Skip to content

docs(analog): the sar_adc record says "reference scan", not "hand scan" (Refs #12117) - #12118

Merged
gHashTag merged 1 commit into
masterfrom
lane/analog-review-fu-12117
Oct 24, 2026
Merged

gHashTag merged 1 commit into
masterfrom
lane/analog-review-fu-12117

Conversation

@gHashTag

Copy link
Copy Markdown
Collaborator

Summary

  • docs/analog/mutation-sar_adc-2026-10-25.md: "hand scan" becomes "reference scan" in the survivor table and in the test list, with one sentence on what the scan is (written in the test, reuses match_count and miss_margins) and what pins the first-of-the-ties case (the third test).
  • Docs only. No spec, seal or number changes.

Why

The t27-bees review of PR #12108 flagged that the test computes its expectation algorithmically while the record said "hand scan". The reviewer is right about the wording; the other items of that review (a sentinel that cannot be reached, a literal 2 that is the par argument also passed to miss_margins, deterministic bisection bounds, line-number coupling that the record already states) were read against the code and need no change.

Refs #12117 (the sar_kick half of that issue needs the pane server back and follows in a second PR).

🤖 Generated with Claude Code

…n" -- the scan is written in the test (Refs #12117)

The t27-bees review of PR #12108 pointed at the phrase: the fit_cp test does not compare against a scan
done by hand, it re-scores all 41 Cp points with match_count and miss_margins inside the test. Say that, and
say what it does and does not check (the choice rule, not those two functions). No number changes.

Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
@github-actions

Copy link
Copy Markdown
Contributor

PR Dashboard

Generated at: 2026-10-24 22:23:47 UTC

Summary

Status Count
Total Open PRs 44
PRs with Failing Checks 40
PRs with All Checks Green 4
READY 1
FAILING 40
PENDING 0
NO CHECKS YET 0

These columns do not partition: 1 + 40 + 0 + 0 = 41, and there are 44 open PRs. A PR is being counted twice or not at all.

Seal Status

  • ⚠️ STALE -- sha256(compiler.rs)=55c0f1b1b32d != manifest seal=87e5cbd3ad94.
    The committed NMSE numbers were certified against an older compiler.rs.
    Run scripts/reseal-check.sh locally for the two-step reseal command (advisory; not a merge gate).

@tri-claw

tri-claw Bot commented Oct 24, 2026

Copy link
Copy Markdown

Summary by t27-bees

This PR updates the SAR ADC mutation documentation to replace "hand scan" with "reference scan" in two places, adding a sentence explaining that the reference scan is computed in the test by reusing match_count and miss_margins, and that the third test pins the first-of-the-ties case. The change corrects terminology flagged in PR #12108 review: the test algorithmically computes its expectation rather than using a manual scan.

Worth a look

  • docs/analog/mutation-sar_adc-2026-10-25.md:48 — The survivor table now says "reference scan"; verify this matches the test's actual behavior (the diff shows the test reuses match_count and miss_margins, which supports the term).
  • docs/analog/mutation-sar_adc-2026-10-25.md:63-66 — The added explanation says the third test "pins the first-of-the-ties case on a model built for it"; confirm the test file actually constructs such a model and asserts the tie-break, otherwise the doc overstates coverage.

Read: 1 files, +5 -3, the whole diff. Head 28f0fc9. Model: openai-compatible/nvidia/nemotron-3-ultra-550b-a55b. Commands: @t27-bees review | summary | help | pause | resume. Free for every repository.

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