From cde3d912a9b1b945a7ea7916c1d70a828a670023 Mon Sep 17 00:00:00 2001
From: Nils Lehnen <30603423+iderex@users.noreply.github.com>
Date: Fri, 21 Aug 2026 05:49:59 +0200
Subject: [PATCH] List the pull-request parameter the parity table does not
 carry (#55)

This does not finish #55. Its remaining leg is the merge-method restriction,
which is a ruleset parameter rather than a file in this tree.

The table under `### The pull-request rule, parameter by parameter` opens by
saying it carries every parameter the command above it prints. It carried eight
rows and the command prints nine names:

    gh api repos/Flowfin/lab/rules/branches/main \
      --jq '[.[] | select(.type=="pull_request") | .parameters | keys[]] | length'
    9

`require_extra_approval_for_unattributed_changes` was the missing one, and it
has never appeared anywhere in this tree:

    git grep -n 'require_extra_approval_for_unattributed_changes' origin/main
    exit=1

Both boards print `true` for it:

    gh api repos/Flowfin/lab/rules/branches/main \
      --jq '.[] | select(.type=="pull_request") | .parameters.require_extra_approval_for_unattributed_changes'
    true
    gh api repos/Flowfin/jellyfin-plugin-sso/rules/branches/main \
      --jq '.[] | select(.type=="pull_request") | .parameters.require_extra_approval_for_unattributed_changes'
    true

What failure this prevents is a reader taking the table for the whole answer
when a parameter that could ask for an approval at a review count of zero was
not in it. The four review-adjacent rows all say they bite only where a review
is required, and this one is not in that family, so its absence read as
agreement with those four.

The row says what the two commands say and claims nothing about what the
parameter does. Whether it holds a merge at a count of zero is a statement about
the platform, no command in this walk answers it, and the row says so rather
than filling the gap with a guess.

The other eight rows and the rule-type table above were re-read at the same time
and every value still holds. How it was found: re-running the three commands the
section prints, before quoting a row from it.

Signed-off-by: Nils Lehnen <30603423+iderex@users.noreply.github.com>
---
 docs/quality-parity.md | 13 ++++++++++---
 1 file changed, 10 insertions(+), 3 deletions(-)

diff --git a/docs/quality-parity.md b/docs/quality-parity.md
index ab8b208..6533cf0 100644
--- a/docs/quality-parity.md
+++ b/docs/quality-parity.md
@@ -141,9 +141,15 @@ requires it to stay empty, so nothing here adds a second rule about it.
 ### The pull-request rule, parameter by parameter
 
 Every parameter the third command prints, with the target's value beside it. On
-2026-08-10 all but the first held the same value on both boards. A setting left
-at its default and a setting chosen deliberately look identical afterwards,
-which is why every row carries a reason and not only a verdict.
+2026-08-10 all but the first held the same value on both boards, and that is
+still the reading on 2026-08-21. A setting left at its default and a setting
+chosen deliberately look identical afterwards, which is why every row carries a
+reason and not only a verdict.
+
+The table listed eight rows and the command prints nine names. Which of the two
+that is, a parameter the platform started printing after the table was written
+or a row the walk missed on the day, is not readable from here, and the last row
+below is the one that was absent either way.
 
 | Parameter | Here | Target | Verdict |
 | --- | --- | --- | --- |
@@ -155,6 +161,7 @@ which is why every row carries a reason and not only a verdict.
 | `require_code_owner_review` | `false` | `false` | Kept. There is no `CODEOWNERS` file in this tree, so requiring a code-owner review here would require an approval nothing can name. |
 | `dismissal_restriction` | `{"allowed_actors":[],"enabled":false}` | The same | Kept. It restricts who may dismiss a review, and there is no required review to dismiss. |
 | `required_reviewers` | `[]` | `[]` | Kept, for the reason in the row above. |
+| `require_extra_approval_for_unattributed_changes` | `true` | `true` | Kept, and it is the row to read carefully. The four rows above turn on a review being required, and this one is written to ask for an approval where a change carries commits the pull request's author is not credited with, so it is the only parameter here that could ask for one at a count of zero. Whether it does is a statement about the platform and no command in this walk answers it, so nothing is claimed. It has never been observed to hold a merge on this board. |
 
 ### Why the merge methods are not a style preference