Update dependency @logtape/logtape to v2.4.0

[renovate]

This PR contains the following updates:

Package	Change	Age	Confidence
@logtape/logtape (source)	2.3.11 → 2.4.0

---

Release Notes

dahlia/logtape (@​logtape/logtape)

v2.4.0

Compare Source

Released on October 7, 2026.

@​logtape/logtape

• Added "forward" mode to the parentSinks option of LoggerConfig.
  Unlike the default "inherit" mode, "forward" inherits ancestors'
  configured sinks without applying each ancestor's lowestLevel
  threshold, so records accepted by the logger's own lowestLevel reach
  the inherited sinks as well. An ancestor configured with
  parentSinks: "override" still stops the inheritance chain. [#​198]

• Added FingersCrossedSink.flush() and FingersCrossedSink.discard()
  methods and the FingersCrossedOptions.bufferAction callback for flushing
  or discarding isolated buffers when a request or job completes. [#​205]

• Added an optional synchronous snapshot(record) callback to
  FingersCrossedOptions to copy records when fingersCrossed() buffers
  them. [#​242, #​250]

• Added getLoggers() function to enumerate every logger in the category
  tree rooted at a given logger (or the root logger by default), including
  the given logger itself, in depth-first pre-order.
  [#​62, #​207 by Jepoy]

• Added inspectLogger() function, which explains how the current
  configuration routes records for a logger without logging anything or
  invoking filters and sinks. It reports the effective category, the
  level gates and sink paths (repeats included) with the categories that
  supplied them, the selected filters, where sink inheritance stops, and
  whether a scoped configuration applies. [#​243, #​251]

  • Added InspectLoggerOptions, LoggerInspection,
    LoggerNodeInspection, SinkPathInspection,
    LevelGateInspection, FilterSetInspection, FilterInspection,
    and InheritanceBoundaryInspection interfaces.
  • Added LoggerInspectionStatus type.

• Added opt-in source location capture for development logging. Setting
  the new captureSourceLocation option of a logger configuration to
  true records where each logging method was called, when the runtime's
  stack trace makes it available, in the new optional
  LogRecord.sourceLocation field, typed as the new SourceLocation
  interface. Child categories inherit the setting, withConfig() and
  withConfigSync() accept it too, and it is off by default. LogTape
  does not resolve source maps, so bundled or minified code may report
  positions in the generated code. [#​134, #​241, #​253]

• Added the TextFormatterOptions.sourceLocation option so that
  getTextFormatter() and getAnsiColorFormatter() can show captured
  source locations, and the optional FormattedValues.sourceLocation
  field for custom format callbacks. [#​134, #​241, #​253]

• Added the getConsoleFormatter() function and the
  ConsoleFormatterOptions interface. Its sourceLocation option shows
  captured source locations in console output; without options it
  formats records like defaultConsoleFormatter(). [#​134, #​241, #​253]

• Added optional onDrop and onError callbacks to non-blocking console
  and stream sinks to report aggregated buffer overflow counts and output
  failures without exposing dropped records. Added the SinkDropReason,
  SinkDropEvent, and SinkErrorEvent types for these notifications.
  [#​245, #​249]

• Fixed concurrent or repeated disposal of a non-blocking stream sink so
  pending output finishes before its writer is closed or released.
  [#​245, #​249]

• Added queue limits and a request lifetime hook to fromAsyncSink().
  Its type is now
  (asyncSink: AsyncSink, options?: AsyncSinkOptions) => Sink & AsyncDisposable
  (was (asyncSink: AsyncSink) => Sink & AsyncDisposable). Without options,
  the adapter behaves as before. [#​238, #​246]

  • The maxQueueSize and overflow options cap how many records wait
    for the async sink, not counting the one being processed, and choose
    whether a full queue drops its oldest waiting record (the default)
    or the incoming one. The limit bounds queued records, not total
    memory: with "drop-oldest", the promise and request-context
    bookkeeping of each dropped record remains until the record being
    processed at that time settles. "drop-newest" leaves nothing
    behind for rejected records.
  • The onDrop callback receives a SinkDropEvent with the number of
    dropped records and the reason, but no record payloads.
  • The waitUntil callback is called synchronously while a record is
    being logged, with a Promise that settles once that record and
    every earlier record have been processed or dropped. Passing
    a platform's waitUntil() function lets serverless functions finish
    sending logs after returning a response.
  • Added the AsyncSinkOptions interface and the
    AsyncSinkOverflowPolicy, SinkDropEvent, and SinkDropReason
    types.

• Added the "bare" preset to the TextFormatterOptions.value option, so
  that getTextFormatter() and getAnsiColorFormatter() can render
  interpolated string values without the quotes and escapes that
  inspect() adds. Other values, including strings nested in objects and
  arrays, are rendered as before. Bare strings are sanitized instead: SGR
  sequences are always escaped, and newlines are escaped unless
  TextFormatterOptions.sanitize explicitly preserves them. [#​236, #​237]

• Changed the type of the TextFormatterOptions.value option to
  "bare" | ((value: unknown, inspect: (value: unknown, options?: { colors?: boolean }) => string) => string)
  (was
  (value: unknown, inspect: (value: unknown, options?: { colors?: boolean }) => string) => string).
  Code that calls the option directly now has to check that it is a function
  first. [#​236, #​237]

• Added the FingersCrossedOptions.afterTrigger option and the
  FingersCrossedAfterTrigger type. Setting it to "buffer" makes
  fingersCrossed() go back to buffering after each trigger instead of
  passing every subsequent record through, so long-running processes can
  output each error together with the records that led up to it.
  [#​234, #​235]

• Added the StreamSinkOptions.closeStream option to dispose stream sinks
  without closing caller-owned streams. [#​203]

• Added the drain() function and the Drainable interface for waiting
  until the records that sinks accepted before the call have settled,
  without disposing of the sinks. A record settles when its sink operation
  finishes, including when the operation fails, or when the sink drops the
  record after accepting it. Records logged after the call are not waited
  for. The sinks returned by getStreamSink() and fromAsyncSink() have
  a drain() method, and withFilter() and fingersCrossed() forward it
  when the wrapped sink has one. Draining a fingersCrossed() sink does
  not release records still waiting for a trigger. [#​239, #​247]

• Changed the type of the getStreamSink() function to
  (stream: WritableStream, options?: StreamSinkOptions) => Sink & AsyncDisposable & Drainable
  (was
  (stream: WritableStream, options?: StreamSinkOptions) => Sink & AsyncDisposable).
  [#​239, #​247]

• Changed the type of the fromAsyncSink() function to
  (asyncSink: AsyncSink, options?: AsyncSinkOptions) => Sink & AsyncDisposable & Drainable
  (was
  (asyncSink: AsyncSink, options?: AsyncSinkOptions) => Sink & AsyncDisposable).
  [#​239, #​247]

• Added overloads to the withFilter() and fingersCrossed() functions
  that keep Drainable in the return type when the wrapped sink is
  Drainable, such as
  withFilter(sink: Sink & Drainable, filter: FilterLike): Sink & Drainable.
  [#​239, #​247]

@​logtape/config

• Added support for the captureSourceLocation option of logger
  configurations, matching @​logtape/logtape 2.4.0. [#​134, #​241, #​253]
• The parentSinks option of logger configurations now accepts
  "forward", matching @​logtape/logtape 2.4.0. [#​198]

@​logtape/elysia

• Added support for Elysia 2 beta while retaining support for Elysia 1.4.
  [#​212]
• Added the completionLevel option to ElysiaLogTapeOptions, which
  chooses the log level of a completed request's log record from the
  response outcome, for example "error" for server errors and
  "warning" for slow responses. Added the CompletionLevelFunction
  type, (ctx, completion) => LogLevel, and the RequestCompletion
  interface, which has the status, responseTime, and, for the record
  written by the plugin's error hook, the error. The callback also
  chooses that error record's level, which otherwise stays "error",
  without adding another record. The level option still applies to
  logRequest logs. When the callback throws or returns an invalid level,
  the record uses level, or "error" for the error record; such failures
  are reported to the meta logger and never change the response.
  [#​240, #​248]

@​logtape/express

• Added the completionLevel option to ExpressLogTapeOptions, which
  chooses the log level of a completed request's log record from the
  response outcome, for example "error" for server errors and
  "warning" for slow responses. Added the CompletionLevelFunction
  type, (req, res, completion) => LogLevel, and the RequestCompletion
  interface, which has the final status and responseTime. The level
  option still applies to immediate logs, and is used when the callback
  throws or returns an invalid level; such failures are reported to
  the meta logger and never change the response. [#​240, #​248]

@​logtape/file

• Added the rotatedFilePath option to getRotatingFileSink() to customize
  rotated log file paths, for example to keep the .log extension.
  [#​215, #​216]
• Changed getRotatingFileSink() to reject positive maxFiles values unless
  they are integers from 1 to 1,000, throwing RangeError during sink
  creation. Update configurations outside this range before upgrading. Zero
  and negative values still disable backups. [#​216]

@​logtape/hono

• Added the completionLevel option to HonoLogTapeOptions, which
  chooses the log level of a completed request's log record from the
  response outcome, for example "error" for server errors and
  "warning" for slow responses. Added the CompletionLevelFunction
  type, (c, completion) => LogLevel, and the RequestCompletion
  interface, which has the status, responseTime, the error from
  Hono's error handler or a failed response body stream, and whether
  the response body was aborted. The level option still applies to
  logRequest logs, and is used when the callback throws or returns
  an invalid level; such failures are reported to the meta logger and never
  change the response. [#​240, #​248]

@​logtape/koa

• Added the completionLevel option to KoaLogTapeOptions, which chooses
  the log level of a completed request's log record from the response
  outcome, for example "error" for server errors and "warning" for slow
  responses. Added the CompletionLevelFunction type,
  (ctx, completion) => LogLevel, and the RequestCompletion interface,
  which has the status and responseTime. The level option still
  applies to logRequest logs, and is used when the callback throws or
  returns an invalid level; such failures are reported to the meta logger
  and never change the response. [#​240, #​248]

@​logtape/lint

• Added an opt-in no-dynamic-message rule to @​logtape/lint for finding
  dynamic expressions passed as log messages. The rule uses available
  TypeScript type information to distinguish properties, callbacks, and
  Error overloads, with local inference as a fallback. Deno Lint users can
  enable it through the @logtape/lint/deno/strict entry point. [#​199]
• Added an opt-in no-unrendered-properties lint rule to
  @​logtape/lint for finding properties that are not referenced by a
  message template and may be omitted by sinks that output only the
  message. The rule is excluded from the recommended preset and the
  default Deno Lint plugin, and is included in
  @logtape/lint/deno/strict. [#​214]

@​logtape/pretty

• Added the PrettyFormatterOptions.value option, which takes the same
  values as TextFormatterOptions.value. Setting it to "bare" renders
  interpolated string values without the quotes and escapes that
  inspect() adds, sanitizing them instead; a function customizes how
  values are rendered, and receives an inspect() function that applies
  the inspectOptions and colors options. [#​236, #​237]

@​logtape/testing

• Added LogRecorder.waitFor() and LogRecorderWaitOptions to wait for
  the first matching record with a timeout and optional cancellation through
  an AbortSignal. [#​244, #​252]

v2.3.12

Compare Source

Released on October 10, 2026.

@​logtape/logtape

• Fixed configureSync() rejecting synchronous disposable sinks and filters
  in browsers without Symbol.dispose or Symbol.asyncDispose, including
  Safari. Sinks and filters now retain their distinct cleanup hooks,
  preventing duplicate cleanup and asynchronous cleanup during resetSync().
  Sink wrappers preserve these hooks, and disposable sinks also work with
  using and await using transpiled with registry-symbol fallbacks
  (Symbol.for("Symbol.dispose") and Symbol.for("Symbol.asyncDispose")).
  Scoped configurations now also preserve cleanup hooks and correctly
  distinguish synchronous and asynchronous resources in withConfig()
  and withConfigSync(). [[#​255], [#​257]]

@​logtape/cloudwatch-logs

• Fixed CloudWatch Logs sinks losing their asynchronous cleanup hooks when
  Symbol.asyncDispose is unavailable. reset() now waits for buffered
  records to be flushed, and configureSync() correctly rejects these
  asynchronous sinks.
  [[#​255], [#​257]]

@​logtape/file

• Fixed file sinks losing their cleanup hooks when Symbol.dispose or
  Symbol.asyncDispose is unavailable. Synchronous file sinks now work with
  configureSync(), and reset() invokes and awaits asynchronous cleanup,
  flushing buffered records for non-blocking and stream file sinks.
  [[#​255], [#​257]]

@​logtape/otel

• Fixed OpenTelemetry sinks losing their asynchronous cleanup hooks in
  browsers without Symbol.asyncDispose, including Safari. reset() now
  waits for provider shutdown, and configureSync() correctly rejects these
  asynchronous sinks.
  [[#​255], [#​257]]

@​logtape/redaction

• Fixed redactByField() dropping sink cleanup hooks in browsers without
  Symbol.dispose or Symbol.asyncDispose, including Safari, preventing
  reset() from flushing or closing wrapped sinks. Wrapped sinks now
  preserve distinct synchronous and asynchronous cleanup hooks.
  [[#​255], [#​257]]
• Fixed redactByFieldAsync() losing its asynchronous cleanup hook when
  disposal symbols are unavailable. reset() now waits for pending
  redaction and wrapped sink cleanup, and configureSync() rejects these
  asynchronous wrappers.
  [[#​255], [#​257]]

@​logtape/syslog

• Fixed Syslog sinks losing their asynchronous cleanup hooks when
  Symbol.asyncDispose is unavailable. reset() now waits for pending
  messages before closing the connection, and configureSync() correctly
  rejects these asynchronous sinks.
  [[#​255], [#​257]]

@​logtape/windows-eventlog

• Fixed Windows Event Log sinks losing their synchronous cleanup hooks when
  Symbol.dispose is unavailable. configureSync() now accepts these
  sinks, and reset() and resetSync() release their Event Log resources.
  [[#​255], [#​257]]

---

Configuration

📅 Schedule: (UTC)

• Branch creation
  • At any time (no schedule defined)
• Automerge
  • Between 07:00 AM and 04:59 PM, only on Monday, Tuesday, Wednesday, and Thursday (* 7-16 * * 1,2,3,4)

🚦 Automerge: Enabled.

♻ Rebasing: Whenever PR becomes conflicted, or you tick the rebase/retry checkbox.

🔕 Ignore: Close this PR and you won't be reminded about this update again.

---

• If you want to rebase/retry this PR, check this box

---

This PR was generated by Mend Renovate. View the repository job log.

[coderabbitai]

Important

Review skipped

Bot user detected.

To trigger a single review, invoke the @coderabbitai review command.

⚙️ Run configuration

• Configuration used: Organization UI
• Review profile: CHILL
• Plan: Advanced
• Run ID: ea470370-6531-4002-a410-77379da5cbe3

---

You can disable this status message by setting the reviews.review_status to false in the CodeRabbit configuration file.

Use the checkbox below for a quick retry:

• 🔍 Trigger review

• Autofix · Keep fixing CodeRabbit findings and required CI, and resolving merge conflicts

---

_{Comment @coderabbitai help to get the list of available commands.}