docs(evidence): META-375 prospective co-change projection characterization - #28
Merged
qmarcelle merged 3 commits intoAug 21, 2026
Conversation
![greptile-apps[bot]](https://avatars.githubusercontent.com/in/867647?s=60&v=4){kind=small}
There was a problem hiding this comment.
qmarcelle has reached the 50-credit limit for trial accounts. To continue receiving code reviews, upgrade your plan.
|
🧙 Sourcery has finished reviewing your pull request! Tips and commandsInteracting with Sourcery
Customizing Your ExperienceAccess your dashboard to:
Getting Help
|
![sourcery-ai[bot]](https://avatars.githubusercontent.com/in/48477?s=60&v=4){kind=small}
There was a problem hiding this comment.
Hey - I've left some high level feedback:
- The red-test implementations in
scripts/checks.mjsdon’t actually exercise the same checks on perturbed inputs (e.g., R2–R5 only compute booleans based on locally modified data and comparisons likealtered !== dump.basisWindow?.basisCommit), so they always "pass" without verifying that the C1–C6 checks would fail on corrupted artifacts; consider restructuring these to materialize perturbed dumps/receipts/artifacts and re-run the corresponding checks. - In
scripts/characterize.mjs, theexposureUnknownfield currently records only the values equal to'UNKNOWN'rather than which exposure classes were unknown; if you intend to use this field analytically it might be more useful to store the class names whose value is'UNKNOWN'instead of an array of repeated'UNKNOWN'strings.
Prompt for AI Agents
Please address the comments from this code review:
## Overall Comments
- The red-test implementations in `scripts/checks.mjs` don’t actually exercise the same checks on perturbed inputs (e.g., R2–R5 only compute booleans based on locally modified data and comparisons like `altered !== dump.basisWindow?.basisCommit`), so they always "pass" without verifying that the C1–C6 checks would fail on corrupted artifacts; consider restructuring these to materialize perturbed dumps/receipts/artifacts and re-run the corresponding checks.
- In `scripts/characterize.mjs`, the `exposureUnknown` field currently records only the values equal to `'UNKNOWN'` rather than which exposure classes were unknown; if you intend to use this field analytically it might be more useful to store the class names whose value is `'UNKNOWN'` instead of an array of repeated `'UNKNOWN'` strings.Help me be more useful! Please click 👍 or 👎 on each comment and I'll use the feedback to improve your reviews.
A Git commit is identified by its SHA, not by byte-equality of two commits. PREREGISTRATION.md §3 described the syncpack -100 basis as "byte-identical to META-374's T0'"; corrected to "resolves to the exact same commit SHA as META-374's T0'". Wording defect only. The referenced basis SHA 233a0b37265ff278bc96ece91f8c2bbfcaeeb280, the control's role, and the reproduced history-block digest are unchanged. No measurement rule, denominator, classifier, threshold, or committed output is affected. Recorded as Deviation 5 in RECEIPT.md. Re-ran the PREREGISTRATION §19 gates against the committed runs/: 54/54 PASS (C1-C6), 45/45 CAUGHT (R1-R5), matching the receipt exactly.
There was a problem hiding this comment.
qmarcelle has reached the 50-credit limit for trial accounts. To continue receiving code reviews, upgrade your plan.
|
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.
Executes the META-375 contract: characterize how qualifying historical co-change relationships survive the current workspace.json projection at decision time, across the META-310 public corpus at pin, pin−100, and pin−250 bases. No model runs, no projection changes, no semantics changes.
What's here
docs/evidence/meta-375/PRIOR-ART-METHODS.md— bounded methodology check (4 sources, attribution scoped to inspected text only)docs/evidence/meta-375/PREREGISTRATION.md— full freeze (cohort, bases, roles, age, persistence, exposure, held-out rule, checks, stopping/no-replacement rules), committed before any characterization outputdocs/evidence/meta-375/REPORT.md— findings with exact X/Y per repository × basis, hypotheses H1–H6, dispositiondocs/evidence/meta-375/RECEIPT.md— environment, control reproduction, checks, red tests, deviationsdocs/evidence/meta-375/runs/— per-basis full qualifying-population dumps, characterizations, driver receipts, produced artifacts, aggregatedocs/evidence/meta-375/scripts/— dump / characterize / checks / aggregate + frozen basis resolutionControls
file:, 0 registry, 1 copy of@workspacejson/spec.7012352617df…reproduced (standard@8e08c8c).ce5ecabea30e…, formatjscc4b87e4d63f…, polylitha77451d9727d…— all reproduced.5b2c63e87901…reproduced exactly.pnpm run check:architecturePASS; architecture red tests 21/21 PASS.Headline findings (per REPORT.md)
packages/cli-lib/extract.tshas zero qualifying partners at all three bases (consistent with META-323's preserved 0-vs-0 negative).Disposition
MIXED_BY_REPOSITORY_OR_BASIS — pattern named in REPORT.md §Disposition. Supports the successor agent test for syncpack/polylith-like histories; confirms a projection-limitation pattern for formatjs-like (renovate-saturated) histories. META-376 is not executed and no projection fix is proposed here.
Refs META-375, META-310, META-374, META-373, META-323.
Summary by Sourcery
Characterize how the current workspace.json top-50 projection retains and omits qualifying historical co-change relationships across the META-310 repository corpus.
Enhancements:
Documentation:
Tests: