v12.0.1

• Fixed: a web SQLite store restores the last commit after a tab died inside a
  commit. The opfs-sahpool VFS of the vendored engine reported a reserved
  lock on every check. SQLite then took the hot journal that a tab killed
  inside a commit left for the journal of a live transaction and never played
  it back, so the next open read the half-written commit. In 12.0.0 and
  earlier, such a store could report database disk image is malformed. The
  patched engine reports no reserved lock, and the next open of such a store
  plays the journal back.

• Fixed: a web SQLite open with the wrong key state leaves the file as it was.
  The open reads the first page of the file through an immutable read-only
  connection with no key before it opens the file. An open with no key of an
  encrypted file, or an open with a key of a plaintext file, rejects with
  file is not a database and plays no hot journal back. In 12.0.0 and
  earlier, such an open of a file with a hot journal played the journal back
  through the wrong key state and left the file corrupt.

• Fixed: ratchetConfig.maxSkip sets the skip limit. The client passes
  SignalProtocolClientConfig.ratchetConfig.maxSkip to the session cipher of
  the protocol manager that it creates. In 12.0.0 and earlier, the client did
  not read the option. A gap above the limit fails with
  TOO_MANY_SKIPPED_MESSAGES. The default is 25000, which is the value that
  the cipher used before. The API reference said 1000 and now says 25000. A
  value that is not a positive integer fails at client creation with
  ratchetConfig.maxSkip must be a positive integer. A protocolManager that
  the caller supplies keeps its own limit. The client does not read
  maxMessageKeysStored or keyExpirationMs. Their reference now states the
  fixed values: 2000 skipped message keys per receiver chain, deleted after 7
  days.