[TimelordUK]

Marks the 46 unreviewed info/ transcripts from #494 as agreed. Only the status lines change.

I read each one against the contract (#503, and option 1 for failed statements):

• A PRINT, or a raiserror below 11, is a result of its own: info, then rowcount 0, after the previous result's rowcount and callback. That holds 60 times over in many-prints*, and inside a procedure in proc-prints.
• A warning raised by a result's rows (8153, 3606, 3607) comes after those rows and before that result's callback. With 120 rows the warning lands on the last fetch batch, and a pause mid-batch or on the last row doesn't lose it.
• Info, then an error, then a result keep that order. A failed statement mid-batch gives rowcount 0.
• fk-on-temp-table-cached: no 1756 on a cached plan, because the server doesn't send it again (see HANDOVER 2026-10-22).

Locally, the info/ contract tests pass: 46 passing.

Independent of #504 and #505, which touch different files.

🤖 Generated with Claude Code

[TimelordUK]

I've also agreed scripts/prepared-info (the owner's call). It's recorded as a limitation: a prepared execution's info (8153) comes before its meta, because the driver hands it over with SQLExecute. Each execution still reports only its own info, once. With this, every transcript in the corpus is agreed or known-bug.