docs(target-extensible-ir-namespaces): consolidate umbrella + replan S1#549
Merged
Conversation
Contributor
|
Caution Review failedThe pull request is closed. ℹ️ Recent review info⚙️ Run configurationConfiguration used: Path: .coderabbit.yml Review profile: CHILL Plan: Pro Run ID: ⛔ Files ignored due to path filters (22)
📒 Files selected for processing (4)
📝 WalkthroughWalkthroughAdds two drive-plan-project trial findings (boundary and win) to drive/retro/findings.md, rewrites drive-discussion's invocation description, and removes disable-model-invocation flags from two skill metadata files to enable orchestrator-driven invocation. ChangesTrial Findings and Orchestration Skill Refinement
Estimated code review effort🎯 2 (Simple) | ⏱️ ~10 minutes Possibly related PRs
Poem
🚥 Pre-merge checks | ✅ 5✅ Passed checks (5 passed)
✏️ Tip: You can configure your own custom pre-merge checks in the settings. ✨ Finishing Touches🧪 Generate unit tests (beta)
Comment |
@prisma-next/extension-author-tools
@prisma-next/mongo-runtime
@prisma-next/family-mongo
@prisma-next/sql-runtime
@prisma-next/family-sql
@prisma-next/extension-arktype-json
@prisma-next/extension-cipherstash
@prisma-next/middleware-cache
@prisma-next/mongo
@prisma-next/extension-paradedb
@prisma-next/extension-pgvector
@prisma-next/extension-postgis
@prisma-next/postgres
@prisma-next/sql-orm-client
@prisma-next/sqlite
@prisma-next/target-mongo
@prisma-next/adapter-mongo
@prisma-next/driver-mongo
@prisma-next/contract
@prisma-next/utils
@prisma-next/config
@prisma-next/errors
@prisma-next/framework-components
@prisma-next/operations
@prisma-next/ts-render
@prisma-next/contract-authoring
@prisma-next/ids
@prisma-next/psl-parser
@prisma-next/psl-printer
@prisma-next/cli
@prisma-next/emitter
@prisma-next/migration-tools
prisma-next
@prisma-next/vite-plugin-contract-emit
@prisma-next/mongo-codec
@prisma-next/mongo-contract
@prisma-next/mongo-value
@prisma-next/mongo-contract-psl
@prisma-next/mongo-contract-ts
@prisma-next/mongo-emitter
@prisma-next/mongo-schema-ir
@prisma-next/mongo-query-ast
@prisma-next/mongo-orm
@prisma-next/mongo-query-builder
@prisma-next/mongo-lowering
@prisma-next/mongo-wire
@prisma-next/sql-contract
@prisma-next/sql-errors
@prisma-next/sql-operations
@prisma-next/sql-schema-ir
@prisma-next/sql-contract-psl
@prisma-next/sql-contract-ts
@prisma-next/sql-contract-emitter
@prisma-next/sql-lane-query-builder
@prisma-next/sql-relational-core
@prisma-next/sql-builder
@prisma-next/target-postgres
@prisma-next/target-sqlite
@prisma-next/adapter-postgres
@prisma-next/adapter-sqlite
@prisma-next/driver-postgres
@prisma-next/driver-sqlite
commit: |
wmadden
changed the title
…personas, github-review-iteration
Remove disable-model-invocation gate on three skills that already had clear
trigger language but were unnecessarily blocked from auto-invoke.
drive-discussion gets a rewritten description that names legitimate
auto-trigger contexts (pre-spec, mid-spec, falsified-assumption, obstacle,
explicit operator request) and the cases where it must NOT fire (simple
clarifying questions, implementation work, pressure-testing finished
artefacts). The old "Use ONLY when the user explicitly invokes this skill /
Never auto-invoke" line was contradicted by the body's own trigger list,
which already named pre-spec / mid-spec / mid-flight as legitimate
auto-trigger contexts and noted that workflow skills route here on
triggers 3/4.
drive-agent-personas: persona library is a load-time resource for other
skills; nothing about it requires gating on explicit invocation.
github-review-iteration: description already names the trigger phrases
("address PR review", "triage review comments", "iterate until review is
clean"). The manual gate added no value.
Signed-off-by: Will Madden <madden@prisma.io>
… branch Pull the existing project spec + ADR 0001 from origin/contract-ir-planes onto the planning branch so the S1 design (TML-2584) can be rewritten in-place rather than coordinated across two branches. The contract-ir-planes branch will be deleted after the rewrite lands. The imported content is not the final state — scope-cuts agreed in this planning round move enums out of the framework-shared types slot and defer RLS / TML-2537 substrate work to separate projects. Rewrite happens next under drive-discussion + drive-specify-project + drive-plan-project. Signed-off-by: Will Madden <madden@prisma.io>
…tity coordinate Rewrite of projects/contract-ir-planes/spec.md absorbing the design discussion (architect + principal-engineer lenses) that pressure-tested the target-pack-contributed entity-kind mechanism and sized the structural refactor against drive's slice budget. Key decisions captured: - D1-D4 kept from the previous spec (two-plane shape, uniform indexing, domain/storage names, object-pair cross-references) - D5 rewritten to front-load the descriptor mechanism: target packs contribute new entity kinds via AuthoringContributions.entityTypes; framework-shared storage.<ns>.types slot deleted as a load-bearing surface; framework-known built-ins (tables, collections) stay hardcoded as deliberate paid debt - D6 new: entity coordinate (namespaceId, entityKind, entityName) becomes the canonical addressing primitive across migration disjoint calculation, planner diffing, validator collision checks, and cross-plane references - D8 new: Tier 1 scope discipline. First-class user affordances (typed Role.member references, db.enums.X surface, codec value-narrowing) live in postgres-enum-finishing; this project ships only the structural relocation that satisfies PDoD3 Added to align with the modern drive-specify-project template: - Functional + Non-Functional Requirements - A1-A7 load-bearing assumptions named for I12 falsification - OQ1-OQ5 implementer degrees of freedom Predecessor PR #534 (TML-2520, namespace exemplar) is now merged at commit 66da80f; spec updates the predecessor framing accordingly. Linked tickets unchanged: subsumes TML-2579, TML-2580, TML-2582; sibling TML-2537 explicitly out of scope; follow-ups TML-2581/TML-2550, TML-2583 unchanged. Signed-off-by: Will Madden <madden@prisma.io>
…ack mechanism Tighter rewrite focusing on the durable architectural decisions: - Decision 1: Two planes (domain, storage) - kept, lightly trimmed - Decision 2: Uniform <plane>.<ns>.<entityKind>.<entityName> shape - kept, updated example to include pack-contributed postgresEnums slot - Decision 3 (new): Canonical entity coordinate (namespaceId, entityKind, entityName) as the addressing primitive; explains the migration disjoint-calc dependency - Decision 4: Object-pair cross-references - kept - Decision 5 (new, was implicit): Pack-contributed entity-kind mechanism via AuthoringContributions.entityTypes; explicitly names the dual-surface (built-ins hardcoded + contributed kinds via descriptor) as deliberate paid debt - Decision 6: Narrowed framework Namespace interface - kept, trimmed Dropped: - Process-meta sections (D6 ADR-worthy, D7 scope-cleavage from PR #534) - these are project-shaping decisions, not architecture decisions; live in spec.md - "How the structural shape was chosen" narrative - chronology of design discussion doesn't belong in a durable architecture record - "Cleanup work absorbed by this project" section - this is project-scope detail; spec.md owns it Added alternatives G/H/I/J/K to cover the new design space the descriptor mechanism and entity-coordinate decisions opened up (hardcoded-per-pack, single generic entities slot, retrofit built-ins, pair-coordinate, string coordinate). This ADR is the durable artifact - migrates to docs/architecture docs/adrs/ at project close-out per PDoD9. The spec.md retains project-scope details (Tier 1 scope discipline, A1-A7 assumptions, OQ1-OQ5 degrees of freedom). Signed-off-by: Will Madden <madden@prisma.io>
Six slices structured as substrate → migration → propagation: Stack (sequential): - Slice 1 — Two-plane IR primitives + entity-coordinate primitive - Slice 2 — Pack-contributed entity-kind descriptor mechanism - Slice 3 — Migrate Postgres enum off framework-shared types slot (PDoD3 satisfied here; project-specific grep gate enforces) Parallel group A (after stack): - Slice 4 — Cross-reference encoding migration (object pairs) - Slice 5 — On-disk contract regeneration + migration replay verification Final: - Slice 6 — Delete subsumed surfaces + close subsumed Linear tickets (grep gate enforces deletion completeness) Each slice is ≤ 1 PR per drive sizing. Realistic budget ~10-12 days. Slice 3 carries the project's single most-likely re-scope trigger: if assumption A6 (no external consumers pinning shape) is falsified during execution, Slice 3 splits into 3a (deprecation shim) + 3b (hard-cut). Plan's working position is hard-cut. Project-DoD coverage map shows every PDoD has at least one delivering slice. Risks register names A4, A6, A7 falsification scenarios as discussion-mode re-entry triggers per invariant I12. Six Linear tickets need creating before slice pickup begins — folded into the Linear audit pass that's coming up at the umbrella level. Tickets intentionally not auto-created here because the umbrella's Linear audit will resolve duplicate/stale tickets in the same pass. Signed-off-by: Will Madden <madden@prisma.io>
…ect spec Scaffold projects/target-extensible-ir-namespaces/ as the drive-project umbrella for the target-extensible IR + first-class namespaces work. The umbrella composes three units: - S1: contract-ir-planes sub-project (TML-2584) — pre-EA - S2: runtime-qualification slice (TML-2605) — pre-EA - S3: explicit-namespace-dsl slice (TML-2550) — post-EA S1 absorbs the Postgres-enum-as-exemplar migration that earlier framings had as a separate S4 — it's exactly the load-bearing proof of S1's descriptor mechanism, not a separately trackable unit. The spec's load-bearing properties: - Pre-EA scope (S1 + S2) ships zero user-facing query-API breakage. The D8 default-namespace family-façade fallback is the EA-protection mechanism: existing single-default-namespace consumers write the same query code they wrote before this umbrella. - S1's substrate makes future pack-contributed entity kinds cheap. RLS policies, roles, sequences, materialised views — every Postgres-only feature the downstream Supabase initiative needs to ship is one descriptor registration through AuthoringContributions.entityTypes. Honest estimate (A2): ~12-15 days for S1 (6 slices) + S2 = pre-EA slip vs. the "few days" framing. Working position is to accept the slip rather than ship breaking IR changes into EA. Scaffold includes the drive-convention dirs (specs/, plans/, slices/, assets/, adrs/) with .gitkeep files. specs/ and plans/ likely stay empty — the umbrella delegates per-slice specs to slices/<slice>/spec.md and per-sub-project specs/plans to the sub-project's own folder. Signed-off-by: Will Madden <madden@prisma.io>
Single stack thread S1 → S2 → S3. No parallelisation at umbrella level — each unit gates the next on substrate, not just convention. S1 (sub-project, contract-ir-planes) decomposes internally into 6 slices per its own plan; the umbrella plan delegates that detail rather than duplicating it. S2 (slice, runtime-qualification, TML-2605) gets a slice spec when picked up via drive-specify-slice, lands at slices/runtime-qualification/. S3 (slice, explicit-namespace-dsl, TML-2550) same pattern. EA cut sits between S2 and S3 — pre-EA must-ship is S1 + S2; S3 is post-EA additive on S2's default-namespace fallback. PDoD coverage map shows every umbrella PDoD has a delivering unit. Risks register names the EA-timeline slip (A2) as the load-bearing project-level conversation; S1 falsification cascade as the most likely re-sequencing trigger; A4 (default-namespace fallback sufficiency) as the EA-protection-mechanism failure mode. Close-out includes archiving the rolled-up sibling project folders (target-extensible-ir/, namespace-exemplar/, contract-ir-planes/) per drive's projects/ transient-folder discipline. Signed-off-by: Will Madden <madden@prisma.io>
…o-ship falsifier Applied the co-ship falsifier (drive/retro/findings.md 2026-05-20 boundary entry) to the contract-ir-planes plan mid-flight, before locking the slice composition into Linear tickets. Plan changes: - Previous Slice 1 (framework primitives) + Slice 2 (descriptor mechanism wiring) → S1.A "Substrate" (2 dispatches). The previous Slice 1 description literally contained the co-ship falsifier: "no behavioural change yet because nothing consumes the new shape." Collapsed. - Previous Slice 3 (enum migration) + half of Slice 5 (fixture regen) → S1.B (3 dispatches). Fixture regen is forced by the slot change, not a deliberate planning step; it's a dispatch within the slice that causes it. - Previous Slice 4 (cross-references) + other half of Slice 5 → S1.C (2 dispatches). Same rationale. - Previous Slice 6 → S1.D (4 dispatches). Stays separate — reviewers can focus on deletion correctness without conflating with structural changes; grep gate is its own quantitative PDoD check. Total budget unchanged (~10-12 days); slice count drops 6 → 4; dispatch count holds at ~11. Retro entry (drive/retro/findings.md) captures the catch as a win — the falsifier was named in the prior boundary finding; operator surfaced the finding mid-discussion; agent re-read its own plan against the rule rather than defending the original framing. The mechanism (operator attaches prior finding → agent applies falsifier → plan revises before mutation) is itself a reusable pattern worth promoting upstream. Group C in the Linear audit proposal now creates 4 tickets (S1.A through S1.D) instead of 6. Signed-off-by: Will Madden <madden@prisma.io>
Linear audit completed 2026-05-20. On-disk plans updated to reference the actual ticket IDs created during the audit (were "TBD" before): - TML-2622 = S1.A (substrate: primitives + descriptor mechanism) - TML-2623 = S1.B (enum migration off framework-shared types slot — PDoD3) - TML-2624 = S1.C (cross-reference encoding migration) - TML-2625 = S1.D (cleanup + close subsumed tickets) All four created with relatedTo TML-2584 (graph link, not sub-issues). Umbrella plan's external dependencies section updated with full audit outcome: mainline tickets refreshed, subsumed tickets noted (close on slice PR merge via GitHub integration), out-of-umbrella tickets left in project with explanatory comments. Signed-off-by: Will Madden <madden@prisma.io>
…-exemplar Both folders are superseded by the consolidated umbrella: - projects/target-extensible-ir/ — substrate predecessor, now rolled into projects/target-extensible-ir-namespaces/ (umbrella spec) + projects/contract-ir-planes/ (S1 sub-project) - projects/namespace-exemplar/ — TML-2520 / PR #534 shipping artifacts, now merged into main at commit 66da80f Their decision content is captured in the umbrella + sub-project specs and ADR. The transient working artifacts (slices/, reviews/, learnings, plans) live in wip/archived-projects-2026-05-20/ on this worktree only (wip/ is gitignored) for local reference if needed. Resolves the "consolidate the projects into one directory" motivation from the planning conversation: the on-disk projects/ tree now has one umbrella + one sub-project for this work, not five overlapping folders. Signed-off-by: Will Madden <madden@prisma.io>
wmadden
force-pushed
the
target-extensible-ir-namespaces-planning
branch
from
700a7f6 to
15f6159
Compare
This was referenced May 21, 2026
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Why
Consolidates the target-extensible-ir-namespaces umbrella work into a single coherent on-disk project + replans its S1 sub-project (contract-ir-planes) with the descriptor mechanism + entity coordinate decisions from the 2026-05-20 planning conversation. Resolves the "multiple overlapping project folders" problem flagged at the start of the planning session.
Planning only — zero implementation work lands in this PR. Demo apps, package code, fixtures all untouched. No ticket attachments intended: planning PRs should not auto-link to the tickets they plan against (which would auto-close them on merge as if implementation work was completed).
What lands
New umbrella project:
projects/target-extensible-ir-namespaces/spec.md— purpose, scope (substrate + runtime qualification + explicit DSL surface only — no PSL substrate, no RLS), PDoD, A1-A5 assumptionsplan.md— single stack thread S1 → S2 → S3; pre-EA scope = S1 + S2; post-EA = S3slices/,adrs/,specs/,plans/,assets/) per drive conventionRewritten S1 sub-project:
projects/contract-ir-planes/spec.md— D1-D8 decisions (two planes, uniform shape, plane names, object-pair refs, descriptor mechanism, entity coordinate, ADR-worthy, Tier 1 scope discipline), A1-A7 assumptions, OQ1-OQ5 implementer degrees of freedomadrs/0001-contract-planes.md— durable architecture decisions (D1-D6, migrates todocs/architecture docs/adrs/at close-out)plan.md— 4-slice composition (substrate → enum migration → cross-references → cleanup); each slice ≤ 1 PR with internal dispatchesLinear audit executed
Archived
projects/target-extensible-ir/andprojects/namespace-exemplar/removed from the tracked tree; content preserved atwip/archived-projects-2026-05-20/on local worktrees (wip/is gitignored)Drive retro
winstanza indrive/retro/findings.mddocumenting the mid-flight application of theco-shipfalsifier from a priorboundaryfinding. Caught a 6-slice composition that should have been 4 slices before locking it into Linear tickets.Bonus: skills frontmatter fix
drive-discussion,drive-agent-personas,github-review-iterationcarrieddisable-model-invocation: truedespite their bodies listing legitimate auto-trigger contexts. Removed the gate; updateddrive-discussion's description to enumerate when it should fire vs. when it should NOT.What does NOT land
Reviewer focus
Where the value is, in order of return-on-attention:
projects/target-extensible-ir-namespaces/spec.md— does the umbrella's purpose + scope + PDoD match what you think we're shipping? Especially: pre-EA scope (S1 + S2 only) honest about the timeline implications.projects/contract-ir-planes/spec.md§ Approach + § Decision log — the descriptor mechanism (D5) and entity coordinate(namespaceId, entityKind, entityName)(D6) are the load-bearing new decisions. The Tier 1 scope discipline (D8) names what's deferred topostgres-enum-finishingso the boundary stays sharp.projects/contract-ir-planes/plan.md§ Composition — 4 slices, each ≤ 1 PR, with internal dispatches. The 4-slice carving was originally drafted as 6 slices and collapsed mid-flight per the co-ship falsifier (seedrive/retro/findings.mdwin entry); re-eyes welcome.projects/contract-ir-planes/adrs/0001-contract-planes.md— durable architecture decisions. Migrates todocs/architecture docs/adrs/at project close-out per PDoD9.Commit history note
The history shows a draft 6-slice plan superseded by a 4-slice rewrite rather than squashing. The messy history is intentional — the retro stanza references it as a worked example of catching the co-ship falsifier in real time.
Linked tickets
The umbrella and its slices are tracked in Linear under Target-Extensible IR + Namespaces. This PR is planning work and does NOT close any of those tickets; their execution PRs will close them individually.
Summary by CodeRabbit
Documentation
Improvements