo) pairing: ids are generated from framework-internal values, and `--no-repo` deliberately makes the same test share one id across a collection's repos. Consumers that need to address a single test case by one value -- links, APIs, webhooks -- have no id to use. Add `gen_test_case_guid(test_collection_id, repo_id, test_case_id)`, a deterministic UUIDv8 derived from that whole identity tuple, so it is unique by construction and inherits the tuple's semantics (including the `--no-repo` collapse to one id per collection). The hash contract is frozen: the three ids as canonical lowercase text, joined with `#`, SHA-256, first 16 bytes, stamped as an RFC 9562 UUIDv8. The stamp is required rather than cosmetic -- consumers validate the value with a UUID matcher that enforces the version and variant nibbles. Two golden vectors are pinned in the Rust tests and mirrored into the JS and Python binding suites, which is what keeps every copy of the contract honest. `gen_info_id` is untouched, so existing bindings and their pinned tests are unaffected. Exported through context-js and context-py alongside it. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
|
Merging to
After your PR is submitted to the merge queue, this comment will be automatically updated with its status. If the PR fails, failure details will also be posted here |
Codecov Report❌ Patch coverage is
Additional details and impacted files@@ Coverage Diff @@
## main #1166 +/- ##
==========================================
+ Coverage 82.97% 83.26% +0.29%
==========================================
Files 71 72 +1
Lines 16044 16269 +225
==========================================
+ Hits 13312 13547 +235
+ Misses 2732 2722 -10 ☔ View full report in Codecov by Harness. 🚀 New features to boost your workflow:
|
|
|
The doc comments restated the algorithm the code already shows. Keep only what is not visible from reading it: why the tuple is the identity, that the contract is frozen, and that the v8 stamp is load-bearing rather than decorative. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Nine new lint findings, all in the two test files this PR adds: - `vitest/prefer-describe-function-title` wants the identifier when a describe title is exactly a function name, while `vitest/valid-title` rejects a non-string title — the two conflict, and trunk's autofix for the first produced a violation of the second. A string that isn't an exact match satisfies both. The `gen_info_id` describe was pre-existing but became a "new" finding once my import shifted its line. - `vitest/require-to-throw-message`: assert the message rather than bare throw. - pytest isn't resolvable in the pyright environment (nothing else here imports it), so the raises test uses try/except instead of depending on it. - `context_py`'s stub is generated rather than committed, so pyright strict can't type anything imported from it. Annotating or `str()`-wrapping just relocates the unknown, so the two call sites and the import carry targeted suppressions, matching the existing precedent in test_parse_codeowners.py. `trunk check` is clean; 113 Rust and 33 context-js tests still pass. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
…uppressions pyright type-checks context-py against context_py.pyi, which is generated by a trunk action rather than committed. Trunk actions run on git hooks, not during `trunk check`, so a developer's tree has the stub and CI never does — every symbol imported from context_py is `Unknown` there, which is most of this repo's standing pyright findings and why adding one import produced a new one. The trunk check job already builds the wasm package for exactly this reason on the JS side (eslint needs the generated pkg/ to resolve). Generating the Python stub is the missing counterpart, and the job already builds the workspace, so it is nearly free. With the stub present the suppressions are unnecessary, so they are gone and the tests keep their natural shape. Existing pyright findings across context-py drop from 70 to 47 as a side effect. The raises test still uses try/except rather than pytest.raises: pytest is not resolvable in the pyright environment either, and nothing else in this directory imports it. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Summary
Adds
gen_test_case_guidto thecontextcrate: a single, globally unique id for a test case in a test collection, derived deterministically from the ids that already identify it.Why
A test case id on its own does not identify a test case within a collection:
--no-repodeliberately makes the same test share one id across them.So the real identity is the
(test_collection_id, repo_id, test_case_id)tuple. Anything that needs to name one test case with one value — a link, a request path, a webhook payload — has nothing to use today.What it does
Hashes the whole tuple, so the result is unique by construction and inherits the tuple's semantics rather than re-deciding them — including
--no-repo, where the nil repo id collapses to one guid per collection, which is the point of that flag.The hash contract is frozen:
#(matchinggen_info_id's convention),Uuid::new_v8),Step 4 is required rather than cosmetic: consumers validate this value with a UUID matcher that enforces the version and variant nibbles, which an unstamped truncated hash fails most of the time. v8 also visibly distinguishes it from the v4/v5 ids next to it.
A repo rename mints a new guid, because
test_case_idembeds the repo name. That is identical to today's behavior for the underlying id, and accepted.Tests
Two golden vectors are pinned in the Rust tests and mirrored into the JS and Python binding suites — the vectors, not a shared build, are what keep the copies honest. Also covered: determinism, uppercase input normalizing to the same value, the v8/variant stamp, each tuple member actually participating, and malformed input erroring rather than being hashed into a plausible-looking id.
Note the
contexttest module is gated on thebindingsfeature, so run these withcargo nextest run -p context --features bindings.Compatibility
gen_info_id's signature and behavior are untouched, so existing bindings and their pinned tests are unaffected. New deps:sha2(already inCargo.locktransitively) and uuid'sv8feature.🤖 Generated with Claude Code