You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
{{ message }}
Repository navigation
PR-size check in PR CI: changed lines per PR, derived files excluded (p90 ≤200 gate) #5733
Refs #5498, #5510. Ordered by aprender-57 (operator plan step 5): every aprender PR is ≤200 changed lines, one ticket per PR; the gate is p90 PR size ≤200.
Loss
Nothing in PR CI measures PR size. Fold PRs and large PRs get past review sizing, hold seats longer and conflict more (#5732 is 12 files / 507 changed lines; the README-count fold conflicts in #5723/#5724/#5727/#5728). #5498 states the ≤200 target but has no check behind it.
Add a PR-size check to PR CI: count changed lines (additions + deletions) of the PR diff, excluding derived files (contracts/witness/*.json, contracts/contracts.nt, contracts/shapes.ttl, ci/ fragments, README counts), and report the number. Failing threshold and the p90 rollup are the flow owner's call at implementation.
Falsifiers (test first)
A fixture diff of 201 changed non-derived lines: the check FAILs and prints the count.
A fixture diff of 200 lines: PASS.
A diff of 5 source lines plus 500 derived-file lines: PASS (derived excluded).
Control: the check run against a PR labeled as a squash fold of two tickets reports the sum, not the largest part.
Process
Gate/workflow change: C324 rule 7 applies (GATE only with the operator's yes, one gate retired per gate added). Quorum + non-author "gate/workflow sign-off: YES" on the final commit, 2 seats. Milestone 0.72.0 (EPIC 20 step 1).
Activity