Skip to content

Review shows one pair of each pattern below the review floor - #147

Merged
mvsbm merged 1 commit into
mainfrom
review-patterns
Nov 15, 2026
Merged

mvsbm merged 1 commit into
mainfrom
review-patterns

Conversation

@mvsbm

@mvsbm mvsbm commented Nov 15, 2026

Copy link
Copy Markdown
Owner

What and why

Reviewers see only pairs from the review floor up, best first. The pairs below it are never asked about, so nobody sees which kinds of near-matches the run leaves apart. Some are worth a look: an ambiguous pattern can be two thirds true. People's answers on them are also the labels a method learns from.

  • nayose list scored --each keeps one pair of each pattern (each comparison's level) that the filters keep.
    • Each pair is drawn by a hash of its IDs.
    • The patterns with the most pairs come first.
    • Each pair carries like: how many pairs its pattern holds.
    • total counts patterns.
    • O(E log E), as the ordering already was.
  • Review page: a checkbox, "One of each pattern below", beside "Review from N points" (this browser keeps it).
    • When on, the queue shows one open pair of each pattern below N, and the tally reads "28 patterns below 4 points, one pair each".
    • Each pair says "one of 119 pairs like it".
    • Pairs are decided as any other. The rail still counts the normal queue.
  • Docs: usage.md's Review row; commands.md regenerated.

Checked in Chromium on a generated workspace of 20,000 records, with screenshots looked at:

  • switching on showed 28 patterns;
  • Different by mouse and Same by key counted down to 26, with 2 to apply;
  • the switch survived a reload;
  • switching off returned "2,054 pairs waiting from 4 points".

Tests:

  • test_a_run_s_pairs_rows_and_entities_…: --each gives one pair per pattern, as many patterns as the plain list holds; like sums to the plain total and is ordered most first.
  • test_review_counts_down_… (browser): the switch on and off, the rail unchanged.

Full suite: 303 passed on 3.14 (-n auto).

Checklist (CONTRIBUTING.md)

  • Reviewed before merging: a review pass on this pull request's changes (rule 10)
  • Unhappy paths tested at the real size: two at once, an input twice, the real data size, a lost server (rule 9)
  • Every new limit names itself (shortcut: …), and every store that grows says what bounds it (rules 3 and 4)
  • Docs and diagrams describe the code as it is

🤖 Generated with Claude Code

…itch: list scored --each keeps one pair of each pattern the filters keep, the patterns of most pairs first, each with the pairs it stands for

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_012o8JNBKSY1aREPbMQDG8g2
@mvsbm
mvsbm merged commit 570fa67 into main Nov 15, 2026
6 checks passed
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant