fix(deps): update all non-major dependencies

[renovate]

ℹ️ Note

This PR body was truncated due to platform limits.

This PR contains the following updates:

Package	Change	Age	Confidence	Type	Update
github.com/jackc/pgx/v5	v5.10.0 → v5.11.0			require	minor
golang.org/x/oauth2	v0.36.0 → v0.37.0			require	minor
golang.org/x/time	v0.15.0 → v0.16.0			require	minor
node (source)	24.20.0 → 24.21.0			volta	minor

---

Release Notes

jackc/pgx (github.com/jackc/pgx/v5)

v5.11.0

Compare Source

This release adds direct PostgreSQL type scanning through database/sql on Go 1.27, improves compatibility with
libpq connection strings and PostgreSQL date/time values, and includes further decoder hardening. See Changes for
connection-string and date/time behavior changes that may affect existing applications.

Features

• stdlib: support Go 1.27's driver.RowsColumnScanner, allowing PostgreSQL types such as arrays and ranges to be
  scanned directly into Go values without pgtype.Map.SQLScanner. Existing database/sql scalar conversions and
  sql.Scanner behavior are preserved. The minimum supported Go version remains 1.25.
• Add Rows.TypeMap to expose the type map used to decode rows, including rows created by RowsFromResultReader
  that have no underlying Conn. Custom implementations of Rows, including mocks, must add this method.
• pgconn: add Config.MaxProtocolMessageBodyLen to configure the maximum incoming protocol message body size
  (carter-ya)
• pgconn: add ErrReadOnlyConnection, ErrReadWriteConnection, ErrPrimaryConnection, and ErrStandbyConnection
  sentinel errors for target_session_attrs validation, allowing callers to use errors.Is (Adrian-Stefan Mares)
• pgxpool: accept pool_ping_timeout in connection strings to configure Config.PingTimeout. The default is zero;
  zero and negative durations mean no timeout (1991santhu)

Changes

• Name-based row-to-struct mapping now matches explicit db tags case-insensitively, with exact matches taking
  precedence so tags can still distinguish quoted column names that differ only by case (AlisinaDevelo)

• pgconn: resolve the OS user account only when no user is supplied by the connection string, environment, or service
  file, avoiding unnecessary account lookups and crashes in some restricted container environments. Home-directory
  defaults for password, service, and TLS files remain available independently of the account lookup. On Unix these
  now use $HOME rather than the OS account's home directory (Mohamed MAACHE)

• pgtype: date, timestamp and timestamptz text values are now parsed and written by a hand-written parser and
  encoder for PostgreSQL's ISO date/time format instead of time.Parse and time.Format. Go's layout language cannot
  express a variable-width year or the BC era, which is the root of the bugs below. The text scan path is roughly 2.5x
  faster for timestamp and timestamptz. Bug fixes:

  • timestamp and timestamptz no longer silently move February 29 of a BC leap year to March 1 when encoding.
    time.Date(-4712, 2, 29, ...) was written as 4713-03-01 BC and is now written as 4713-02-29 BC. This
    affected ordinary four-digit BC years, not only extended-range ones. date was never affected.
  • timestamp and timestamptz can now scan BC leap days. 4713-02-29 BC previously failed with
    day out of range. date could already scan them.
  • Years past 9999 can now be scanned. 10000-01-02 03:04:05 previously failed to parse, so timestamp and
    timestamptz values at the high end of PostgreSQL's range were unreadable over the simple protocol and in any
    other text-format result.
  • time.Time arguments in the simple protocol now encode BC dates correctly, using the same timestamp encoder.
  • Fractional seconds beyond microsecond precision are rounded the way the server rounds them (round half to even,
    carrying into the rest of the value) instead of being kept at full precision. PostgreSQL never sends more than six
    fractional digits, so this only affects values from other sources.

  Behavior changes:

  • date now rejects impossible dates instead of normalizing them. 2024-02-30 returned 2024-03-01 and
    2024-13-01 returned 2025-01-01; both are now errors. timestamp and timestamptz already rejected them.
  • All three types now reject values outside PostgreSQL's range for that type, in the binary format as well as the
    text format. PostgreSQL never sends out-of-range dates, so this only affects corrupt or hand-built input; the range
    is checked in both formats so that whether a value is accepted does not depend on QueryExecMode.
    timestamptz also rejects time zone displacements outside PostgreSQL's signed 32-bit seconds range, while accepting
    the wider offsets emitted for POSIX time zones, such as +16.
  • timestamptz values scanned from the text format are now returned in time.Local, or in ScanLocation when it is
    set, matching what the binary format has always returned. Previously the text path kept whatever location
    time.Parse derived from the offset the server sent, so the same value scanned in the two formats could report a
    different Location() and Zone(). The instant is unchanged, but everything that renders the location changes
    with it: Timestamptz.MarshalJSON now writes the client's offset rather than the server's, so a value the server
    sent as +05:30 marshals as 2024-01-01T13:34:05-08:00 on a UTC-8 client instead of 2024-01-02T03:04:05+05:30,
    and DecodeDatabaseSQLValue hands database/sql a time.Time in that same location. Set the codec's
    ScanLocation to time.UTC to pin the location regardless of the client's zone.
  • Error messages from these paths have changed.

• pgconn: connection URIs (postgres://...) are now parsed by a new parser designed to exactly match libpq's URI
  parser behavior instead of net/url,
  making pgx accept and reject exactly the same URIs as libpq (verified by differential fuzzing against libpq itself).
  Most connection strings are unaffected. Edge-case behavior changes, all matching libpq:

  • + in query values is literal, no longer decoded as a space.
  • Malformed percent-encoding is a parse error instead of the parameter being silently dropped. %00 is rejected.
  • Leading/trailing spaces in URI components are trimmed; interior spaces are a parse error (encode them as %20).
  • # is ordinary data, not a fragment delimiter.
  • The userinfo terminator is the first @ before any / (previously the last @).
  • When a query parameter is repeated, the last occurrence wins (previously the first).
  • ssl=true is accepted as an alias for sslmode=require in URIs (JDBC compatibility). A repeated ssl key
    follows the same last-occurrence-wins rule as other repeated parameters, even across the rewrite to sslmode. If
    the final ssl value is not true, an independent explicit sslmode remains in effect.
  • Multiple hosts with mixed port specs are positionally aligned: postgres://h1,h2:5433/db now means h1:5432 and
    h2:5433 (previously both hosts got port 5433). A port list that is neither a single port nor exactly one port per
    host is an error (could not match N port numbers to M hosts), also for keyword/value connection strings.
  • An IPv6 address in a URI must be enclosed in brackets. A bare postgres://::1/db was previously accepted as host
    ::1; it is now read as an empty host followed by port :1 and fails with an invalid port error. Write it as
    postgres://[::1]/db.
  • Empty host list elements (e.g. h1,,h2) get the default host instead of being dropped. Likewise, an empty host in
    a keyword/value string (host=) now means the default host -- typically the Unix socket directory -- where it
    previously meant a TCP connection to an empty hostname.
  • An empty port (?port= in a URI or port= in a keyword/value string) now means the default port 5432 for the
    affected hosts; previously it was an invalid port error. Like any connection-string port, a present-but-empty port
    takes precedence over PGPORT.
  • ASCII control characters (tab, newline, ...) in a URI are ordinary data bytes, as they are to libpq; net/url
    rejected any URI containing one. The exception is a literal NUL byte, which is still rejected, as net/url did.
    (libpq never sees one -- C strings end at the first NUL -- but in Go a raw NUL could otherwise pass through into
    the NUL-delimited startup message and inject extra parameters.)

  Unlike libpq, unrecognized URI query parameters are still accepted (they become runtime parameters or pgx-specific
  options). Parse error messages avoid quoting the unredacted connection string and redact recognizable password
  fields on a best-effort basis. Invalid connection strings can be structurally ambiguous, so password redaction
  cannot be guaranteed for every malformed input.

• pgconn: keyword/value connection strings (host=... user=...) now match libpq's parser exactly, the same treatment
  the URI parser received above and verified the same way, by differential fuzzing against libpq itself. Most
  connection strings are unaffected. Behavior changes, all matching libpq:

  • A backslash escapes whatever character follows it and is dropped, where previously only \\ and \' were
    unescaped and every other backslash was kept. A value containing a backslash must now escape it, as libpq
    requires: sslcert=C:\path\to\cert reads as C:pathtocert and has to be written sslcert=C:\\path\\to\\cert.
    This mainly affects Windows certificate and key paths, which previously came through intact without doubling.
  • A trailing backslash in an unquoted value escapes the end of the string, so it is dropped and the value ends
    there; it was previously rejected with invalid backslash. Inside a quoted value the escaped terminator leaves
    the string unterminated, which is still an error.
  • Whitespace inside a keyword is an error (missing "=" after "us" in connection info string) instead of becoming
    part of the key. Whitespace around the = is unaffected. This most often shows up with an unquoted value
    containing a space: application_name=my app host=x previously set neither parameter and sent app host to the
    server as a runtime parameter, and now fails to parse.

  As with URIs, unrecognized keywords are still accepted where libpq rejects them, and an empty user= is still
  dropped so that PGUSER and the OS user still apply.

Fixes

• Keep the connection open after a recoverable PostgreSQL error from Begin or BeginTx
  (Victor Alejandro Sanz Ararat)
• Call TraceQueryEnd when Exec fails while deallocating invalidated cached statements (Chris Bandy)
• Deallocate a failed prepare using the statement name actually sent to the server, and skip cleanup if Parse never
  completed, avoiding leaked prepared statements and unnecessary cleanup errors (Eliran Ben-Zikri)
• Fix LoadTypes overwriting scalar codecs such as box and point with an incorrect ArrayCodec (Arsen Ozhetov)
• pgconn: retrieve field descriptions when cached descriptions are empty, such as for cursor FETCH statements,
  including batch and pipeline execution (water)
• pgconn: keep batch statement descriptions and result formats aligned when commands return no rows or when
  Batch.ExecStatement is mixed with other batch commands; preserve field descriptions for empty results
• pgconn: handle empty and comment-only queries in pipeline mode, discard stale statement data after bind errors,
  and return a nil result from Pipeline.GetResults on error
• pgconn: skip reading the password file when a password is already set (Jared Fowkes)
• pgxpool: treat non-positive MaxConnLifetime values as unlimited instead of immediately expiring connections
  (Aurelien Pillevesse)
• pgtype: support non-comma text array delimiters through ArrayCodec.Delimiter, including the semicolon delimiter
  used by box[]. LoadType and LoadTypes now load the delimiter from PostgreSQL (Sueun Cho)
• pgtype: quote text array elements containing internal whitespace (Louisa Huang)
• pgtype: quote and escape text range bounds containing delimiters, quotes, or backslashes, and distinguish empty
  string bounds from unbounded ranges (Sueun Cho)
• pgtype: preserve decimal precision in Numeric.ScanScientific and accept scientific notation in
  Numeric.UnmarshalJSON; reject out-of-range scientific exponents and preserve the original input in parse errors
  (Sueun Cho)
• pgtype: encode and decode numeric infinity in JSON as "Infinity" and "-Infinity" instead of encoding it as zero
  (Vladimir Saraikin)
• pgtype: treat a valid Numeric with a nil Int as zero in Int64Value, and return errors when converting NaN or
  infinity to an integer instead of panicking (Vladimir Saraikin)
• pgtype: fix an infinite loop when decoding binary numeric zero with a nonzero digit count (Vladimir Saraikin)
• pgtype: fix binary numeric digit-count overflow and trailing-byte handling. Binary encoding now rejects values
  whose digit count, weight, or scale cannot fit the wire format, while accepting the full unsigned digit-count range.
• pgtype: fix scanning through multiple pointer levels, including SQL NULL and XML values, and return an error
  instead of panicking when a pointer-to-pointer scan destination is nil (Rangel Reale)
• pgtype: use bounds-checked binary reads throughout the codecs and reject malformed lengths, counts, and trailing
  data. This includes fixes for panics on malformed records and truncated multiranges (Vladimir Saraikin), and
  validation of bit / varbit bit lengths against the actual data (g3m0sis).
• pgtype: return errors instead of panicking on malformed interval text (greymoth-jp), unterminated composite text
  fields, and text arrays whose dimensions and element counts disagree
• pgtype: cap the initial allocation estimate when parsing hstore text to avoid excessive allocation from unvalidated
  separator counts; valid hstores may still contain any number of pairs (AshSgDe29071999)
• pgtype: correct reversed bounds in integer scan error messages
• pgconn: a backslash as the last byte of a quoted value in a keyword/value connection string no longer panics with
  slice bounds out of range. host='a\ -- and the shorter ='\, reachable through pgx.ParseConfig and
  pgxpool.ParseConfig -- now return unterminated quoted string in connection info string, libpq's own message for
  the same input. The unquoted branch has been guarded since be69c1c; the quoted branch carried the same unguarded
  increment since the parser was ported from pgx v3. Found by fuzzing (Maxim Korotkov)
• pgconn: error messages that embed the connection string now also redact password and sslpassword values supplied
  as URI query parameters; previously only the userinfo password was redacted. Redaction matches keys the way the
  parser does -- percent-encoded spellings such as pass%77ord= are recognized -- and masks the entire raw value, so
  a password containing a space cannot leak its tail into the error message. Credentials stranded outside the
  userinfo by a malformed URI are masked whole, and invalid-port errors no longer embed the offending text (which in
  a malformed URI can be a mislaid password). Redaction of invalid connection strings is necessarily best effort:
  their structure may be ambiguous, so some malformed inputs can still expose password text in an error.
• pgconn: ParseConfigOptions.ConnStringAllowedKeys no longer exempts an explicitly supplied empty port (?port= in
  a URI or port= in a keyword/value string) from the allow-list. Only the implied all-empty port list of a
  multi-host URI without ports (postgres://h1,h2/db) is exempt. An explicit empty port shadows PGPORT even though
  it is empty, so it must be allowed like any other user-supplied key. The URI-only ssl=true alias is accepted when
  either ssl or sslmode is allowed, and every ssl/sslmode spelling written in the URI is validated --
  including occurrences superseded by later repeated parameters.
• pgconn: drain socket before close in asyncClose so context cancellation produces a TCP FIN instead of RST, avoiding "connection reset by peer" on the server / proxy (Sean Chittenden at CrowdStrike, Inc.)
• pgproto3: StartupMessage.Encode rejects a NUL byte in any parameter name or value instead of writing it. The
  startup message body is a run of NUL-delimited strings whose length is data-driven, so a NUL in a value ends that
  parameter and everything after it is read by the server as further parameters -- an application_name of
  x\x00user\x00admin changed the role the connection logged in as. libpq cannot reach this state because its
  parameters are NUL-terminated C strings. Connect now fails with nothing written to the wire, which covers
  settings that bypass connection string parsing: service files and direct assignment to Config.RuntimeParams,
  Config.User, or Config.Database.
• pgconn: keyword/value connection strings containing a NUL byte are rejected by ParseConfig, as URIs already were.

nodejs/node (node)

v24.21.0: 2026-09-08, Version 24.21.0 'Krypton' (LTS), @​aduh95

Compare Source

Notable Changes

• [71106e1f17] - crypto: update root certificates to NSS 3.126 (Node.js GitHub Bot) #​65495
• [afca0a912d] - (SEMVER-MINOR) crypto: support loading private keys through STORE loaders (Filip Skokan) #​63949
• [6274fccbd9] - deps: update OpenSSL to 3.5.8 (Node.js GitHub Bot) #​65542
• [53cba013c7] - deps: update Undici to 7.29.1 (Node.js GitHub Bot) #​65789
• [0529772798] - (SEMVER-MINOR) lib,src: improve histogram implementation (James M Snell) #​65024
• [41c7062b81] - (SEMVER-MINOR) net: improve performance of net.BlockList (James M Snell) #​64974
• [5197b5a3c5] - (SEMVER-MINOR) perf_hooks: add statistical hypothesis testing to histogram (James M Snell) #​65416
• [35c635b032] - (SEMVER-MINOR) util: add non-throwing MIMEType.parse (James M Snell) #​64965

Commits

• [84d706cb9b] - assert: improve documentation wording (Kamal Rawal) #​64953
• [90d127db33] - (SEMVER-MINOR) benchmark: add --analyze mode to compare.js (James M Snell) #​65416
• [ad1d7884c3] - benchmark: add test-only and mock timers cases (Luan Muniz) #​64097
• [1d8f045914] - benchmark: apply highWaterMark in webstreams pipe-to (Matteo Collina) #​65138
• [ed7ba3c993] - benchmark: complete the sqlite is-transaction fix (Edy Silva) #​65218
• [c8837e1aa3] - benchmark: add test runner hooks and options (Luan Muniz) #​63754
• [2ba8661e23] - buffer: prevent string write offset overflow (Matteo Collina) #​65043
• [cc9ee6c2ae] - buffer: treat detached ArrayBuffers as empty (Archkon) #​64504
• [5a5d73e4c5] - build: pass target architecture to small-icu genccode (ulofiai) #​65095
• [273e72d1a5] - build: deprecate always enabled --enable-static (Chengzhong Wu) #​65103
• [89a67246e3] - build: check FIPS option value in node.gyp (Filip Skokan) #​64982
• [2d21f41cd5] - build: handle malformed OpenSSL macros (Filip Skokan) #​64982
• [f62bc0f862] - build,win: add PGO workload scripts (Stefan Stojanovic) #​63696
• [33d0c7dc12] - child_process: keep SIGWINCH from killing on Win (Kirill Saied) #​64510
• [71106e1f17] - crypto: update root certificates to NSS 3.126 (Node.js GitHub Bot) #​65495
• [419af8b86d] - crypto: fix missing error checks on ASN1_STRING_to_UTF8() (Nora Dossche) #​65200
• [8029383f3f] - crypto: use available BoringSSL APIs (Filip Skokan) #​65423
• [a9bd780e19] - crypto: remove obsolete BoringSSL shims (Filip Skokan) #​65423
• [7defefad3f] - crypto: read WebCrypto inputs through primordials (Filip Skokan) #​65115
• [6ed1e38627] - crypto: fix disabling FIPS mode (Filip Skokan) #​64982
• [afca0a912d] - (SEMVER-MINOR) crypto: support loading private keys through STORE loaders (Filip Skokan) #​63949
• [9b9dd6e9cf] - debugger: wait for target startup (Filip Skokan) #​65194
• [07faaeeffd] - deps: update corepack to 0.36.0 (Node.js GitHub Bot) #​65653
• [53cba013c7] - deps: update undici to 7.29.1 (Node.js GitHub Bot) #​65789
• [0268ca547c] - deps: update archs files for openssl-3.5.8 (Node.js GitHub Bot) #​65542
• [6274fccbd9] - deps: upgrade openssl sources to openssl-3.5.8 (Node.js GitHub Bot) #​65542
• [6bdcd121fa] - deps: update zlib to 1.3.2.1-motley-8002e91 (Node.js GitHub Bot) #​65316
• [c2aa446b6d] - deps: update simdjson to 4.6.7 (Node.js GitHub Bot) #​65318
• [3e58e48ea8] - deps: update googletest to 49495ea (Node.js GitHub Bot) #​65317
• [670b3665c0] - deps: cherry-pick libuv/libuv@e640dc9 (ulofiai) #​65118
• [ca1c67b021] - deps: float ICU-23262 patch for icu78 (René) #​64678
• [4ad043b0aa] - deps: enable AVX-512 OpenSSL asm with clang (Daniel Lemire) #​65136
• [96b4af109b] - deps: update googletest to d89aac5 (Node.js GitHub Bot) #​65153
• [774f663c56] - dgram: don't swallow bind errors when callback is provided (armanmikoyan) #​62602
• [94b118d62e] - diagnostics_channel: validate before channel activation (Trivikram Kamat) #​65313
• [09788665bd] - dns: validate address type in lookupService (Lazizbek Ergashev) #​64878
• [15f95fc0e2] - dns: validate port range in setServers() (René) #​65021
• [37b9e9a154] - dns: fix crash on setServers with port 0 (Lazizbek Ergashev) #​65009
• [cd6205fa0d] - doc: update AHAFS reference link (Taeuk Ha) #​65481
• [0e6f9ae42e] - doc: fix property names in os.networkInterfaces() example (Jihwan) #​65469
• [034a827b41] - doc: fix broken links in cli.md (Donghoon Kang) #​65412
• [eb364621d4] - doc: remove outdated WASI version fallback (이혜미) #​65303
• [b13f425bf8] - doc: fix broken GYP link in n-api.md (Donghoon Kang) #​65413
• [45c4011067] - doc: document that an empty OPENSSL_CONF skips config loading (Orgad Shaneh) #​64949
• [fde6776c5f] - doc: fix broken TLS security level example (soreavis) #​65391
• [290c1fec04] - doc: clarify socket destroyed behavior (Dayun) #​65395
• [a7e8269947] - doc: update outdated nodejs.org guide links (Donghoon Kang) #​65394
• [7809f11249] - doc: clarify that ipv4 mapped to ipv6 are classified as ipv6 (Vedant Kulkarni) #​62117
• [64cd3a6e95] - doc: clarify how fs.Dirent file types are determined (soreavis) #​64532
• [9b92fdce14] - doc: update security release prepare command (Rafael Gonzaga) #​64699
• [6f9b9df3c1] - doc: clarify copyFile symlink behavior (T) #​62941
• [2480acb550] - doc: document setRawMode write access on Windows (Erik Demaine) #​63856
• [a1e9c3a5db] - doc: add missing return types in fs.md (Chaseton Collins) #​65307
• [4533572040] - doc: add missing return types in buffer.md (Yuya Inoue) #​65308
• [39ecedbbd2] - doc: fix lint clean command (greenhead) #​65274
• [0b1fb8fcd8] - doc: fix typo in onboarding.md (서울민트초코) #​65295
• [3165b5d38a] - doc: add missing added: tags to fs.lchmod (Lazizbek Ergashev) #​65283
• [113b808e59] - doc: fix SQLite changeset constant descriptions (greenhead) #​65265
• [c3eb51d5a1] - doc: document open pull request limit (Matteo Collina) #​65250
• [d7accdcd52] - doc: document http2 header constants (Harjoth Khara) #​64548
• [f47111416f] - doc: create ai-guidelines and include to CONTRIBUTING (Rafael Gonzaga) #​62105
• [c3b120e737] - doc: update synopsis (Augustin Mauroy) #​65171
• [39f4c831fd] - doc: fix broken internal links (greenhead) #​64901
• [be25cdd69e] - doc: report proper return type on urlPattern.test (Brian Muenzenmeyer) #​64831
• [1bf7737810] - doc: fix permission documentation examples (greenhead) #​64897
• [b7932e68a1] - doc: document sqlite parameter binding (Guilherme Araújo) #​65089
• [5234a5169c] - doc: finalize statements in sqlite examples (Guilherme Araújo) #​65088
• [39ea929da7] - doc: document quic stopSending() and resetStream() (Issac) #​64888
• [2ba198db73] - doc: clarify sqlite bare parameter default (Sumit Kumar Das) #​62009
• [314f9b200f] - doc: remove usage of util.inherits (Augustin Mauroy) #​60817
• [d82a61662c] - doc: fix grammar in worker_threads.md (이혜미) #​64913
• [44c0c8ff5b] - doc: clarify OpenSSL FIPS configuration (Filip Skokan) #​64982
• [9b2ca70e0d] - doc: remove --expose-gc flag from CLI documentation (Dario Piotrowicz) #​58909
• [a0a12397b9] - doc: document ArrayBuffer support in pbkd2Sync (kyungrae2002) #​64976
• [b48699e077] - doc: correct default highWaterMark values (Yilong Li) #​64617
• [0312ee133c] - esm: avoid super-linear data URL MIME regex (Sumit Kumar Das) #​61951
• [cd84d55c81] - esm: only register text format when enabled (Efe Karasakal) #​64992
• [c0a8ef611e] - esm: fix wasm import name in error message (이혜미) #​64950
• [e6c34f90c2] - events: inline iterationCondition hybrid dispatch closure (Szymon Łągiewka) #​64473
• [6ee4b40c91] - events: inline createEvent hybrid dispatch closure (Szymon Łągiewka) #​64473
• [367549eed5] - fs: use sized reads for large files in readFileUtf8 (Shelley Vohr) #​65328
• [8a5b1ae4c2] - fs: fix realpath of namespaced drive paths (Jason Zhang) #​65378
• [c9233b950d] - fs: fix glob early return skipping sibling entries (Srinu desetti) #​64895
• [bc54dd8905] - fs: pass symlink type in cp when filter is provided (Jerry Zhao) #​62654
• [fcb4333aca] - fs: allocate FSReqPromise stat arrays lazily (Samuel Attard) #​63886
• [c35876154e] - fs: fix out-of-bounds write in mkdtemp for long prefixes (Hierax_Umbra) #​64770
• [b269616936] - fs: treat std::errc::permission_denied as EPERM error (Kirill Saied) #​64698
• [212fe77e76] - fs: add windowsHandle option to file streams (Kirill Saied) #​63851
• [75df6cb435] - http: improve performance with known-length calls to end() (Tim Perry) #​65466
• [48d9cd4a28] - http: cache maxHeaderPairs per header section (GetThatCookie) #​64988
• [04785c8f43] - http: fix keylog listener setup on existing agent sockets (Shani Singh) #​65066
• [4b90031534] - http: emit drain on socket takeover and avoid stale HWM reuse (Naman Trivedi) #​64991
• [83a27559cd] - http2: adapt receive deferral for Node.js 24 (Matteo Collina) #​65093
• [f43bed0ecc] - http2: avoid uaf while receiving and sending rst_stream (esgor) #​64166
• [b42d664321] - inspector: avoid calling into JS from V8 interrupts (Joyee Cheung) #​65028
• [6bf852197d] - lib: use bracket notation instead of startsWith/endsWith for single char (Taejin Kim) #​61500
• [151ca7e104] - lib: harden webidl dictionary member reads (Filip Skokan) #​65115
• [7e8c2c9f44] - lib: use validateArray for array arguments (greenhead) #​64959
• [424fe2bc5a] - lib: add and test [EnforceRange] in webcrypto dictionaries (Filip Skokan) #​65091
• [0529772798] - (SEMVER-MINOR) lib,src: improve histogram implementation (James M Snell) #​65024
• [186e1b76e8] - meta: move targos to emeritus (Michaël Zasso) #​65393
• [b41b07c72a] - meta: add unified http api initiative (James M Snell) #​65139
• [f85b6ecd67] - meta: move one or more collaborators to emeritus (Node.js GitHub Bot) #​65182
• [8f3d01bdce] - meta: add Aviv Keller to .mailmap (Aviv Keller) #​65048
• [cfad1d5b28] - meta: update sccache to 0.17.0 (René) #​64985
• [349c53c441] - module: report unreadable package.json (Paul Bouchon) #​65223
• [bd21e6706e] - module: cache nearest parent package.json per directory (Shelley Vohr) #​65326
• [961bd04370] - module: fix --check on ambiguous ESM files (Paul Bouchon) #​65203
• [f45ef73420] - net: handle undefined parent in _unrefTimer and _destroy (Shivay-98) #​64644
• [41c7062b81] - (SEMVER-MINOR) net: improve performance of net.BlockList (James M Snell) #​64974
• [5197b5a3c5] - (SEMVER-MINOR) perf_hooks: add statistical hypothesis testing to histogram (James M Snell) #​65416
• [1722ddac28] - permission: enforce addon permission in GetLinkedBinding (Rafael Gonzaga) #​65432
• [dff2b675db] - process: validate resource stats array offsets (Archkon) #​65098
• [f277983e7b] - quic: changes for nghttp3_conn_close_stream2 (Marten Richter) #​64574
• [a4c770c78e] - quic: mark drain promise handled (James M Snell) #​65319
• [2460b171c5] - quic: reset rejected HTTP/3 request streams with H3_REQUEST_REJECTED (trivenay) #​65442
• [18a7ccf302] - quic: write desired size needs update on maxstream (Marten Richter) #​64768
• [9017f4a780] - quic: do not destroy incoming streams that have a consumer (trivenay) #​65335
• [ddc41c1ef4] - quic: fix wake up blob (Marten Richter) #​64044
• [88bee43d7c] - quic: convert incoming :status header to number (Hallison Pereira Melo) #​63589
• [cd776fe97c] - quic: fix infinite loop if STOP_SENDING received on a buffering stream (Tim Perry) #​64715
• [4f6eda3c23] - repl: keep entries added while history file is loading (Mhayk Whandson) #​64513
• [cd1e6ce29b] - repl: add benchmarks (Aviv Keller) #​64590
• [2ba740669d] - sea: avoid dangling CLI option pointers (Archkon) #​64755
• [ec5e2d6856] - sea: handle NUL bytes in asset keys (Archkon) #​64773
• [703b854293] - sea: reject trailing content in config JSON (Archkon) #​64774
• [a61a5fdd1c] - sqlite: prevent reentrant session.close() (Trivikram Kamat) #​65349
• [6ae81be0a3] - sqlite: reject statement-less SQL in prepare() (Trevor Burnham) #​65157
• [043dfe4996] - sqlite: reject statement-less SQL in SQLTagStore (Trevor Burnham) #​65157
• [b8faee02e1] - sqlite: check null returns from sqlite value functions (Nora Dossche) #​63288
• [d6d2a71bee] - sqlite: validate maxSize argument in createTagStore() (Anshika Jain) #​63792
• [61a046309f] - sqlite: reject non-positive backup rates (Trivikram Kamat) #​64893
• [514e3f30fb] - sqlite: clear SQLTagStore bindings (Matteo Collina) #​65041
• [c1542255b8] - sqlite: bind Boolean (mike-git374) #​62001
• [cdb732beb5] - sqlite: fix undefined behaviour in Session::Changeset() (Nora Dossche) #​63637
• [fbe8861111] - sqlite: bind ArrayBuffer (mike-git374) #​62061
• [e3c6bd6bc8] - src: add missing vector include (Filip Skokan) #​65622
• [793cf69df8] - src: fix heap value deduplication in embedder graph (Ilyas Shabi) #​64801
• [ea5935b04c] - src: fix out-of-bounds write when transcoding odd-length ucs2 (nashit hayat) #​64512
• [22e5023f4d] - src: escape Windows environment variables in task runner (Antoine du Hamel) #​65217
• [a0f3c62bd9] - src: simplify c++ diagnostics channel API (James M Snell) #​65158
• [8e831a3d2e] - src: make minor cleanup to permission checks (James M Snell) #​65158
• [968b2ca3a3] - src: use DictionaryTemplate for permission diag channel message (James M Snell) #​65158
• [18e16e7f8d] - src: cache permission strings (James M Snell) #​65158
• [c8625a4b6f] - src: add SetAbortHandler (Max H Fisher) #​64684
• [c990140d60] - src: match cmd.exe case-insensitively in task runner (Archkon) #​64907
• [d7463b9dbc] - src: reuse cached env strings in remaining files (Seongeun Lee) #​65039
• [b055a43b93] - src: expose Windows-only fs open flags (Kirill Saied) #​64775
• [7f21e37496] - src: report why --enable-fips failed (Filip Skokan) #​64979
• [8c11beae27] - src: update repeated use strings to env (James M Snell) #​64760
• [9e2c477e26] - stream: normalize fused stateless transform results (Trivikram Kamat) #​65367
• [107e96dd41] - stream: encode whole chunks in TextEncoderStream (Matteo Collina) #​65414
• [164068279b] - stream: prevent share from eagerly draining source (Trivikram Kamat) #​65338
• [93d822bdc1] - stream: drain pending writes before broadcast end (Trivikram Kamat) #​65334
• [6b8b9c362f] - stream: reuse unexposed managed read buffers (GetThatCookie) #​64990
• [4ec4fab367] - stream: avoid duplicated endReadableNT scheduling (Matteo Collina) #​65310
• [86d1196ebf] - stream: decouple transform backpressure changes (Matteo Collina) #​65143
• [b8a7a75b19] - stream: reject pull on signal abort during flush (Trivikram Kamat) #​65346
• [386ed6a06d] - stream: avoid leaking consumers on signal failure (Trivikram Kamat) #​65299
• [74d53bd73b] - stream: use validateObject for zlib/iter params (greenhead) #​65015
• [3a174ce16b] - stream: use validateNumber for BYOB reader options.min (greenhead) #​65014
• [859ea01cb2] - stream: consolidate non-op algorithm callbacks (Matteo Collina) #​65138
• [c577669825] - stream: cut promise churn in webstreams hot paths (Matteo Collina) #​65138
• [d631e910db] - stream: preserve falsy cancellation reasons (Trivikram Kamat) #​64705
• [f9c21eabbd] - stream: use validateBuffer for BYOB reader view (greenhead) #​65046
• [b89c8f5d1e] - stream: fix recursive WritableStream abort (Jeong SeokChan) #​64825
• [39e0457e86] - test: fix link-local dgram scope assertion (Filip Sko

❗ Important

✂ PR body was truncated to here.

---

Configuration

📅 Schedule: (in timezone Asia/Jakarta)

• Branch creation
  • "on saturday"
• Automerge
  • At any time (no schedule defined)

🚦 Automerge: Disabled by config. Please merge this manually once you are satisfied.

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

👻 Immortal: This PR will be recreated if closed unmerged. Get config help if that's undesired.

---

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

---

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