Bun's `fetch`, which `~/bin/quak` runs under, rejects with a plain `Error`
that has no `cause` and a code of bun's own. Running the built binary against
local listeners gave `ConnectionRefused` for a refused connection and a failed
TLS handshake, and `ConnectionRefused` or `FailedToOpenSocket` for a name that
does not resolve. Both join `TRANSPORT_CODES` in `src/retry.ts`, so they are
retried and called transient. They stay out of `CONNECT_CODES`, so a `POST` or
`PUT` is not replayed on them, because bun also gives them when it resends a
request on a new connection. `classify` in `src/backup.ts` records a
`TimeoutError` as transient, so a stalled download is no longer `unknown`.
Not changed: a certificate bun rejects is not retried.
Model: opus-5-5
Node's `fetch` rejects every failure below HTTP as `fetch failed` and puts
the reason on `cause`. `describeError` in `src/errors.ts` gives the message
followed by the messages on the cause chain, and for an `AggregateError`,
which Node gives when every address of a host fails, the message of each
address. `run` prints it, and the backup logs and records it. `classify`
reads that text and calls a failure transient when `hasTransportCode`, now
exported by `src/retry.ts`, finds a transport errno on the chain. A failed
ML data fetch keeps its error as the cause. The bounded walk along the
chain moved from `src/retry.ts` to `src/errors.ts`.
A truncated download's message leaves the authentication failure to its
cause.
Model: opus-5-5
The backup awaited each original before starting the next, so it used
one of the content pool's five slots. lib.backup(), and so quak backup,
now runs as many workers as the content pool's limit (5 by default, or
the contentConcurrency of the pools the library was opened with), each
taking the next file from one shared list, so the limit covers the
--verify hashing too. Once a worker has thrown or the lock is lost, no
worker takes another file; the backup waits for the files under way and
checks its lock before the ML data fetch. runBackup rejects a
contentConcurrency that is not a positive integer before it takes the
lock. Failed files are listed in the order they failed.
Model: opus-5-5
proper-lockfile's default onCompromised throws from a timer, so a backup
whose backup.lock was removed or taken over by another backup died with
a stack trace. lockBackupDirectory now returns the lock, its release
function and why it was lost, and passes a handler that only records
the reason. The backup checks it after the refresh and before each
original, thumbnail, file's JSON and album, and before it writes
failures.json, and then fails with "lost the lock on DIR: REASON",
which quak backup prints as one line, exiting 1. lib.backup() takes a
caller's lock as lock, in place of lockHeld: true. Releasing a lost lock
leaves backup.lock alone, since it may be another backup's.
Model: opus-5-5
test/cli/backup-end-to-end.test.ts runs backupCommand through run, as
bin/quak.ts does, with the real loadSession reading a session saved after
logging in, and every request answered by a fake of the Ente HTTP API serving
data encrypted with libsodium. It covers the backup tree of an account with
two albums of its own and one shared with it, a second run that downloads
nothing, a download that succeeds after two 503s, a 404 recorded in
failures.json with exit 1, a refresh answered 401 with exit 3, and --verify
putting back an original changed on disk.
The fake server and the account it logs in moved from the backup-metadata
test into test/ente-server.ts, unchanged, so both tests use them.
Model: opus-5-5
clawbot