Summary
§10.3 says an Attested Computation may carry its computation inline as "a single fenced code block in the body under # Computation." But §10.2's own worked example puts the computation in a 4-space-indented code block, not a fence:
# Computation
SELECT SUM(amount) AS revenue
FROM finance.recognized_revenue
WHERE fiscal_year = @year
A consumer that detects the inline computation by looking for a fenced block under # Computation — the literal §10.3 wording — would miss the spec's own example, and for the §10.3 "inline or a computation: path" rule would wrongly conclude the computation is provided in neither place.
The question
Should §10.3 read "a fenced or indented code block," or should §10.2's example be rewritten as a fenced block? Either resolves it — the mismatch between the normative text and the worked example is the problem.
Context
Same shape as #234 (the actor convention): the normative text and a worked example disagree, so a strict reading of one flags the other. This surfaced building the same structural validator; we currently key the inline-vs-path check on the presence of the # Computation heading rather than the code block's style, so both forms pass.
Summary
§10.3 says an Attested Computation may carry its computation inline as "a single fenced code block in the body under
# Computation." But §10.2's own worked example puts the computation in a 4-space-indented code block, not a fence:A consumer that detects the inline computation by looking for a fenced block under
# Computation— the literal §10.3 wording — would miss the spec's own example, and for the §10.3 "inline or acomputation:path" rule would wrongly conclude the computation is provided in neither place.The question
Should §10.3 read "a fenced or indented code block," or should §10.2's example be rewritten as a fenced block? Either resolves it — the mismatch between the normative text and the worked example is the problem.
Context
Same shape as #234 (the actor convention): the normative text and a worked example disagree, so a strict reading of one flags the other. This surfaced building the same structural validator; we currently key the inline-vs-path check on the presence of the
# Computationheading rather than the code block's style, so both forms pass.