Skip to content

feat(docs): estimated SIG rating and levers per app on the quality matrix - #912

Merged
rubenvdlinde merged 8 commits into
mainfrom
feat/quality-matrix-rating
Nov 7, 2026
Merged

rubenvdlinde merged 8 commits into
mainfrom
feat/quality-matrix-rating

Conversation

@rubenvdlinde

Copy link
Copy Markdown
Contributor

Base: this branch stacks on #910 (feat/quality-matrix-page, head 755d16a). It also merges #911 (the gate 122 rating estimate), because the builder takes the estimate from gate 122. Review #909, #910 and #911 first; until they land this diff also shows their commits.

What changes

  • The quality matrix opens with an Estimated SIG rating section, labelled as an estimate:
    • the overall rating as a range, power mean p = -1 to p = 1, on 0.5 to 5.5
    • the five sub-characteristics
    • the seven properties, with the per-language ratings in the cell title
  • A levers panel per app: the top 5 levers, the gain from the next 1,000 lines under both exponents, the lines needed for +0.5, and whether the lever is sure (top 3 under both exponents).
  • The page explains the method in five steps and lists the sources and every assumption.
  • build-sig-matrix.py writes the rating section, apps.<app>.rating and levers.
  • Data rebuilt for all 20 apps, measured 2026-11-05.

Inputs used

  • 19 apps: the gate 122 reports in ~/sig-lanes/matrix/sig, the checkouts in ~/sig-lanes/matrix/clones, the gate runs in ~/sig-lanes/matrix/out.
  • openregister: the fresher report ~/sig-lanes/doc3/out/openregister.json and its checkout, both at development 6891dae. Its gate rows still come from the full gate run at 28cabe73, because no gate run exists for 6891dae.

Accessibility

  • Rating cells reuse the star colours, which pass WCAG AA in both themes. Each cell also shows the stars as text and the value.
  • The sure mark reuses the 4-star colour pair, adds a check mark, the word "sure" and a ring in the text colour. "No" carries screen-reader text that explains it.
  • Every lever table has column and row headers, and each app panel is a labelled section.

Verified

Not verified

  • The page was not viewed in a browser. The browser service was down.
  • No comparison with a real Sigrid rating. Every assumption is listed on the page.
  • No CI run was watched.

🤖 Generated with Claude Code

…he gates

The SIG scores page becomes the quality matrix: SIG maintainability on the
v18.1 limits with all five star levels, SIG's Open Source Health and
Architecture Quality rows where we measure them, our own banded measures for
tests, debt, security and accessibility, and one row per gate category that
opens to its gates. A table below the matrix states why each own measure is
there and the basis of its bands.

Assisted-by: ClaudeCode:claude-opus-5-5
…trix

The matrix opens with an estimated SIG rating section: the overall range
over power mean p = -1 and p = 1, the five sub-characteristics and the seven
properties, all from gate 122's rating estimate. A levers panel shows each
app's top 5 levers with the gain from the next 1,000 lines, the lines for
+0.5 and whether the lever is sure. The page explains the method and lists
the sources and every assumption.

Data rebuilt for all 20 apps. openregister uses the fresher report and
checkout at development 6891dae; its gate rows are still from 28cabe73.

Assisted-by: ClaudeCode:claude-opus-5-5
…ating

Assisted-by: ClaudeCode:claude-opus-5-5
Assisted-by: ClaudeCode:claude-opus-5-5
…ating

Assisted-by: ClaudeCode:claude-opus-5-5
@rubenvdlinde
rubenvdlinde merged commit 3077ebf into main Nov 7, 2026
56 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