Repository navigation
Add reproducible validator studies with MateProbe compatibility replay - #1
Merged
Merged
Conversation
added 10 commits
September 29, 2026 18:17
…-evidence # Conflicts: # mkdocs.yml
pablomate4b
changed the title
pablomate4b
changed the title
pablomate4b
marked this pull request as ready for review
September 29, 2026 22:56
Codex Review SummaryThis comment shows the latest Codex review activity on this pull request.
ℹ️ About Codex in GitHubYour team has set up Codex to review pull requests in this repo. Reviews are triggered when you
Codex reacts with 👀 while any review is running, comments if it has suggestions, and reacts with 👍 once all reviews finish with no findings. |
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
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
The technical report now includes controlled validator comparisons and retrospective reproductions of public bugs. This integrates that evidence with the current MateProbe a4 namespace and preserves all frozen a3 study sources, protocols, inputs, captures and reports unchanged.
The controlled study covers 144 pairs under eight authored validator profiles. Detailed Python assertions and the audit agree on all 1,152 classifications; boolean-only accounting loses attribution for 96 wrong-ID rejections, while valid controls expose 24 regressions. The historical study screens ten candidates and executes four comparisons; an explicitly documented compatibility follow-up reproduces three fixes across jsonschema and Marshmallow. The selected IDN case remains unsuccessful, and all setup failures, corrections, versions, locks and artifact hashes remain available.
Current users can run
python benchmarks/study_replay.py --output /tmp/mateprobe-study-replay. A scoped import bridge runs the original hash-checked drivers against MateProbe. Result comparison permits only explicit library-version metadata changes; changes to outcomes, evidence or report membership fail. CI separately verifies byte-identical replay with the published historical a3 package, compatibility with published MateProbe a4, and the current checkout. Documentation distinguishes these replays from fresh third-party executions and provides an isolated a3 environment for historical commands.Validation:
No core library behavior, published version, or frozen a2 evidence changes. These author-run studies support reproducible evidence accounting, not greater detection power than careful pytest, natural-language truthfulness, independent adoption, or measured authoring-cost savings. The paper makes those limitations explicit.