++

Author: subtleGradient

.tasks/done/TASK-037-zig-sqlite-upstream-feedback-blocked.md

@@ -0,0 +1,45 @@
+# TASK-037: zig-sqlite Upstream Feedback Capture (blocked)
+
+## Status
+- [x] Planned
+- [ ] Assigned
+- [ ] In Progress
+- [ ] Blocked
+- [x] Complete (cancelled per Tom)
+
+## Priority
+low
+
+## Assigned To
+subagent (explore)
+
+## Parent Docs / Cross-links
+- Wish: `.wishes/gather-upstream-feedback.md`
+- Zig reference analysis: `research/zig-cr/20-zig-sqlite-capabilities.md`
+- Candidate touchpoints in our code: `zig/src/sqlite/value.zig`, `zig/src/sqlite/vtab.zig`
+
+## Description
+Capture specific, actionable upstream feedback ideas for `.refs/zig-sqlite`.
+
+Per `.wishes/gather-upstream-feedback.md`, these ideas should be recorded as individual markdown cards under `.wishes/blocked-on-tom/` so Tom can later choose what (if anything) to contribute upstream.
+
+This task is blocked because it requires Tom to confirm the intended scope (how many cards, and whether we’re allowed to touch `.refs/` at all).
+
+Decision-capture task:
+- `.tasks/backlog/TASK-055-tom-scope-upstream-feedback.md`
+
+## Files to Modify
+- `.wishes/blocked-on-tom/*.md` (new idea cards)
+- `research/zig-cr/92-gap-backlog.md` (optional link section)
+
+## Acceptance Criteria
+- [ ] At least 5 concrete upstream idea cards created.
+- [ ] Each card links to the exact file/line area in our repo motivating it.
+- [ ] Each card includes a "why it matters" section.
+
+## Progress Log
+### 2025-12-14
+- Task card created; blocked pending Tom direction
+
+## Completion Notes
+**2025-12-17**: Cancelled per Tom — "skip all zig-sqlite stuff"

.tasks/done/TASK-055-tom-scope-upstream-feedback.md

@@ -0,0 +1,39 @@
+# TASK-055: Tom decision — Scope zig-sqlite upstream feedback capture
+
+## Status
+- [x] Planned
+- [ ] Assigned
+- [ ] In Progress
+- [ ] Blocked
+- [x] Complete (cancelled per Tom)
+
+## Priority
+low
+
+## Assigned To
+Tom
+
+## Parent Docs / Cross-links
+- Blocked task: `.tasks/backlog/TASK-037-zig-sqlite-upstream-feedback-blocked.md`
+- Wish: `.wishes/gather-upstream-feedback.md`
+- Gap backlog: `research/zig-cr/92-gap-backlog.md`
+
+## Description
+TASK-037 is blocked on scope from Tom. Tom’s current direction: this is low priority until our own product work is done; upstream feedback capture may happen later, if ever.
+
+## Files to Modify
+- `.tasks/backlog/TASK-037-zig-sqlite-upstream-feedback-blocked.md`
+
+## Acceptance Criteria
+- [ ] Update `.tasks/backlog/TASK-037-zig-sqlite-upstream-feedback-blocked.md` with:
+  - Number of upstream idea cards desired (e.g., 3 / 5 / 10)
+  - Whether we may reference `.refs/zig-sqlite` files directly in the cards
+  - Whether we’re allowed to patch `.refs/zig-sqlite` locally for experiments (still no upstream PR)
+  - Any “must include” topics
+
+## Progress Log
+### 2025-12-15
+- Task created from Round 33 “blocked on scope” status.
+
+## Completion Notes
+**2025-12-17**: Cancelled per Tom — "skip all zig-sqlite stuff"

.wishes/done/gather-upstream-feedback.md

@@ -0,0 +1,17 @@
+.refs/zig-sqlite is a good project
+let's help them
+gather a bunch of ideas for how to make zig-sqlite better
+add them as separate cards to .wishes/blocked-on-tom/*.md
+cross-link them to the relevant places in our codebase
+e.g. zig/src/sqlite/value.zig
+
+eventually, Tom may assign someone to contribute those fixes back upstream
+but for now, let's just focus on capturing these things as potential work way may consider doing
+
+---
+
+## Cancelled
+
+- **Date**: 2025-12-17
+- **Decision**: Per Tom — "skip all zig-sqlite stuff"
+- **Tasks cancelled**: TASK-055, TASK-037

.wishes/done/scratchpad.md

@@ -0,0 +1,41 @@
+i wish my scratchpad playgrounds were all wired up
+
+## scratch/browser-scratchpad ✅
+
+Done (Round 40, 2025-12-16)
+- `bun --hot scratch/browser-scratchpad/src/index.ts`
+- React app with SharedWorker multi-tab coordination
+- Shows cross-tab item sync, provider election, db_version tracking
+
+## scratch/bun-scratchpad ✅
+
+Done (Round 40, 2025-12-16)
+- `bun run scratch/bun-scratchpad/index.ts`
+- CR-SQLite demo with bun:sqlite + extension loading
+- Shows CRR tables, CRUD, db_version, crsql_changes, site_id
+
+## scratch/effect-bun-scratchpad ⏸️
+
+Blocked on Tom / spec-gate (TypeScript work must go through effect-native/.specs/)
+See: `.wishes/blocked-on-tom/effect-bun-scratchpad.md`
+
+should be a simple Effect-TS project with bun
+using @effect/platform @effect/platform-bun @effect/sql @effect/sql-sqlite-bun
+using the @effect/language-service
+see .refs/effect/tsconfig.base.json
+using @effect/vitest @effect/eslint-plugin
+see .refs/effect/package.json
+using all our new effect-native packages
+
+should have some kind of realistic example that does something neat
+
+---
+
+## Satisfied
+
+- **Date**: 2025-12-17
+- **What changed**: Bun + browser scratchpads wired up in Round 40
+- **Evidence**: 
+  - `bun run scratch/bun-scratchpad/index.ts` — shows full CR-SQLite demo
+  - `bun --hot scratch/browser-scratchpad/src/index.ts` — React multi-tab demo
+- **Remaining**: Effect-TS scratchpad blocked on spec-gate (documented in `.wishes/blocked-on-tom/effect-bun-scratchpad.md`)

.wishes/done/small-prod-builds.md

@@ -0,0 +1,16 @@
+the old libcrsqlite.{dylib,so,dll} build(s) of the old rust cr-sqlite implementation was chonkier than sqlite itself
+i believe this is because rust includes some about of runtime stuff that gets included in the build
+i expect that a properly optimized build of our zig-based re-implementation should be significantly smaller than sqlite itself
+if our zig build is big, that smells like we screwed something up somehow
+that, or I'm confused about something
+
+---
+
+## Satisfied
+
+- **Date**: 2025-12-17
+- **What changed**: Size report implemented in Round 38 (TASK-068)
+- **Evidence**: 
+  - `make -C zig size-report` shows: Zig crsqlite is 105.72% of SQLite (~103KB overhead)
+  - This is dramatically smaller than the Rust build (which was ~2-3x SQLite)
+- **Conclusion**: Zig build is healthy — extension overhead is minimal (~6% over baseline SQLite)