feat(admin): deleting an account stops deleting other people's work

`tasks.created_by` was ON DELETE CASCADE, so retiring a colleague removed every
task they had ever written — including in projects somebody else owns. The
interface said so in plain words before you pressed the button, which is a
warning rather than a fix. It was the last remaining way to lose other people's
work by accident.

SET NULL instead: the task stays and loses its author, the same shape
`assignee_id` and `completed_by` already have.

SQLite cannot alter a foreign key action, so the table is rebuilt — and with
foreign keys on, the drop in the middle of a rebuild is a cascading delete that
takes every row referencing `tasks` with it. `PRAGMA foreign_keys` is a no-op
inside a transaction, so migrations now run on a connection of their own, and a
migration can opt out of enforcement with `-- verdande:rebuild-tables`.
`PRAGMA foreign_key_check` runs before the commit, so a rebuild that leaves a
dangling reference fails the migration rather than writing it.

The test fills a database at 0007, upgrades it, and counts. Without the marker it
loses a task, a label link and a comment — which is exactly what it is for.

The admin list now says two numbers instead of one: what disappears, and what
stays behind unattributed. They are different sentences and the second is a
reassurance.

Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
This commit is contained in:
Kristian Wind
2026-08-18 09:03:59 +02:00
co-authored by Claude Opus 5
parent a29534c811
commit ebdc8eacf2
8 changed files with 541 additions and 79 deletions
@@ -0,0 +1,101 @@
-- `tasks.created_by` stops cascading.
--
-- It was ON DELETE CASCADE, which made deleting an account delete every task that
-- account ever wrote — including in projects somebody else owns. Retiring a
-- colleague therefore removed their contributions from shared work rather than
-- leaving it behind. It is the last remaining way to lose other people's work by
-- accident, and the administrator's page saying so in plain words is a warning, not
-- a fix.
--
-- SET NULL instead: the task stays and loses its author. `created_by` becomes
-- nullable to make that expressible, which is the same shape `assignee_id` and
-- `completed_by` already have for the same reason.
--
-- SQLite cannot alter a foreign key action, so the table is rebuilt. The FTS
-- triggers come along: they are dropped first so that DROP TABLE and ALTER TABLE
-- see a schema they can parse, and recreated below. `tasks_fts` is an
-- external-content table keyed on `tasks.rowid`, and a rebuilt table hands out new
-- rowids, so the index is rebuilt at the end rather than left pointing at rows that
-- have moved.
--
-- verdande:rebuild-tables
DROP TRIGGER tasks_fts_insert;
DROP TRIGGER tasks_fts_delete;
DROP TRIGGER tasks_fts_update;
CREATE TABLE tasks_rebuilt (
id TEXT PRIMARY KEY,
project_id TEXT NOT NULL REFERENCES projects (id) ON DELETE CASCADE,
section_id TEXT REFERENCES sections (id) ON DELETE SET NULL,
parent_id TEXT REFERENCES tasks (id) ON DELETE CASCADE,
content TEXT NOT NULL,
description TEXT NOT NULL DEFAULT '',
priority INTEGER NOT NULL DEFAULT 4,
due_date TEXT,
due_datetime INTEGER,
due_timezone TEXT,
duration_min INTEGER,
recurrence_rule TEXT,
assignee_id TEXT REFERENCES users (id) ON DELETE SET NULL,
-- Nullable, and SET NULL: an account can go without its work going with it.
created_by TEXT REFERENCES users (id) ON DELETE SET NULL,
completed_at INTEGER,
completed_by TEXT REFERENCES users (id) ON DELETE SET NULL,
sort_order REAL NOT NULL DEFAULT 0,
created_at INTEGER NOT NULL,
updated_at INTEGER NOT NULL,
deleted_at INTEGER,
fold TEXT GENERATED ALWAYS AS (
replace(replace(replace(replace(replace(replace(
content || ' ' || description,
'ø', 'o'), 'Ø', 'o'),
'æ', 'ae'), 'Æ', 'ae'),
'å', 'aa'), 'Å', 'aa')
) STORED,
CHECK (priority BETWEEN 1 AND 4)
);
INSERT INTO tasks_rebuilt
(id, project_id, section_id, parent_id, content, description, priority,
due_date, due_datetime, due_timezone, duration_min, recurrence_rule,
assignee_id, created_by, completed_at, completed_by, sort_order,
created_at, updated_at, deleted_at)
SELECT
id, project_id, section_id, parent_id, content, description, priority,
due_date, due_datetime, due_timezone, duration_min, recurrence_rule,
assignee_id, created_by, completed_at, completed_by, sort_order,
created_at, updated_at, deleted_at
FROM tasks;
DROP TABLE tasks;
ALTER TABLE tasks_rebuilt RENAME TO tasks;
CREATE INDEX idx_tasks_project ON tasks (project_id, sort_order)
WHERE deleted_at IS NULL AND completed_at IS NULL;
CREATE INDEX idx_tasks_due ON tasks (due_date)
WHERE deleted_at IS NULL AND completed_at IS NULL AND due_date IS NOT NULL;
CREATE INDEX idx_tasks_assignee ON tasks (assignee_id)
WHERE deleted_at IS NULL AND completed_at IS NULL AND assignee_id IS NOT NULL;
CREATE INDEX idx_tasks_parent ON tasks (parent_id) WHERE parent_id IS NOT NULL;
CREATE TRIGGER tasks_fts_insert AFTER INSERT ON tasks BEGIN
INSERT INTO tasks_fts (rowid, content, description, fold, id, project_id)
VALUES (new.rowid, new.content, new.description, new.fold, new.id, new.project_id);
END;
CREATE TRIGGER tasks_fts_delete AFTER DELETE ON tasks BEGIN
INSERT INTO tasks_fts (tasks_fts, rowid, content, description, fold, id, project_id)
VALUES ('delete', old.rowid, old.content, old.description, old.fold, old.id, old.project_id);
END;
CREATE TRIGGER tasks_fts_update AFTER UPDATE ON tasks BEGIN
INSERT INTO tasks_fts (tasks_fts, rowid, content, description, fold, id, project_id)
VALUES ('delete', old.rowid, old.content, old.description, old.fold, old.id, old.project_id);
INSERT INTO tasks_fts (rowid, content, description, fold, id, project_id)
VALUES (new.rowid, new.content, new.description, new.fold, new.id, new.project_id);
END;
-- The rowids the index was built against are gone with the old table.
INSERT INTO tasks_fts (tasks_fts) VALUES ('rebuild');