Repository navigation
Comparing changes
Open a pull request
base repository: go-openapi/validate
base: v0.26.4
head repository: go-openapi/validate
compare: v0.26.5
- 7 commits
- 6 files changed
- 2 contributors
Commits on Aug 25, 2026
-
test(messages): halt on the first failing fixture only in dev mode
Both Test_MessageQuality*_Issue44 passed haltOnErrors=true, while testWalkSpecs runs every fixture with t.Parallel(). checkMustHalt therefore stopped the run on whichever fixture lost the race and the rest were never reported, so the fixture named in the failure changed from run to run. haltOnErrors now follows DebugTest, which already decides whether the fixtures run serially. A normal run reports every failing fixture; a run with SWAGGER_DEBUG_TEST=1 still stops at the first one, serially, which is what that mode is for. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com> Signed-off-by: Frederic BIDON <fredbi@yahoo.com>
Configuration menu - View commit details
-
Copy full SHA for ce91d00 - Browse repository at this point
Copy the full SHA ce91d00View commit details -
test(messages): add TestGenExpectations to rebuild the expectations file
expected_messages.yaml pins some 700 messages over 50 fixtures and had no way to be re-derived, so it drifted: the -enable-long lane has been red since v0.23.0, growing from 7 failing fixtures to 17. TestGenExpectations runs every fixture through the validator in both modes and rewrites the file. It edits the yaml.Node tree rather than re-marshalling the document, so the 48 comments in the file survive: an expectation that still matches keeps its node and its spelling, one that matches nothing is deleted, an actual message nothing covers is appended, and withContinueOnErrors is set from whether the message also appears in the stop-on-errors run. It skips unless SWAGGER_GEN is set, and running it twice changes nothing. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com> Signed-off-by: Frederic BIDON <fredbi@yahoo.com>
Configuration menu - View commit details
-
Copy full SHA for e83999e - Browse repository at this point
Copy the full SHA e83999eView commit details -
test(messages): refresh expected_messages.yaml from the validators
The messages now name the parameter they are about - "parameters.0.in" in the stop-on-errors run and "parameters.myid.in" in the continue-on-errors one, where the file still expected "parameters.in" - and example validation reports warnings that did not exist when the file was written. Regenerated with TestGenExpectations. The two "could not resolve reference" expectations were widened by hand for the "cannot load spec:" prefix loads now adds, rather than replaced by a literal carrying an absolute path. Three fixtures record a defect instead of a message, each with a todo saying so: - fixture-581.yaml and fixture-additional-items-3.yaml no longer load at all, and have not since v0.23.0, so both now expect a load error. The first spells "maximum: 18446744073709551615", a YAML integer beyond int64, which the conversion refuses; the second spells "properties: {v: string}", which is not a schema and spec.Schema refuses. Both were written to exercise validation and reach no validator today. What fixture-581.yaml now demonstrates is the unmarshalling error itself, which names the offending value but not the field it came from. - fixture-constraints-on-numbers.yaml does not report "param3 in query should be a multiple of 10", though the fixture's own comment says it should. MultipleOf compares data/factor through conv.IsFloat64AJSONInteger, whose tolerance is relative: at a quotient of 2.1e8 it accepts a remainder up to 0.2, and 2147483648/10 = 214748364.8 falls inside that. MultipleOfInt and MultipleOfUint decide the same numbers correctly, so only a value arriving as a float64 is affected. The todo says why this is not a tolerance to retune: multipleOf is only decidable with decimal arithmetic. The -enable-long lane is green. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com> Signed-off-by: Frederic BIDON <fredbi@yahoo.com>Configuration menu - View commit details
-
Copy full SHA for da10635 - Browse repository at this point
Copy the full SHA da10635View commit details -
ci: re-instate error message stability test in ci
Signed-off-by: Frederic BIDON <fredbi@yahoo.com>
Configuration menu - View commit details
-
Copy full SHA for ab4461e - Browse repository at this point
Copy the full SHA ab4461eView commit details -
Merge pull request #292 from fredbi/fix/message-quality-expectations
Fix/message quality expectations
Configuration menu - View commit details
-
Copy full SHA for 07d44fe - Browse repository at this point
Copy the full SHA 07d44feView commit details -
fix: onboard spec url normalization fix (#293)
Signed-off-by: Frederic BIDON <fredbi@yahoo.com>
Configuration menu - View commit details
-
Copy full SHA for 251c9c5 - Browse repository at this point
Copy the full SHA 251c9c5View commit details -
Signed-off-by: Frederic BIDON <fredbi@yahoo.com>
Configuration menu - View commit details
-
Copy full SHA for a214a69 - Browse repository at this point
Copy the full SHA a214a69View commit details
This comparison is taking too long to generate.
Unfortunately it looks like we can’t render this comparison for you right now. It might be too big, or there might be something weird with your repository.
You can try running this command locally to see the comparison on your machine:
git diff v0.26.4...v0.26.5