Files
ironclaw/migrations/V17__document_versions.sql
Illia Polosukhin a68358086a fix(db): resolve V15 migration numbering conflict (#1923)
* fix(db): resolve V15 migration numbering conflict between user_identities and conversation_source_channel

A merge conflict left two PostgreSQL migrations at V15. This renumbers them
(V15=user_identities, V16=conversation_source_channel, V17=document_versions)
to match the libSQL incremental ordering. Adds user_identities, document_versions,
and source_channel to the libSQL base schema so fresh databases get all tables.
Includes a one-time repair for existing databases where V15 was mis-recorded.

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

* style: fix rustfmt formatting in repair_misnumbered_v15

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

* fix(db): address PR review — proper error handling and tighter repair condition

- Replace .ok().flatten() with explicit error propagation via .map_err()?
  so DB errors during V15 repair are surfaced, not silently swallowed
- Tighten repair condition from `!= "user_identities"` to `== "document_versions"`
  to only fix the specific known-bad case from the merge conflict

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

---------

Co-authored-by: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
2026-04-02 11:06:27 -07:00

25 lines
980 B
SQL

-- Document version history for workspace files.
-- Every content update saves the previous content as a version,
-- enabling rollback and audit trails.
CREATE TABLE memory_document_versions (
id UUID PRIMARY KEY DEFAULT gen_random_uuid(),
document_id UUID NOT NULL REFERENCES memory_documents(id) ON DELETE CASCADE,
version INTEGER NOT NULL,
content TEXT NOT NULL,
content_hash TEXT NOT NULL,
created_at TIMESTAMPTZ NOT NULL DEFAULT NOW(),
changed_by TEXT,
UNIQUE(document_id, version)
);
CREATE INDEX idx_doc_versions_lookup
ON memory_document_versions(document_id, version DESC);
-- GIN index on metadata for future JSON containment queries
-- (e.g., WHERE metadata @> '{"hygiene": {"enabled": true}}').
-- Not used by current queries (which use path LIKE) but enables
-- efficient metadata-based filtering without a full table scan.
CREATE INDEX idx_memory_documents_metadata
ON memory_documents USING GIN (metadata jsonb_path_ops);