Say what these two empty registers are worth, not what the tree used to lack

[iderex]

The issue this belongs to

Closes #406

What changed

The paragraph in .github/lint/excluded-lints and
.github/thread-detector/suppressions that explains why each register is empty now
gives a reading of this tree, anchored to a commit, instead of an absence of code.
Each also says what it claimed before, and the two say it differently, because the
lint sentence drifted and the detector sentence never held.

Both registers stay empty. No rule, fixture, threshold, expression or setting of
either leg changes, and neither script's parser sees anything new: every added line
is a comment.

What failure it prevents

An empty debt register reading as worthless. Both files say of themselves that a
line in them is a debt rather than a dispensation, so what their emptiness is
evidence of is the thing a reader takes away. Empty beside "there is almost no code
here yet" is something a reader discounts. Empty over 17,200 lines of code that the
analyser found nothing to excuse in, and over seventeen files that take a lock or
share an atomic that the detector found nothing to suppress in, is a statement
about the gate. Both files were giving that statement away.

This is a correction and the failure has already happened. What was wrong is the
reason each paragraph gives. How it was found: by re-reading the paragraph in each
register against the tree it describes, after the same class of stale claim was
repaired in #404.

Evidence

Read at a194071f3207d3903f6b40e27feef80aa1ad3d9a, which is the base of this
branch, and the head being pushed is 51a2d01.

The two paragraphs as they stood.

git rev-parse origin/main
a194071f3207d3903f6b40e27feef80aa1ad3d9a
git show a194071f:./.github/lint/excluded-lints | sed -n '12,14p'
# THIS REGISTER IS EMPTY OF ENTRIES TODAY, AND THAT IS A FACT ABOUT THE TREE
# RATHER THAN ABOUT THE RULE. There is almost no code here yet, so nothing has
# had a chance to need an exclusion. The refusal above is proven by the fixtures
git show a194071f:./.github/thread-detector/suppressions | sed -n '19,21p'
# THIS REGISTER IS EMPTY OF ENTRIES TODAY, AND THAT IS A FACT ABOUT THE TREE
# RATHER THAN ABOUT THE RULE. The suite this leg runs holds no code that takes a
# lock or shares a value across threads yet, so nothing has had a chance to need

Every number written into the two files.

git ls-tree -r --name-only a194071f src/ | wc -l
46
git ls-tree -r --name-only a194071f src/ | xargs -I{} git show "a194071f:{}" | grep -vE '^\s*(//.*)?$' | wc -l
17200
git grep -lE 'Mutex|RwLock|Arc<|thread::spawn|AtomicU|AtomicBool' a194071f -- src/ tests/ | wc -l
17
git ls-tree -r --name-only 3b236ce src/ | wc -l
9
git ls-tree -r --name-only 3b236ce src/ | xargs -I{} git show "3b236ce:{}" | wc -l
238

The 17,200 drops blank lines and whole-line comments, so it counts lines that are
neither. It is a floor for what the analyser read, not a measure of it.

That each register is still empty, which is the half of each sentence that
survives.

grep -v '^#' .github/lint/excluded-lints | grep -c . ; echo "exit=$?"
0
exit=1
grep -v '^#' .github/thread-detector/suppressions | grep -c . ; echo "exit=$?"
0
exit=1

That both legs are green at the commit the two readings are anchored to, which
is what makes an empty register a statement rather than a silence.

gh api repos/Flowfin/core/commits/a194071f3207d3903f6b40e27feef80aa1ad3d9a/check-runs --jq '.check_runs[] | select(.name=="thread-detector" or .name=="lint") | "\(.name) \(.status) \(.conclusion)"'
lint completed success
thread-detector completed success

The lint sentence was true and drifted.

git log -S'There is almost no code here yet' --format='%h %ad %s' --date=short a194071f -- .github/lint/excluded-lints
3b236ce 2026-08-25 Analyse the tree with the language's own analyser (#17)

Nine files and 238 lines at that commit, counted above, against 46 files today.

The detector sentence never held.

git log -S'lock or shares a value across threads yet' --format='%h %ad %s' --date=short a194071f -- .github/thread-detector/suppressions
bb128ae 2026-08-26 Run the suite under a thread detector, and prove the detector is on (#117) (#197)
git show bb128ae^:src/measurement/mod.rs | grep -nE 'AtomicU64|Mutex' | head -5
31:use core::sync::atomic::{AtomicU64, Ordering};
259:    next_id: AtomicU64,
286:            next_id: AtomicU64::new(1),
403:    use std::sync::Mutex;
413:        state: Mutex<ControlledState>,
git show bb128ae --stat --format='' | grep -c 'src/measurement/mod.rs' ; echo "exit=$?"
0
exit=1

The first of those is the commit BEFORE the paragraph landed, and the third shows
that the change writing the paragraph did not open that file.

The sibling whose form the two now take.

git show a194071f:./.github/codeql/excluded-rules | sed -n '12,14p'
# THIS REGISTER IS EMPTY OF ENTRIES TODAY, AND THAT IS A FACT ABOUT THE TREE
# RATHER THAN ABOUT THE RULE. The analysis reports nothing against this tree at
# the commit this landed on, so nothing has had a chance to need an exclusion.

Both register parsers still hold, run at this head.

bash .github/lint/lint.sh selftest | tail -1
Every fixture held. The rules the gate applies are the rules these fixtures ran.
bash .github/thread-detector/thread-detector.sh selftest | tail -1
Every fixture held. The rules the gate applies are the rules these fixtures ran.

The paths this change touches.

git diff --name-only origin/main...HEAD
.github/lint/excluded-lints
.github/thread-detector/suppressions

The means. The change is prose inside the comment header of two register
files, in the format those files already are. It adds no language, no runtime and
no dependency.

What this does not cover

• Whether either register should hold an entry. Nothing here runs either
  analyser looking for one, and both registers are still empty after this change.
• .github/codeql/excluded-rules and .github/format/unformatted-paths. They
  carry the same paragraph. The first is already anchored and the second's reason
  still holds today; both are read in Two debt registers explain their emptiness by an absence of code the tree no longer has, and one of the two never had it #406 and neither is edited here.
• Any rule, fixture, threshold, expression or setting of either leg. No run
  behaves differently because of this change. Nothing here adds or edits a guard,
  so this pull request has no section showing one bite.
• The same sentence in
  docs/decisions/0113-the-target-triples-the-gate-compiles-for.md. It is prose
  in a landed record, and whether such a sentence going false because the tree
  moved is inside 0269's subject is the decision 0001 says no check reads a decision record and pastes five workflow files, and a check over docs/decisions/ has run since 2026-09-02 #380 is waiting on. I moved none
  of it.
• Any other claim in either file. I read the paragraph that explains the
  emptiness. That is a sweep of one paragraph rather than of the two files.
• The other legs' verdicts. I ran the two selftest verbs above on this
  machine and nothing else. Whether the rest of the gate passes is what this pull
  request's own runs report, and I have not predicted them here.

Who has read it

Nobody other than the author has read this change. There is no second reader on
this board tonight, and the evidence above stands in place of one rather than the
question being left open.