blob: a587288ef24f253f24eca81d6a78c97e2fb2164f (
plain) (
blame)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
|
-- Tmp directive system: every task must belong to a directive going forward.
--
-- Background: the unified-surface UI previously surfaced "orphan" tasks
-- (tasks with directive_id NULL) under a synthetic /tmp/ folder. That
-- accumulated stale junk over time and made the UI noisy. The new model:
--
-- * Add `is_tmp` to directives — at most one per owner, marks the
-- special "scratchpad" directive that holds otherwise-orphan tasks.
-- * Delete every existing orphan task. The user explicitly asked for
-- a clean slate: "ALSO there are TOO MANY old tasks in tmp, we need
-- to remove all of them as well."
-- * Going forward, ephemeral / standalone task creation paths attach
-- to the caller's tmp directive (auto-created on first use by the
-- repository helper, not by this migration — owners may not exist
-- yet at migration time, but every owner gets one as soon as a
-- standalone task is requested).
-- * A 30-day expiry sweep in the directive reconciler deletes tasks
-- in tmp directives once they age out.
-- 1. New flag column on directives. Default false; only set true on the
-- auto-created scratchpad directive.
ALTER TABLE directives
ADD COLUMN is_tmp BOOLEAN NOT NULL DEFAULT false;
-- Partial unique index — at most ONE tmp directive per owner.
CREATE UNIQUE INDEX idx_directives_owner_tmp_unique
ON directives(owner_id)
WHERE is_tmp;
-- 2. Clear out every existing orphan task. Per the user's spec these are
-- discardable scratch work; pre-existing valuable tasks are already
-- attached to a directive and will not be touched.
--
-- Cascades: task_events delete via FK; daemon links go to NULL; nothing
-- in the contracts/directive_steps tables references orphan tasks (a
-- contract-backed step always has a directive_id by construction).
DELETE FROM tasks
WHERE directive_id IS NULL
AND parent_task_id IS NULL;
|
