A session has a number: given at creation, frozen, never reused (#760, D86, lane 4) #807

Merged

claude-bot merged 6 commits from build/760-session-number into main 2026-09-24 06:45:03 +00:00

Conversation 0 Commits 6 Files changed 60 +1938 -166

claude-bot commented 2026-09-24 06:16:37 +00:00

Collaborator

Copy link

Closes #760. This is lane 4's first server PR (#730). D86: a number, given when the session is created, frozen and never reused; a cancelled session keeps its number, so the sequence has gaps.

Server

• Migration f3a4b5c6d7e8 (on e1f2a3b4c5d6, single head):

  • Adds campaigns.next_session_number and sessions.number (NOT NULL) with UNIQUE (campaign_id, number).
  • Backfills in created_at order, as #760 asks. A tie, such as an import's shared created_at, is broken by when the session took place. Cancelled sessions are numbered.
  • The docstring says the backfill is a reconstruction, not a record.
  • It is re-runnable, and the downgrade drops everything.

• A counter, not max + 1, so a deleted newest session's number is never reused. One UPDATE … RETURNING in the creating transaction, whose row lock serialises concurrent creates. session_numbering.allocate_session_number is called by:

  • create_session (and the one-shot);
  • the import;
  • series materialisation, which skips existing occurrences before allocating, so its hourly run doesn't burn numbers.

• A before_flush safety net numbers any other ORM-created session from the same counter. That covers a future creation path, the eval seed and test fixtures.

• Import and export: export writes number. Import keeps valid, unclaimed numbers, numbers the rest in the backfill's order, and leaves the counter above them all.

• number is exposed on:

  • the session and session-list responses;
  • This week;
  • Home and Needs you (session_number);
  • the live endpoint;
  • search;
  • the cold open;
  • series sessions;
  • the timeline;
  • notifications.

  It is also on /api/bot/*, additively: no contract bump, and the bot is untouched.

The UI: "Session 14" where the drawing is tight

Everything goes through sessionLabel() in vocabulary.js:

• the breadcrumb and phone title, and the tab title (Session 14 · Editing it · The Amberwake Accord · Quest Board);
• the Contents sub-head;
• the live bar;
• the bell's sentences ("Session 14 is on: Mon 22 Sep, 8:00 PM");
• Home's Needs you and campaign card ("Last time: Session 13, “The Drowned Halls”");
• This week (label This week · Session 14 above the title heading);
• the Sessions list (a number column; the title stays the row's name and link).

Journeys

• The fixture titles lose their "Session N: " prefixes, and the seed checks the server's numbers (1/2/3).
• J0, J1, J2 and J6 now assert the real numbers. J6's "there is no numbering (#760)" becomes This week · Session 3.

Checks

• Backend, full suite: 4172 passed, 13 skipped.
• Frontend: 144 files, 2027 tests. Lint: 0 errors, and the warning count is unchanged at 228.
• Journeys: 27 passed and 3 skipped (J2's lane-4 half, by design), on desktop, phone and keyboard.
• The migration was run for real on seeded data: upgrade, downgrade, upgrade. The numbers came out as expected and the counters were right.
• After merging main (#805, #806): ruff clean, and the numbering, erasure, migration and campaign tests pass (178).

Not in this PR

• The Storyline page, the Search panel's rows and MergeNotes' options don't draw the number yet; they're outside D86's places.
• docs/API.md doesn't list the new fields.

Lane 4's page PRs pick these up.

🤖 Generated with Claude Code

Closes #760. This is lane 4's first server PR (#730). D86: a number, given when the session is created, frozen and never reused; a cancelled session keeps its number, so the sequence has gaps. ## Server - **Migration `f3a4b5c6d7e8`** (on `e1f2a3b4c5d6`, single head): - Adds `campaigns.next_session_number` and `sessions.number` (NOT NULL) with `UNIQUE (campaign_id, number)`. - Backfills in `created_at` order, as #760 asks. A tie, such as an import's shared `created_at`, is broken by when the session took place. Cancelled sessions are numbered. - The docstring says the backfill is a reconstruction, not a record. - It is re-runnable, and the downgrade drops everything. - **A counter, not `max + 1`,** so a deleted newest session's number is never reused. One `UPDATE … RETURNING` in the creating transaction, whose row lock serialises concurrent creates. `session_numbering.allocate_session_number` is called by: - `create_session` (and the one-shot); - the import; - series materialisation, which skips existing occurrences before allocating, so its hourly run doesn't burn numbers. - **A `before_flush` safety net** numbers any other ORM-created session from the same counter. That covers a future creation path, the eval seed and test fixtures. - **Import and export:** export writes `number`. Import keeps valid, unclaimed numbers, numbers the rest in the backfill's order, and leaves the counter above them all. - **`number` is exposed** on: - the session and session-list responses; - This week; - Home and Needs you (`session_number`); - the live endpoint; - search; - the cold open; - series sessions; - the timeline; - notifications. It is also on `/api/bot/*`, additively: **no contract bump**, and the bot is untouched. ## The UI: "Session 14" where the drawing is tight Everything goes through `sessionLabel()` in `vocabulary.js`: - the breadcrumb and phone title, and the tab title (`Session 14 · Editing it · The Amberwake Accord · Quest Board`); - the Contents sub-head; - the live bar; - the bell's sentences ("Session 14 is on: Mon 22 Sep, 8:00 PM"); - Home's Needs you and campaign card ("Last time: Session 13, “The Drowned Halls”"); - This week (label `This week · Session 14` above the title heading); - the Sessions list (a number column; the title stays the row's name and link). ## Journeys - The fixture titles lose their "Session N: " prefixes, and the seed checks the server's numbers (1/2/3). - J0, J1, J2 and J6 now assert the real numbers. J6's "there is no numbering (#760)" becomes `This week · Session 3`. ## Checks - Backend, full suite: 4172 passed, 13 skipped. - Frontend: 144 files, 2027 tests. Lint: 0 errors, and the warning count is unchanged at 228. - Journeys: 27 passed and 3 skipped (J2's lane-4 half, by design), on desktop, phone and keyboard. - The migration was run for real on seeded data: upgrade, downgrade, upgrade. The numbers came out as expected and the counters were right. - After merging main (#805, #806): ruff clean, and the numbering, erasure, migration and campaign tests pass (178). ## Not in this PR - The Storyline page, the Search panel's rows and MergeNotes' options don't draw the number yet; they're outside D86's places. - `docs/API.md` doesn't list the new fields. Lane 4's page PRs pick these up. 🤖 Generated with [Claude Code](https://claude.com/claude-code)

claude-bot added 5 commits 2026-09-24 06:16:38 +00:00

feat(backend): a session has a number, given at creation and never reused (#760, D86) ... 8aae853572

`sessions.number` (NOT NULL, UNIQUE per campaign) and
`campaigns.next_session_number`, in migration f3a4b5c6d7e8 on top of
e1f2a3b4c5d6. The backfill numbers every session, cancelled ones
included, by created_at and then by when it was (confirmed time, else
its earliest proposed time) to break an import's tie, and says it is a
reconstruction rather than a record.

Numbers come from a counter, not max + 1, so deleting the newest session
never hands its number on: one UPDATE ... RETURNING in the creating
transaction, whose row lock serialises concurrent creates.
`allocate_session_number` is called by create_session (and so the
one-shot), the import and series materialisation; a before-flush hook
numbers any other ORM insert from the same counter. Materialisation asks
which occurrences exist before it allocates, so the hourly run burns
nothing.

Export writes the number; import keeps a usable one and numbers the rest
after it in the backfill's order. The number is on the session detail
and list, Home, This week, Needs you, the live endpoint, both searches,
the cold open, a series' sessions, the timeline and the bell, whose rows
now say "Session 14". Additive on /api/bot/*: no contract bump.

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>

feat(webapp): "Session 14" where the drawing is tight (#760, D86) ... 10a4661e63

`sessionLabel(session)` in vocabulary.js: "Session N", the title only when
the number is missing (an old cached response), else "Session".

- The crumb and the tab: a session page declares its session with its
  campaign, and the frame names it "Session 14" (shelf, table and edit
  alike); "Session" until the page has said which.
- The Contents: on a session's own pages Sessions is current, with the
  session as its sub-head. Contents now draws a sub-head on its own.
- The live bar reads through sessionLabel.
- Home: Needs you rows name the session by number beside the campaign;
  a card's This week leads with it; Last time reads
  Session 13, "The Drowned Halls".
- This week: the title stays the heading, with "This week · Session 14"
  above it; an untitled session is headed by its number; Last time and
  the table card name it by number.
- Sessions list: the drawing's leading "S14" column is back, read out as
  "Session 14"; an untitled row reads "Session 15 · no title yet".

The bell's rows are the server's sentences, which now say "Session 14".

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>

test(e2e): the journeys read the session's real number (#760, D86) ... 72e9d42725

The fixture titles carried "Session 12: " and so on because nothing
numbered a session. Now the product does, so the prefixes go (they would
read "Session 1 · Session 12: The Salt Marshal") and identities.json
carries the numbers the seed's three sessions get in a fresh campaign —
1, 2 and 3 — which the seed checks against what the server said.

J0 and J2 find the upcoming session on Home's card, in Needs you and in
the bell by its number; J6 turns "there is no numbering" into the label
"This week · Session 3" and reads Last time as number and title; J1
asserts its fresh campaign's sessions are Session 1 and Session 2, on
This week and in the Sessions list.

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>

docs(changelog): a session has a number (#760, D86) ... c62fa3c888

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>

Merge remote-tracking branch 'origin/main' into build/760-session-number c6f4d3b24a

rbrooks added 1 commit 2026-09-24 06:25:04 +00:00

ci: re-run after the runner's Docker daemon dropped mid-job ... 070250d738

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

claude-bot merged commit cfdb2ddcab into main 2026-09-24 06:45:03 +00:00

claude-bot referenced this pull request from a commit 2026-09-24 06:45:05 +00:00

Merge PR #807: a session has a number — given at creation, frozen, never reused (#760, D86, lane 4)

claude-bot referenced this pull request 2026-09-24 06:47:15 +00:00

[Backend][UX] A session has a number (D86): given at creation, frozen, gaps allowed #760

claude-bot referenced this pull request 2026-09-24 07:18:16 +00:00

Corrections that wait, and every version of a summary (D31, IS-379/D51, lane 4) #808

claude-bot referenced this pull request 2026-09-24 08:02:42 +00:00

Scheduling: a move re-asks everyone and carries the series with it; pencilled in → a vote; why it was cancelled (D99, D100, D101, #804, lane 4) #809

claude-bot referenced this pull request 2026-09-24 08:20:36 +00:00

Build lane 4: sessions (SESS-1 to SESS-7) #730

claude-bot referenced this pull request from a commit 2026-09-24 15:23:03 +00:00

fix(backend): the transcript_failed migration takes an unused id (D96, #730)

claude-bot referenced this pull request 2026-09-25 17:22:40 +00:00

Build lane 4: sessions (SESS-1 to SESS-7) #730

Sign in to join this conversation.

Reviewers

No reviewers

Labels

Clear labels   

backend

Issue related to the backend

  

bot

Issue related to the bot

  

bug

Something is not working

  

duplicate

This issue or pull request already exists

  

enhancement

New feature

  

frontend

Issue related to the frontend

  

help wanted

Need some help

  

invalid

Something is wrong

  

question

More information is needed

  

recording

Issue related to the recording system

  

security

Security hardening / vulnerability work

  

wontfix

This won't be fixed

No labels

backend

bot

bug

duplicate

enhancement

frontend

help wanted

invalid

question

recording

security

wontfix

Milestone

Clear milestone

No items

No milestone

Projects

Clear projects

No items

No project

Assignees

Clear assignees

No assignees

2 participants

Notifications

Subscribe

Due date

The due date is invalid or out of range. Please use the format "yyyy-mm-dd".

No due date set.

Dependencies

No dependencies set

Reference

rbrooks/Quest-Board!807

No description provided.