Default MinBatchSize of 50 left quiet admin traffic in memory
forever because MaxFlushWait was unset. Force a timed flush so
the logs page can show recent authenticated requests.
Global Router.Use middleware is applied at HTTP Start instead of
being snapshotted when each route is registered, so risk_control
still wraps admin APIs that mount earlier. Access-log collection
is enabled by default on SQLite/Postgres, not only ClickHouse.
Public config now comes from admin as a flat visibility=1 map instead of
a cross-plugin query that compared an integer column to "visible". Login
and register resolve CaptchaService per request so CAP is not skipped
when user applies before cap.
Wire plugin services through Bind/InjectFrom and AppContext so HTTP and
workers resolve dependencies after Apply. Register TaskHandler objects
with persisted results, and implement send_email_code, mail:send,
cleanup_inactive_users, and dispatch_bot_msg.
Asynq func handlers now go through ProcessTask so admin execution
rows leave pending. The in-process worker resolves admin type
identifiers and writes the same w_task_executions table. Remove
the no-op admin system_cleanup that shadowed the upload handler.
- implement route whitelist registration and wildcard matching in RouterExtension
- add cookie store session fallback when Redis is disabled in driver_http
- actively register public auth endpoints to whitelist in auth plugin
- update user handlers to persist session and clear cookie on logout
- document router whitelist mechanism in AGENTS.md and new-api skill
* autoresearch iter 23: fail-closed auth guarding for user/message_gateway
Both plugins resolve contracts.AuthService in Apply to build their route
middleware, but declared only DBService in Inject(). The kernel gates a
plugin's Apply solely on declared deps, and cmd/app.go registers user
before auth, so user mounted first, core.Inject failed, and loginMW
silently degraded to a pass-through closure — leaving /api/v1/user
change-password, profile and access-tokens unguarded. message_gateway
was saved only by its later list position.
Declare AuthService in Inject() for both, and pin the property with a
reconcile-level test that mirrors production registration order and
asserts the real auth middleware reaches the route table.
* autoresearch iter 24: make auth middleware fallbacks fail closed
user, message_gateway and admin each fell back to a c.Next() closure when
contracts.AuthService could not be resolved, so a route would be served as
if authenticated. For admin this is reachable at runtime: OnDispose calls
service.ResetServices(), which nils the global the per-request guard reads,
so requests still in flight during dispose bypass authorization entirely.
Add ginutil.AuthUnavailable() and bind every fallback to it, with a test
that drives each plugin's registered guard without an auth service present
and asserts the request is aborted rather than passed through.
* chore(autoresearch): log iter 23 (fail-open auth ordering, proven)
* autoresearch iter 24 follow-up: let staticcheck infer the auth guard type
* docs(autoresearch): log iters 24-25 and lessons 9-11 (declared-dep bug class, gate discipline)
24 of the 96 nolint directives were dead: they covered findings that no
longer exist. A stale suppression is not inert — it silently claims any
future finding for that linter in that scope, so a real problem raised
there would vanish without anyone noticing. Explanatory prose was kept as
ordinary comments.
Two directives proved load-bearing under the project gate even though
nolintlint reported them unused, and removing them exposed verified
contextcheck false positives: App.Run does forward a sigCtx derived from
the caller's context to Start, and the migration lock renewal must keep
its own deadline because the task context may already be canceled. Both
were restored, narrowed to the live linter, and given the reason the
originals lacked.
downloadMedia created a fresh os.MkdirTemp for every private message carrying
a photo or document, and no code path anywhere reads Attachment.Path, so each
message permanently grew the disk while burning a Bot API download. The
handler now removes the directory once onInbound returns.
No mechanical proof is possible here: exercising downloadMedia needs a live
telebot download. Verified by reading every consumer of InboundMessage
.Attachments instead.
telebot types LongPoller.Timeout as time.Duration and sends
int(timeout / time.Second) to getUpdates, so the literal 10 meant ten
nanoseconds: Telegram received timeout=0, long polling never held the
connection, and the adapter polled the Bot API in a tight loop instead.
Use 10 seconds and extract the settings so the conversion is asserted.
The adapter also has no media temp-dir cleanup (downloadMedia creates an
MkdirTemp per attachment and nothing removes it); that is left as a separate
change rather than bundled here.
upload's four real background tasks (system cleanup, stats rebuild, storage
migration, image warmup) plus the admin and user stubs registered handlers
typed as func(ctx, *asynq.Task) error. Only the asynq worker accepts that
shape; the Redis-free in-process worker's invokeHandler rejects it with
'unsupported handler type', so none of those tasks could ever run in that
deployment mode. Take payload bytes instead, which both drivers support.
Adds architecture gate check 7 forbidding asynq imports from business and
infrastructure plugins. It deliberately does not cover robfig/cron: the admin
plugin uses cron.ParseStandard only to validate a user-entered spec, which is
a library call rather than a driver binding, and the in-process scheduler
already normalizes 5-field specs.
The user plugin registered a cron dispatching to user:daily_audit, a task
pattern it never registers, and no audit logic exists anywhere in the plugin.
The daily run therefore went nowhere while a test asserted the schedule was
registered — proving the wiring existed, not that it worked. Implementing a
real daily audit is unstarted functionality, so the schedule is removed rather
than stubbed.
The combined domain test now asserts the real invariant across all applied
plugins: every schedule's task type must have a registered handler.
message_gateway scheduled message_gateway:cleanup_pairing_codes every 10
minutes but never registered a task under that pattern, so every dispatch
went to a task type with no handler and expired pairing rows accumulated
forever, even though repository.DeleteExpiredPairingCodes already existed.
Add a test that fails for any schedule whose task pattern is unregistered:
it reports the exact orphan rather than relying on a schedule-exists assert.
enrichAccessLogsWithUsers preferred the UserService contract over the local
repository — correct layering, but it looped GetUserByID and issued up to a
page-size worth of separate SELECTs against w_users, while the single-query
WHERE id IN variant was only reached in the no-contract fallback branch.
Give the contract a GetUsersByIDs so callers can keep the layering and drop
the N+1. The test asserts 1 query batched against 3 per-id, so the counting
itself is checked.
isOriginAllowed read server_address from w_system_configs for every request
carrying an Origin header — one uncached primary-DB round-trip plus a split
and trim loop per browser request, while sibling config reads in the storage
driver are already TTL cached. Read it through the shared CacheService with
the same 5s window, falling back to the database when no cache is bound.
driver_http now binds CacheService in Apply the way it already binds DBService.
AppendTaskExecutionLog discarded the error from its read of the buffer, so a
transient cache failure looked like an empty buffer and the very next write
replaced the whole accumulated log with just the newest line. Flush already
distinguished miss from failure; append now does the same.
GetExecution returned (nil, nil) under the in-process driver where the asynq
driver returns an error, so the same contract call meant 'empty' in one
deployment mode and 'failed' in the other.
Authenticate, CreateAuthSource, UpdateAuthSource and ToggleAuthSource claimed
success with a nil record, so any test that reached them surfaced a nil
pointer dereference instead of the actual cause. Full suite confirms no test
relied on the silent behaviour.
RunPushTest flattened channel validation failures with %v, telegram's
fallback path discarded the original send error, and the config loader's
type assertion on viper.ConfigFileNotFoundError would miss a wrapped form
and fatally abort over a merely missing file. errorlint now reports zero.
loadTaskExecutionLog and loadTaskExecutionLogs could never fail, yet four
call sites branched on their error as if they could, presenting unreachable
code as error handling.
bodyContent's fallback ranged over the body map, and Go randomizes map
iteration, so the same notification rendered its fields in a different order
on every send. Observed failing before the fix: the second call already
reordered the output. Iterate sorted keys instead.
email, telegram and lark each re-implemented the title/content/level lookup
with only their markup differing, and each carried a dead content := ""
initialization that every branch overwrote. Three small helpers in template.go
now own that logic.
ryan