Skip to content

DAILY_SUMMARY_TEMPLATE.md

Latest commit

 

History

History
92 lines (61 loc) · 2.67 KB

DAILY_SUMMARY_TEMPLATE.md

File metadata and controls

92 lines (61 loc) · 2.67 KB
Raw

Work-session summary template

Use this checklist at the end of a substantive Sharp Runtime session. Do not prepend another chronological diary to NEXT.md; update its existing snapshot, recent-changes, limitations, and next-task sections in place. Git history preserves older handoffs.

Durable detail belongs in three places:

  • The affected ticket.notes row in local plan.sqlite3.
  • A focused commit message with the ticket number and validation result.
  • NEXT.md only when the cold-start facts or ordering actually changed.

Keep the summary factual: exact commits, test counts, graph counts, toolchain versions, and confirmed behavior. Distinguish a fixed bug from a clean audit or a deliberately deferred gap.

## Session result — YYYY-MM-DD: <short theme>

**Branch/code baseline:** `feature/work` at `<short-hash>`

**Validation:**

- Module graph: <N> physical modules, <N> production edges
- Build: <0 warnings, 0 errors or exact blocker>
- Tests: <N> passed across <N> executables
- Extra checks: <selective matrix / TSan / ASan / UBSan / cross-build / none>

**Ticket state:**

- Ticket #<N>: `<before-status>` → `<after-status>`
- Queue after the session: <todo/doing/blocked/needs_user counts>

**Behavior changed:**

- <What was wrong, the relevant .NET reference path, what changed, regression
  test, and commit hash.>

**Audited with no behavior change:**

- <What was checked and the evidence for a clean result.>

**Deferred or newly discovered:**

- <Exact gap, why it was not included, and the proposed next bounded step.>

**Documentation updated:**

- <README/NEXT/plan/component catalogue/module README, or “not needed”.>

**Recommended next task:**

- <One bounded task and why it is next.>

End-of-session checks

Run the full local gate:

scripts/local_ci_check.sh build

If component metadata changed, also run the complete selective matrix:

scripts/check_selective_components.sh

Then verify the queues:

sqlite3 plan.sqlite3 \
  "SELECT status, priority, COUNT(*) FROM ticket GROUP BY status, priority ORDER BY priority, status;"

sqlite3 plan.sqlite3 \
  "SELECT status, COUNT(*) FROM task GROUP BY status ORDER BY status;"

Update NEXT.md when any of these facts changed:

  • Code baseline or verified test count.
  • Component/module/edge count.
  • CI or platform evidence.
  • A known limitation or permanent scope decision.
  • The ordered next-task list.

Do not copy routine command output into NEXT.md; record only the measured result. If a test failed because the execution environment denied local networking, distinguish that from a product failure and rerun in an environment that permits the suite's socket/HTTP/ping tests.