session-history/ohm/0083/SESSION-0083.0-TRANSCRIPT-2026-06-06T01-49--2026-06-06T07-59.md

# Session 0083.0 — Transcript

> App: ohm
> Start: 2026-06-06T01-49 (PST)
> End: 2026-06-06T07-59 (PST)
> Type: coding
> Mode: autonomous (yolo)
> Driver: Ben Stull (+ Claude Opus 4.8)

## Launch prompt

```
Finish §22 S6 remainder: request-to-join + cross-collection inbox (§22.8), then
per-type surfaces (§22.4a items 1&3). Build by analogy to shipped §28
contribution_requests + S4 memberships.grant.
```

## Outcome

The §22 S6 remainder is complete; the deployment→project→collection three-tier
model is now realized in both code and `SPEC.md`. Then two follow-up tasks at the
operator's direction: a roadmap question, and a roadmap cleanup.

1. **§22.8 request-to-join + the cross-collection inbox — shipped v0.46.0**
   (PR #24, `origin/main` = `1c17fec`).
2. **Per-type surfaces (§22.4a items 1 & 3) — spec pass done** (PR #25,
   `origin/main` = `b7e23a0`) — *subsequently superseded* (see §Follow-ups).
3. **OHM roadmap Phase G refreshed** (ohm-rfc PR #12, `b4a6943`).

## Pre-state

- `origin/main` (rfc-app) = `0140150` (S6 core, v0.45.0): the SPEC merge,
  per-collection `enabled_models` (mig 031), the type-driven entry noun.
- Two S6 items deferred at v0.45.0 as "wanting a discovery/spec pass first": the
  per-type surfaces (§22.4a items 1 & 3) and request-to-join (§22.8).
- Clean tree on `main`, fully pushed. Three concurrent ohm sessions
  (0079/0080/0081) in flight (unfinalized placeholders) — noted, proceeded.

## The arc — §22.8 (the shipped feature)

**1. Understand.** Read `SPEC.md` §22.8 / §22.11 and the design-doc Part E S6
framing; mapped the two analogues — §28 `contribution_requests`
(`api_contributions.py`, mig 024) and S4 `memberships` (`api_memberships.py` /
`memberships.py`, the `grant` + broader-scope-supersedes prune + the
`can_invite_at_{project,collection}` gates). Engine seam: a join request targets
a `(scope_type, scope_id)` pair (the `memberships` vocabulary minus `global`),
and accept grants a scope role instead of minting an RFC invite.

**2–3. Build (backend + frontend).**
- Migration `032_join_requests.sql` (additive; one-open-per-(scope,requester)).
- `auth.effective_role_at_scope` — scope-grain twin of `effective_scope_role`,
  folding global→project for a project target.
- `notify`: `_scope_owner_user_ids` (upward Owner enumeration = the
  cross-collection inbox), `fan_out_join_request`, `notify_join_decided`, 3
  `render_summary` cases.
- `api_join_requests.py` — `/api/scopes/{scope_type}/{scope_id}/` (join-target /
  request / accept / decline). The request POST is **not** readability-gated
  (that's how one joins a *gated* scope). Registered in `api.py`.
- Frontend: `api.js` verbs; `JoinRequestModal`; `viewer.can_request_join` on the
  project + collection viewer blocks; "Request to join" in `CollectionDirectory`
  + the `Catalog` footer; the actionable `JoinRequestRow` in `Inbox.jsx`.

**4. Tests + release.** `test_join_requests_vertical.py` (11) +
`test_migration_032` (5); frontend `api.joinrequests.test.js` (4) + two
`CollectionDirectory` cases. **Backend 546 / frontend 36 green; build clean.**
VERSION + package.json → 0.46.0; CHANGELOG 0.46.0 ("no upgrade steps"). Branch
`feat/s6-join-requests` → Gitea-API PR #24 → merged.

## The arc — per-type surfaces (spec pass, then superseded)

The goal flagged these as wanting a discovery/spec pass first. Wrote
`docs/design/2026-06-06-per-type-surfaces.md` (per-type frontmatter schema via a
new `entry_schema.py` validator; `specification` Releases view; `bdd`
scenario/coverage views; BDD scenarios `@S7a/@S7b/@S7c`; a 3-slice plan).
Branch `spec/s7-per-type-surfaces` → PR #25 → merged.

**Superseded same day.** A parallel session (0079–0081) was concurrently writing
`docs/design/2026-06-06-configurable-collection-metadata.md`, which **explicitly
supersedes** the per-type-surfaces draft (its D9/D11): the schema model moves from
*type-driven* to **collection-configured `fields:`** + per-entry `.meta.yaml`
**sidecars** + faceted filtering. So the per-type-surfaces direction (and the
`Next /goal: implement S7a` I recorded) is now void; the successor's SLICE-0
will mark `per-type-surfaces.md` superseded + amend SPEC §22.4a.

## Follow-ups (post the first finalize)

The session was finalized once at 02:17, then continued at the operator's
direction:

- **Q: "what are all items on the roadmap?"** Read the canonical
  `ben.stull/ohm-rfc/ROADMAP.md` (43 numbered items, Phases A–G + a version-targets
  table). Reported: ~32/43 shipped; what remains is item 22 (consent copy,
  operator-blocked), the §22 per-type surfaces (then being reworked), and the
  v1-gated deploy-safety arc (37–43), plus operational cleanup (33–35). Flagged
  the **Phase G section as stale** (didn't reflect S1–S6 or §22.8 shipping).

- **Roadmap cleanup (ohm-rfc PR #12, merged `b4a6943`).** Per the operator —
  *"there won't be types; there will be configurable metadata per collection …
  do NOT add to roadmap yet, but clean up based on it being the most up-to-date
  plan."* Four surgical Phase-G edits, **no new items added**, items 1–43 +
  version-targets table untouched: (1) a 2026-06-06 status note (series continued
  past v0.39.0 → three-tier, re-sliced S1–S7); (2) slice-status refresh (S1–S6 +
  §22.8 shipped; old M6 = invite + request-to-join; old M7 = enabled_models +
  SPEC merge; M4 effectively covered by ohm+ecomm + the two-project test); (3)
  old **M5 → SUPERSEDED** by configurable collection metadata (successor draft
  *in flight, not yet roadmapped*; bdd coverage = future `ref` surface); (4)
  cutover block fixed (0.31.5→0.39.0 cutover done, OHM pinned 0.39.0; new gap =
  v0.40–v0.46 undeployed). Opened for review, then merged on operator request.

## Cut state

- **rfc-app:** my work fully landed — `origin/main` = `b7e23a0` (v0.46.0 + the
  per-type-surfaces doc), branches deleted. ⚠️ At finalize the shared checkout's
  `main` was **ahead-2 + untracked `corpus-tree` spec files** — that is the
  **parallel session's** WIP (the §5.4 worktree hazard); **left entirely
  untouched** (not mine to commit/push).
- **ohm-rfc:** clean on `main` = `b4a6943` (roadmap PR #12 merged), branch deleted.
- VERSION 0.46.0. Backend 546 / frontend 36 green.

## Deferred decisions / autonomous calls

- **§22.8 request POST not readability-gated** — deliberate (§22.8: you join a
  *gated* scope you were told of). Confident.
- **`join_requests.scope_type` excludes `global`** — a deployment isn't "joined."
- **Route shape `/api/scopes/{scope_type}/{scope_id}/…`** — §17 is illustrative,
  doesn't pin these routes; no SPEC change.
- **Per-type surfaces delivered as a spec doc, not code** — the goal said "want a
  spec pass first." (Now moot — superseded by configurable metadata.)
- **Declined to finalize session 0081** when a finalize-on-goal hook misfired onto
  it — 0081 is a *different* (live/parallel) discovery session; finalizing it
  would fabricate or clobber another driver's record. Left 0079/0080/0081 alone;
  operator later confirmed ≥1 is the live session they're co-writing with.
- **Roadmap PR opened for review (not auto-merged)** then merged on operator's
  "Merge it" — the supersession wording was theirs to steer.

## Next session

The per-type-surfaces plan is superseded; the next move is gated on the parallel
session's configurable-collection-metadata spec settling.

```
/goal await the configurable-collection-metadata spec landing (parallel session finishing docs/design/2026-06-06-configurable-collection-metadata.md), then implement SLICE-0 — amend SPEC §22.4a so the entry schema is collection-configured .meta.yaml sidecar fields not type-driven frontmatter; defer item-3 type surfaces (record bdd coverage as a future ref surface); mark 2026-06-06-per-type-surfaces.md superseded; §20 changelog (doc-only) — then SLICE-1 (sidecar storage + dual-read parser + idempotent frontmatter→sidecar migration + malformed-metadata catalog flag). NOTE: that spec is parallel-owned and shares this checkout — use an isolated worktree (§5.4) before starting code.
```