Pre-launch — Gluecron is in final validation. Public signups and git hosting for non-owner users open after launch review.
CodeIssuesDiscussionsWikiPull RequestsProjectsCommitsActionsReleasesContributorsPulse● GatesSecuritySettingsDeploymentsPipelineInsightsAgents✨ Explain✨ Ask AI✨ Workspace✨ Spec✨ Tests▓ Debt Map✨ NL Search🏛 Archaeology
Blame · Line-by-line history

0063_multi_repo_refactors.sql

Each line is annotated with the commit that last touched it. Click any SHA to jump to that commit and see the surrounding change.

0063_multi_repo_refactors.sqlBlame64 lines · 1 contributor
23d0abfClaude1-- Multi-repo refactor agent — one English request fan-outs to coordinated
2-- PRs across every affected repo, then merges them in order once all PRs
3-- are green + approved.
4--
5-- The classic example: "rename `getUserById` to `findUser`". A user owns
6-- 8 repos that import the helper; the agent walks each of them, opens an
7-- AI-authored PR with the rename, and tags every PR with a single
8-- `multi-repo:refactor:<id>` label so the UI can group them.
9--
10-- Two tables:
11-- `multi_repo_refactors` — the parent record. One row per user-issued
12-- request. Lifecycle: planning → building →
13-- ready_for_review → merged | failed.
14-- `multi_repo_refactor_prs` — one row per affected repo. Holds the FK to
15-- the `pull_requests` row (nullable until the
16-- per-repo PR is actually opened) plus the
17-- per-repo status (pending|building|opened|
18-- failed) and an error message if the build
19-- step bailed for that repo.
20--
21-- Coordinated merge: when the user (or autopilot) triggers a merge on the
22-- parent refactor, the orchestrator walks the child rows in insertion order;
23-- if any child fails the rest stay open so the user can intervene. The
24-- merge driver lives in `src/lib/multi-repo-refactor.ts`.
25
26CREATE TABLE IF NOT EXISTS multi_repo_refactors (
27 id uuid PRIMARY KEY DEFAULT gen_random_uuid(),
28 owner_user_id uuid NOT NULL REFERENCES users(id) ON DELETE CASCADE,
29 title text NOT NULL,
30 description text NOT NULL,
31 -- 'planning' | 'building' | 'ready_for_review' | 'merged' | 'failed'
32 status text NOT NULL DEFAULT 'planning',
33 created_at timestamp NOT NULL DEFAULT now(),
34 updated_at timestamp NOT NULL DEFAULT now()
35);
36
37CREATE INDEX IF NOT EXISTS multi_repo_refactors_owner
38 ON multi_repo_refactors (owner_user_id, created_at DESC);
39
40CREATE INDEX IF NOT EXISTS multi_repo_refactors_status
41 ON multi_repo_refactors (status, updated_at DESC);
42
43CREATE TABLE IF NOT EXISTS multi_repo_refactor_prs (
44 id uuid PRIMARY KEY DEFAULT gen_random_uuid(),
45 refactor_id uuid NOT NULL REFERENCES multi_repo_refactors(id) ON DELETE CASCADE,
46 repository_id uuid NOT NULL REFERENCES repositories(id) ON DELETE CASCADE,
47 -- Nullable: stays NULL until the per-repo PR is actually opened.
48 pull_request_id uuid REFERENCES pull_requests(id) ON DELETE SET NULL,
49 -- 'pending' | 'building' | 'opened' | 'failed'
50 status text NOT NULL DEFAULT 'pending',
51 error_message text,
52 created_at timestamp NOT NULL DEFAULT now(),
53 updated_at timestamp NOT NULL DEFAULT now()
54);
55
56-- One row per (refactor, repo). A refactor never targets the same repo twice.
57CREATE UNIQUE INDEX IF NOT EXISTS multi_repo_refactor_prs_unique
58 ON multi_repo_refactor_prs (refactor_id, repository_id);
59
60CREATE INDEX IF NOT EXISTS multi_repo_refactor_prs_refactor
61 ON multi_repo_refactor_prs (refactor_id, status);
62
63CREATE INDEX IF NOT EXISTS multi_repo_refactor_prs_repo
64 ON multi_repo_refactor_prs (repository_id);