• docs/v322_new.md src/sbbs

    From Rob Swindell (on Windows@VERT to Git commit to main/sbbs/m on Fri Jul 17 03:05:00 2026
    https://gitlab.synchro.net/main/sbbs/-/commit/9fc5505cbc4dfc9aeaab00ff
    Modified Files:
    docs/v322_new.md src/sbbs3/ctrl/build.bat src/sbbs3/jsexec.vcxproj ntsvcs.vcxproj sbbscon.vcxproj
    Log Message:
    Windows: build all 32-bit Synchronet hosts /LARGEADDRESSAWARE

    Mark sbbs.exe (sbbscon.vcxproj), sbbsNTsvcs.exe (ntsvcs.vcxproj),
    jsexec.exe (jsexec.vcxproj), and sbbsctrl.exe (ctrl/build.bat) as large-address-aware, raising their usable address space from 2GB to
    ~4GB on 64-bit Windows.

    This is a stopgap for GitLab #1185: an access violation in mozjs185
    seen only on 32-bit Windows under JS heap pressure (a web-scrape
    attack), never on Linux-x64. It raises the ceiling; it does not fix
    the underlying pointer corruption, which the SM128 / 64-bit migration
    addresses properly.

    The MSVC projects get <LargeAddressAware>true</LargeAddressAware> in
    both Debug|Win32 and Release|Win32 link steps. Borland's ilink32 has
    no LAA option, so build.bat flips the PE-header bit post-build
    instead.

    Validated: the 2006-era Borland host runs stably under LAA with
    system-wide top-down allocation (Synchronet DLLs executing above 2GB,
    all servers listening, no errors), and jsexec exercised the mozjs GC
    and trace-JIT with ~3.4GB live above 2GB without failure. Sustained
    behavior under real scrape load is not yet proven.

    Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

    ---
    þ Synchronet þ Vertrauen þ Home of Synchronet þ [vert/cvs/bbs].synchro.net
  • From Rob Swindell (on Windows@VERT to Git commit to main/sbbs/m on Sat Jul 18 22:02:00 2026
    https://gitlab.synchro.net/main/sbbs/-/commit/3a5e9cb3628031f8812840c5
    Modified Files:
    docs/v322_new.md src/sbbs3/scfg/scfg.c scfg.h scfgmsg.c
    Log Message:
    Restore SCFG import/export of subs.txt, alongside subs.ini

    subs.txt import/export was removed in mayor-19-caps (e6acb3269a) and its
    menu slots later repurposed for *subs.ini in reef-31-feed (cfb32e03f5).
    Some sysops still rely on the legacy text format (especially for import),
    so this restores it as an additional format while keeping *subs.ini as
    the preferred format going forward.

    Message Areas > Export/Import Areas list subs.txt again, the -import=
    command line recognizes a subs.txt filename once more, and the
    line-oriented parser/writer are back.

    Rather than the old memcpy-over-the-existing-sub (which zeroed every
    field the format doesn't carry - qwk_conf, print modes, area_tag, ...),
    the import now copies only the fields subs.txt actually provides, leaving
    the rest of an over-written sub intact. ptridx is likewise preserved
    (the exported value is meaningless on the importing system).

    Partially addresses issue #1128.

    Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

    ---
    þ Synchronet þ Vertrauen þ Home of Synchronet þ [vert/cvs/bbs].synchro.net
  • From Rob Swindell (on Debian L@VERT to Git commit to main/sbbs/m on Sun Jul 19 15:25:00 2026
    https://gitlab.synchro.net/main/sbbs/-/commit/8a79c1ff830f811d824cced5
    Modified Files:
    docs/v322_new.md src/sbbs3/nodedefs.h
    Log Message:
    Document node-rerun-on-recycle behavior; fix stale NODE_RRUN comment

    Add a v3.22a release note for the terminal-server change that flags
    active nodes to re-read their configuration when a recycle is signaled,
    so configuration changes reach new callers without waiting for every
    node to disconnect.

    Also correct the NODE_RRUN definition comment. Its only remaining
    effect in the threaded terminal server is to make a node re-read its configuration on the next logon; nothing is "re-run" and there is no logoff-time action (unlike its neighbors NODE_EVENT and NODE_DOWN).
    The old "Re-run this node when log off" wording described the DOS/OS2 separate-process node model that no longer exists.

    Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

    ---
    þ Synchronet þ Vertrauen þ Home of Synchronet þ [vert/cvs/bbs].synchro.net
  • From Rob Swindell (on Debian L@VERT to Git commit to main/sbbs/m on Fri Jul 24 11:16:00 2026
    https://gitlab.synchro.net/main/sbbs/-/commit/04b47329aa0cfbc839ec1a4c
    Modified Files:
    docs/v322_new.md src/sbbs3/zmodem.c
    Log Message:
    zmodem: fix SIGFPE when the transmit window is under 4 blocks (#1197)

    zmodem_send_from() picks the ack-request interval as

    subpkts_sent % (max_window_size / block_size / 4)

    which is a division by zero whenever the window is narrower than four
    blocks: with -8 -w8192 the divisor is 8192 / 8192 / 4, that is 1 / 4,
    that is 0. sexyz then dies with SIGFPE right after the ZRQINIT/ZRINIT handshake, having put about 3 KB of handshake on the wire and no file
    data at all. Also affected: -8 -w16384, -2 -w4096, and the same window
    sizes set through MaxWindowSize in sexyz.ini.

    Clamp the interval to a minimum of one, so a window narrower than four
    blocks requests an ACK on every subpacket, which is the only sensible
    reading of "every quarter window" at that size.

    Verified against lrz over the bench harness: -8 -w16384 and -2 -w4096
    both used to die on signal 8 and now complete with a matching SHA-256, at
    3.08 and 0.71 MB/s. Windows that already worked are unchanged, -w32768
    at 4.88 MB/s and -w65536 at 5.03, as is streaming without -w at 11.6.

    SyncTERM shares zmodem.c but never sets max_window_size, so it leaves the
    value at zero, short-circuits the branch, and cannot reach this. The
    exposure is sexyz only.

    A window exactly equal to the block size (-8 -w8192) no longer crashes
    but is still pathologically slow, roughly one subpacket per second with a one-second receive timeout each iteration. That is tracked separately in
    the issue; it is not a regression, since it used to be a crash.

    zmodem_ver bumped 2.3 -> 2.4. Release notes updated, including the already-shipped 2 GB windowed fix and Deuce's send-path speedups, which
    had no v322 entry.

    Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

    ---
    þ Synchronet þ Vertrauen þ Home of Synchronet þ [vert/cvs/bbs].synchro.net
  • From Rob Swindell (on Debian L@VERT to Git commit to main/sbbs/m on Fri Jul 24 23:28:00 2026
    https://gitlab.synchro.net/main/sbbs/-/commit/aaf394a0d833acd89a64d3a9
    Modified Files:
    docs/v322_new.md src/sbbs3/sexyz.c
    Log Message:
    sexyz: keep the transmit window >= 4x the block size, like lsz (#1197)

    sexyz's -w set only the ZMODEM transmit window and left the block size independent (from -2/-4/-8, default 1 KB), so a window at or below the
    block size was configurable -- and it stalled the transfer to roughly one window per second. lsz and Forsberg sz never allow this: when -w is
    given they force the block down to a quarter of the window
    (blkopt = Txwspac = Txwindow/4), guaranteeing window >= 4*block so the ack-every-quarter-window scheme always has several blocks in flight.

    Do the same after argument parsing, so -8 -w8192 and -w8192 -8 both
    resolve the same way: floor the window to 256, round to a multiple of 64,
    and reduce max_block_size to window/4 when it exceeds it (and the start block_size with it). A one-line notice reports the reduction.

    Effect: the formerly-hanging window<=block configurations now complete --
    -8 -w8192 reduces the block to 2048 and transfers, -4 -w4096 to 1024,
    and so on -- while windows that already left four blocks of headroom
    (-w32768 and up with -8) are untouched and unchanged (4.75 and 4.88 MB/s
    here, streaming still 11.3). This also makes the divide-by-zero guarded
    in 04b47329aa unreachable by configuration, since window/block/4 is now
    always >= 1.

    This does not make sexyz's windowed throughput competitive with lsz --
    at the same block and window sexyz is still many times slower, and very
    small windows (which force tiny blocks) remain slow with occasional
    one-second stalls. That is the separate sexyz windowed-send overhead
    tracked in #1195; this change is only about not permitting the degenerate window<=block configuration that lsz forbids by construction.

    sexyz.c version 3.3 -> 3.4. No zmodem.c change, so SyncTERM is
    unaffected (it never sets a transmit window anyway).

    Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

    ---
    þ Synchronet þ Vertrauen þ Home of Synchronet þ [vert/cvs/bbs].synchro.net
  • From Rob Swindell (on Windows@VERT to Git commit to main/sbbs/m on Sat Jul 25 22:24:00 2026
    https://gitlab.synchro.net/main/sbbs/-/commit/b17b1d3e3e45bbbc4df47ae6
    Modified Files:
    docs/v322_new.md src/sbbs3/atcodes.cpp getkey.cpp js_bbs.cpp putmsg.cpp sbbs.h terminal.h
    Log Message:
    Unicode @-codes: return the expansion instead of printing it

    @U+XXXX@, @CHECKMARK@, @ELLIPSIS@, @COPY@, @SOUNDCOPY@, @REGISTERED@, @TRADEMARK@, @DEGREE_C@, @DEGREE_F@ and @WIDE:text@ wrote their glyph
    straight to the terminal from inside atcode() and returned an empty
    string. Format modifiers were therefore applied to "": @CHECKMARK-R10@
    emitted the glyph and then right-aligned nothing in a 10-column field.
    Callers that expand an @-code into a buffer for a non-printing reason -- bbs.atcode(), bbs.expand_atcodes(), format_text(), mnemonics(),
    putnode(), notify() and putmsg()'s @CENTER@ -- got an empty expansion
    plus stray output on the current node's terminal.

    These codes now encode into the caller's buffer and set P_UTF8, so show_atcode()'s existing display-width compensation makes alignment and truncation work. formatted_atcode() and expand_atcodes() take an
    optional pmode out-param so the buffer-expansion callers can learn the
    result is UTF-8; @CENTER@ and mnemonics() use it, the latter needing the
    P_UTF8 branch bputs() already has, since it emits byte-by-byte through outchar(). Terminal::center() measures with bstrlen(str, mode & P_UTF8):
    the other code sets bstrlen() recognizes are translated by putmsg(), not bputs(), and a centered line bypasses putmsg()'s character loop.

    The |W (doubled) modifier prints a UTF-8 expansion as-is rather than
    feeding it to wide(), which would double each byte rather than the glyph. Output buffers grew to fit @WIDE:, which triples byte length.

    Introduced in 9374bfe37e (pill-4-pencil), a9eea57362 (begun-3-ended) and 43c793a14b (hold-20-maps).

    Fixes #1198

    Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>

    ---
    þ Synchronet þ Vertrauen þ Home of Synchronet þ [vert/cvs/bbs].synchro.net
  • From Rob Swindell (on Windows@VERT to Git commit to main/sbbs/m on Sat Jul 25 22:24:00 2026
    https://gitlab.synchro.net/main/sbbs/-/commit/dbc654505f7d822a8ece7c0d
    Modified Files:
    docs/v322_new.md src/sbbs3/con_out.cpp terminal.cpp
    Log Message:
    Count UTF-8 columns once, in Terminal::utf8_increment()

    outcp() wrote the codepoint with term_out() and then advanced the column itself, but term_out() -> parse_output() -> Terminal::utf8_increment()
    had already counted it -- so every codepoint outcp() emitted to a UTF-8 terminal advanced the column counter by twice the glyph's display width.
    The counter ran ahead of the cursor, making word-wrap, right-margin
    truncation, centering and the auto-pause line counter wrap early on any
    line containing such text. The call was also unconditional while
    term_out() returns early under CON_ECHO_OFF, so the column advanced for suppressed output too.

    utf8_increment() is now the only place that counts, and it passes sbbs->unicode_zerowidth rather than a hardcoded 0, preserving the
    zero-width quirk the ZWNBSP probe detects at logon. That applies the
    detected value to all UTF-8 output, bputs() included, not just outcp().

    outcp()'s inc_column() was correct when written in 9374bfe37e
    (pill-4-pencil): it paired with putcom(), which does no column
    accounting. It became a double-count in 374282bf81 (skins-28-shop),
    which switched the write to term_out() and introduced utf8_increment().

    Fixes #1200

    Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>

    ---
    þ Synchronet þ Vertrauen þ Home of Synchronet þ [vert/cvs/bbs].synchro.net
  • From Rob Swindell (on Windows@VERT to Git commit to main/sbbs/m on Sat Jul 25 23:32:00 2026
    https://gitlab.synchro.net/main/sbbs/-/commit/005460da2b098c8ec796464a
    Modified Files:
    docs/v322_new.md src/sbbs3/useredit.cpp
    Log Message:
    Uedit: newline before the non-alpha commands that prompt

    getkeys() runs with K_NOCRLF and only alpha commands got a newline
    before their handler, so the '~' (leech), '+' (adjust credits), '*'
    (adjust minutes), '$' (credits) and '/' (AR string) prompts were echoed
    onto the tail of the "User edit (?=Menu) (n of m): " line.

    Present since ca499c9550 (silence-13-chorus) introduced the alpha-only
    newline.

    Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>

    ---
    þ Synchronet þ Vertrauen þ Home of Synchronet þ [vert/cvs/bbs].synchro.net
  • From Rob Swindell (on Debian L@VERT to Git commit to main/sbbs/m on Tue Jul 28 18:06:00 2026
    https://gitlab.synchro.net/main/sbbs/-/commit/58dc4d2b39ca0c227d5427aa
    Modified Files:
    docs/v322_new.md src/sbbs3/js_global.cpp src/xpdev/ini_file.h
    Log Message:
    Correct the .local rationale: ctrl/ is not overwritten by an upgrade

    The commit that added the .local filename variation justified it as protecting a sysop's edits from being overwritten by an upgrade. That is not true of the ctrl directory, which is where most of the configuration files it applies to live: ctrl is populated when the BBS is installed and an upgrade leaves it alone. The Windows upgrade script copies exactly one file into it, text.dat.

    What an upgrade does refresh is the distributed program directories -- exec, text and docs unconditionally, and xtrn when the sysop accepts the installer's "Upgrade Externals" option -- so the overwrite argument holds for a file we ship in one of those, and for any installation run from a source checkout, where updating means pulling the whole tree.

    The variation is still worth having for a ctrl file; the honest reason is separation rather than rescue. Reworded accordingly on the wiki (config:ini_files, howto:multihost) and in the release notes, where the claim was made. xpdev is a general-purpose library, so its header documents the search order and what .local means and leaves the rationale to those.

    Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>

    ---
    þ Synchronet þ Vertrauen þ Home of Synchronet þ [vert/cvs/bbs].synchro.net
  • From Rob Swindell (on Debian L@VERT to Git commit to main/sbbs/m on Tue Jul 28 19:54:00 2026
    https://gitlab.synchro.net/main/sbbs/-/commit/d91e670aaee817fea9820e9e
    Modified Files:
    docs/v322_new.md src/sbbs3/con_out.cpp str_util.c str_util.h terminal.h Log Message:
    bprintf: measure %s field widths in columns rather than bytes

    printf counts a field width in bytes, but a terminal renders columns.
    For a UTF-8 argument the two differ, so a byte-counted field prints
    short and whatever follows it slides left. Every columnar format string
    sent to a UTF-8 terminal is affected; message listings are the most
    visible case.

    Rewrite the numbers instead of the formatting. fmt_col_widths() walks
    the conversions in a format string, and for each %s carrying a literal
    width or precision re-states them in bytes such that printf's
    truncate-then-pad arrives at the requested number of columns: the new
    precision is the byte count of the leading N columns, and the new width
    is padded by the byte-minus-column difference. vsnprintf still does all
    the conversion. Sysop-customized text.dat files therefore keep working untouched, and output to a non-UTF-8 terminal is unchanged, the pre-pass running only for P_UTF8.

    A '*' width or precision is left alone: atcode() computes those itself
    and has already accounted for the difference, so rewriting them would compensate twice. An unrecognized conversion, a NULL argument, or a
    rewrite that would overflow the buffer all return the format string
    unmodified.

    Widths are measured with Terminal::bstrlen(), which already knows what
    occupies a column -- Ctrl-A codes, the PCBoard, Wildcat, Renegade,
    Celerity and WWIV sequences, backspace, and UTF-8 code-point widths.
    It gains an optional column limit and a byte-count out-parameter, which
    is what lets a field be truncated on a column boundary without splitting
    a multi-byte sequence.

    Issue #1204

    Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>

    ---
    þ Synchronet þ Vertrauen þ Home of Synchronet þ [vert/cvs/bbs].synchro.net
  • From Rob Swindell (on Debian L@VERT to Git commit to main/sbbs/m on Fri Jul 31 21:43:00 2026
    https://gitlab.synchro.net/main/sbbs/-/commit/72303693ef1b8dc2c11ff006
    Modified Files:
    docs/v322_new.md src/sbbs3/websrvr.cpp
    Log Message:
    websrvr: enforce download credits for unauthenticated file requests

    An anonymous request through the file-area virtual path
    (FileVPathPrefix) was authorized by user_can_download() alone, with no
    credit check, so every file in a directory with empty ARS downloaded
    free of charge regardless of its cost. The FTP server applies the
    credit gate to anonymous users, and issue #446 was closed with that
    being the intended policy.

    Both authorization paths now share check_file_download(), which applies
    the credit gate before the ARS check. Putting the two callers on one
    predicate is the point of the helper: the bypass existed because the
    anonymous branch and the authenticated one had drifted apart.

    The gate measures cost against user_available_credits() rather than
    user.cdt, so a user's free credits now count toward a download, as they
    already do on the Terminal and FTP servers, and as subtract_cdt()
    already assumed when charging: it spends free credits first.

    A refused anonymous request is logged at LOG_DEBUG. An insufficient
    credit refusal now reports NotEnoughCredits; both this and the existing authenticated failure log previously reported CantDownloadFromDir for
    it, naming the wrong cause.

    Fixes #1192

    Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>

    ---
    þ Synchronet þ Vertrauen þ Home of Synchronet þ [vert/cvs/bbs].synchro.net
  • From Rob Swindell (on Debian L@VERT to Git commit to main/sbbs/m on Sat Aug 1 20:33:00 2026
    https://gitlab.synchro.net/main/sbbs/-/commit/4de1032086f038890a75559e
    Modified Files:
    docs/v322_new.md src/sbbs3/websrvr.cpp
    Log Message:
    websrvr: load a file's cost before checking download credits (#1192)

    The credit gate added in 72303693ef (2026-07-31) compared against a
    cost that was always zero. resolve_vpath() loaded the file record at file_detail_index, and smb_getfile() skips smb_getmsghdr() entirely at
    that detail level, while cost is only ever set from the SMB_COST header
    field. Every comparison against it therefore passed: costed files
    downloaded free of charge for authenticated users as well as anonymous
    ones, while user_downloaded_file() still billed the account afterward
    (it reloads the record itself at file_detail_normal), so a user could
    go credit-negative.

    Read the header record when the directory's files can cost something. A
    free directory has none to report, since loadfile() zeroes the cost and download_is_free() short-circuits on the same DIR_FREE flag before any
    credit comparison, so the index record alone still suffices there and loadfile() is not asked to open the base twice.

    Verified against a live file base: an unauthenticated request for a 2,639,434-credit file is refused with 401, a zero-cost file in the same non-free directory returns 200, and a file in a free directory
    transfers in full.

    Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>

    ---
    þ Synchronet þ Vertrauen þ Home of Synchronet þ [vert/cvs/bbs].synchro.net
  • From Rob Swindell (on Debian L@VERT to Git commit to main/sbbs/m on Sat Aug 1 23:40:00 2026
    https://gitlab.synchro.net/main/sbbs/-/commit/7a0dabc5613b3f1a5e4ff8be
    Modified Files:
    docs/v322_new.md src/sbbs3/js_file.cpp
    Log Message:
    Parse an .ini once in File.iniGetAllObjects()

    It asked for each section's keys by name, and that lookup scans the file
    from the top -- so the cost was O(sections * lines). SyncArcade's lobby
    reads games.ini on every run for the cabinet titles, and at 4740 sections
    that read took two seconds on an idle machine; a loaded one paid seven,
    which reads as the lobby rescanning its ROMs when it is doing nothing of
    the kind.

    Parse the file into per-section lists once instead, the way scfglib1.c
    and scfglib2.c already read the configuration: 2005ms -> 66ms on that
    file. The section list still comes from iniGetSectionList(), so a
    duplicate section name is still collapsed to its first occurrence.

    Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>

    ---
    þ Synchronet þ Vertrauen þ Home of Synchronet þ [vert/cvs/bbs].synchro.net
  • From Rob Swindell (on Windows@VERT to Git commit to main/sbbs/m on Sun Aug 9 07:19:00 2026
    https://gitlab.synchro.net/main/sbbs/-/commit/d02c34e2d43c073fb16fc610
    Modified Files:
    docs/v322_new.md src/sbbs3/xtrn_sec.cpp
    Log Message:
    Populate CHAIN.TXT lines 24 and 25 (call time and time used)

    These two WWIV drop-file fields have been hard-coded to 0 since the
    initial v3.00c check-in (7e3e47141a, 2000-10-10). Line 24 is the time
    of the call expressed as seconds past midnight; line 25 is the number
    of seconds the user has been on the system so far, so doors that show
    "time on" -- or recover the current time from the pair -- saw zeros.

    Line 25 uses timeon() rather than a raw now-starttime subtraction: it
    measures against the same logon timestamp line 24 is derived from, so
    the two stay consistent, and it clamps a negative result instead of
    printing it through an unsigned conversion.

    Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>

    ---
    þ Synchronet þ Vertrauen þ Home of Synchronet þ [vert/cvs/bbs].synchro.net
  • From Rob Swindell (on Windows@VERT to Git commit to main/sbbs/m on Sun Aug 9 07:19:00 2026
    https://gitlab.synchro.net/main/sbbs/-/commit/abb64a33b74f9fcbe37e57ec
    Modified Files:
    docs/v322_new.md src/sbbs3/pcbdefs.hpp xtrn_sec.cpp
    Log Message:
    Fill in three PCBOARD.SYS fields left empty by the drop-file writer

    The struct-based PCBOARD.SYS writer added in 973b7f0a35 (2022-03-13)
    never assigned these, so each was written as whatever the value-
    initialized struct held:

    - Allowed K-Bytes for Download (offset 63) was 0, which tells a door
    the caller may download nothing. It now carries the user's
    available credits in kilobytes, the same source DOOR.SYS line 31
    and CALLINFO.BBS line 16 already use, saturated at 32767 -- the
    value PCBoard reads as "unlimited".

    - Conference Area user was in (offset 65) was 0. It now carries the
    current message group. The group number is used rather than the
    sub-board number (which offset 142 already carries) because it is
    an index into the user's group list and so always fits in a byte.

    - Language Extension (offset 80) was four NUL bytes; PCBoard writes
    four spaces when no alternate language is selected. The default is
    set in the structure definition alongside the other fields whose
    empty value is not zero.

    Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>

    ---
    þ Synchronet þ Vertrauen þ Home of Synchronet þ [vert/cvs/bbs].synchro.net
  • From Rob Swindell (on Debian L@VERT to Git commit to main/sbbs/m on Sun Aug 9 23:03:00 2026
    https://gitlab.synchro.net/main/sbbs/-/commit/2c49eafa1f3c5bfce980fbe4
    Modified Files:
    docs/v322_new.md src/sbbs3/xtrn_sec.cpp
    Log Message:
    Report GR in DOOR.SYS line 20 whenever ANSI is supported

    Line 20 required both ANSI and COLOR, so a caller using an ANSI
    terminal in monochrome mode was told "NG" and doors ran with ANSI
    disabled entirely.

    The monochrome distinction belongs to line 39 ("ANSI supported but
    NG mode"), which is still computed, but no door kit consults it.
    OpenDoors and xpdoor (both in this tree), MBSE's doorlib, and the
    Turbo Pascal DoorDriver Plus kit all decide ANSI solely from line
    20 -- DoorDriver Plus closes the file immediately after reading it,
    and doorlib names line 39 "allwaysN". Writers agree: WWIV emits GR
    on ANSI alone, and the published field definitions describe line 39
    as "always set to N".

    The ANSI-plus-COLOR test dates to the initial v3.00c check-in and
    was never revisited. 554c2ca395 (days-21-coins, 2020-12-11) moved
    it from useron.misc to the negotiated terminal flags for this same
    symptom -- doors losing ANSI for an ANSI-capable caller -- but kept
    the COLOR requirement.

    Affects both the 52-line and 31-line drop-file forms; line 20 is in
    the section common to both.

    Closes #1218

    Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>

    ---
    þ Synchronet þ Vertrauen þ Home of Synchronet þ [vert/cvs/bbs].synchro.net
  • From Rob Swindell (on Debian L@VERT to Git commit to main/sbbs/m on Mon Aug 10 03:17:00 2026
    https://gitlab.synchro.net/main/sbbs/-/commit/b8332a4618efd80300b671d6
    Modified Files:
    docs/v322_new.md src/sbbs3/mailsrvr.cpp mailsrvr.h
    Log Message:
    Treat the submission ports as an MSA, not as port 25 (#1221)

    The message submission ports (587, and 465 for implicit TLS) ran the same
    code path as the SMTP transfer port, so nothing downstream could tell the
    two apart. Three consequences, all reported by Deuce:

    - An authenticated user submitting an original message to an external
    recipient was judged by the port-25 rule and refused as an illegal
    relay unless the sysop enabled ALLOW_RELAY, a server-wide option.
    - The submission ports accepted unauthenticated mail, contrary to
    RFC 6409 section 4.3, which requires a 530 error response to MAIL on
    an unauthenticated session.
    - Once ALLOW_RELAY was set, every sender check was skipped for an
    authenticated session, so a user could submit with any From: address,
    including one at a domain the system does not own. The SMTP envelope
    sender was already derived from the user record, but the RFC822 From:
    header is emitted verbatim, and is now also DKIM-signed.

    Listening sockets previously passed their protocol name through xpms as
    the callback data, and the accept loop recovered each socket's properties
    by comparing that pointer against the name it was created from. Those comparisons work only because the names differ: the compiler pools
    identical string literals, so the transfer and submission agents, both
    named "SMTP", share one pointer and cannot be told apart. Pass a
    descriptor instead, carrying the display name alongside the properties
    the accept loop wants, so identity no longer depends on the name.

    An authenticated user sending to an external address via a submission
    port is a submission rather than a relay and no longer consults
    ALLOW_RELAY, which continues to govern the transfer port. The G and M restrictions still apply on every port, so a sysop retains per-user
    control over outbound mail.

    A sender address is authorized when its domain is one of this system's
    and it resolves to the authenticated user. Both questions are answered by smtp_resolve_user(), extracted from the RCPT TO handler, so any address
    that reaches a user -- alias, real name, alias.cfg entry, sub-address
    tag, user number -- also works as their sender address, and stays correct
    as the sysop edits alias.cfg. An alias resolving off-system is a forced
    relay rather than a local address, so it authorizes nobody, and the
    DefaultUser catch-all is suppressed: it exists to make unknown recipients deliverable, and honoring it would let that one account send as any
    address at the system's domains. MAIL FROM and the From: header are both checked. A null reverse-path is accepted, per RFC 6409 section 3.2.

    A refusal suggests the account's own address rather than prescribing it,
    since several addresses may be acceptable and the likeliest cause is a
    mail client configured with a misspelling of the sender's own address.

    The transfer port is unaffected in every respect, so only sysops who have deliberately enabled a submission port see any change.

    Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>

    ---
    þ Synchronet þ Vertrauen þ Home of Synchronet þ [vert/cvs/bbs].synchro.net
  • From Rob Swindell (on Debian L@VERT to Git commit to main/sbbs/m on Tue Aug 18 22:04:00 2026
    https://gitlab.synchro.net/main/sbbs/-/commit/f8d0bd2f64b441e91fe3e662
    Modified Files:
    docs/v322_new.md src/sbbs3/mailsrvr.cpp
    Log Message:
    mailsrvr: honor ALLOW_RELAY for foreign submission sender addresses

    Since b8332a4618 (chart-3-cause, 2026-08-10), the submission ports verify
    that an authenticated user's MAIL FROM and From: address is at one of the system's domains and resolves to that user. That rejects the sysop who
    points a mail client at another BBS as a smarthost and submits as their own domain -- which worked on 587 before that commit, and still works on port 25 wherever ALLOW_RELAY is set.

    Both of Vertrauen's authenticated submissions on 2026-08-18 were refused
    this way, one of them from a domain whose SPF record already authorizes Vertrauen to send for it. The reply named an address the submitter had no reason to want, so it read as a problem with the server hostname rather than with the sender address.

    A sender address in a foreign domain is a relay, so honor the permission the sysop already granted for relaying: with ALLOW_RELAY enabled, accept one
    from an authenticated user who is not restricted from relaying (G/M). An address in one of the system's own domains must still resolve to the authenticated user, so this is no license to forge a local address -- the
    case that would lend the system's DKIM reputation to the forgery.

    RFC 6409 section 6.1 (Enforce Submission Rights) is a MAY, in a section
    titled "Optional Actions", so declining to enforce it is conformant. The authentication requirement of section 4.3, a MUST, is unaffected.

    Issue #1221 asked for sysop-selectable strictness here. ALLOW_RELAY is a
    coarse stand-in for that -- a dedicated option or an ARS (issue #107) would express it properly -- but it does pick out the systems that mean it.

    Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>

    ---
    þ Synchronet þ Vertrauen þ Home of Synchronet þ [vert/cvs/bbs].synchro.net
  • From Rob Swindell (on Debian L@VERT to Git commit to main/sbbs/m on Tue Aug 25 01:12:00 2026
    https://gitlab.synchro.net/main/sbbs/-/commit/222dbfc7f6fd9921e2754adc
    Modified Files:
    docs/v322_new.md src/sbbs3/sexyz.c
    Log Message:
    sexyz: expose the ZMODEM file-management option, and -e for ESCCTL

    The engine has carried management_protect / management_clobber / management_newer for years and honors all three in zmodem_send_zfile(),
    but nothing ever set them -- not sexyz, not SyncTERM -- so every send
    fell through to the default and announced ZF1_ZMCRC ("transfer if
    different CRC or length"). The protocol work was already done; only
    the plumbing was missing.

    Add SendManagement to sexyz.ini's [ZMODEM] section, taking crc (the
    default, unchanged), clobber, protect or newer. One enumerated key
    rather than three booleans, because the engine treats them as mutually exclusive in a fixed precedence and three independent flags could be
    set to a combination that silently resolves to one of them.

    On the command line, -p and -n are new and match lsz. -y keeps its
    current meaning when receiving -- may we overwrite a file already here
    -- and gains the matching one when sending: tell the receiver to
    overwrite its own. That is exactly the lrzsz split, where lsz -y sets
    ZMCLOB and lrz -y clobbers locally, and -y was previously documented as applying only to receives, so nothing that worked before changes. Note
    the two act on different disks: receiving, the destination is this
    machine; sending, it is the user's. Synchronet's own receiver ignores
    a ZFILE's management field entirely, so this reaches third-party
    terminals only.

    Also add -e, the command-line equivalent of the existing
    EscapeCtrlChars ini key, for links that do not pass control characters
    intact. Verified against lrzsz: receiving with -e, the wire grows from 4,312,067 to 5,245,286 bytes for the same 4 MB file, within 0.005 % of
    what lrz -e negotiates from the same sender, and verifies
    byte-identical. It took the hex-header terminator fix in the preceding
    commit to work at all.

    When sending, -e is subject to the receiver's ZRINIT, which is what the
    engine derives escape_ctrl_chars from, so the usage text says the
    receiver's request governs rather than promising something the protocol
    layer overrides. Sending with ESCCTL actually negotiated is broken for
    a separate, pre-existing reason noted in the preceding commit.

    Options parse after the ini is read, so a flag overrides the file;
    verified with SendManagement=protect in the ini and -y on the command
    line yielding "overwriting destination".

    ---
    þ Synchronet þ Vertrauen þ Home of Synchronet þ [vert/cvs/bbs].synchro.net
  • From Rob Swindell (on Debian L@VERT to Git commit to main/sbbs/m on Tue Aug 25 01:12:00 2026
    https://gitlab.synchro.net/main/sbbs/-/commit/fd038d9dd3cba68b4d95d026
    Modified Files:
    docs/v322_new.md src/sbbs3/zmodem.c
    Log Message:
    zmodem: escape CR unconditionally when ESCCTL is negotiated

    A sender with ESCCTL negotiated escaped every control character
    except carriage return, which it routed to a conditional rule --
    escape only when the previous byte sent was '@' -- and otherwise put
    on the wire raw. A receiver that requested ESCCTL discards unescaped
    control characters by definition, so it discarded those CRs, every
    subpacket carrying one failed its CRC, and the transfer never
    advanced: the sender ZRPOSed to zero until it spent its error budget
    and gave up. With random data that is any subpacket at all. A
    download to any receiver requesting ESCCTL was therefore impossible,
    not merely slow.

    lrzsz settles what the rule should be. In zsendline_init() (zm.c) the
    table entry for 015/0215 is:

    if (Zctlesc) tab = 1; /* always escape */
    else if (!turbo_escape) tab = 2; /* the conditional '@' rule */
    else tab = 0;

    So the conditional rule is the NON-ESCCTL case, and ESCCTL escapes CR
    like any other control character. This engine had the two swapped:
    the conditional rule was reachable only with ESCCTL on, where it is
    wrong, and never reachable with ESCCTL off, where it would be right.
    Since it could not fire correctly in either state, drop it -- fold CR
    into the control class and let the escape mask decide, which is
    exactly lrzsz's behaviour for both states. The '@' rule can be added
    properly later if wanted; the last_sent field it read is left in
    place, being public.

    The unescaped path is provably untouched: without ESCCTL neither the
    old CR class nor the new one is in the active mask, so CR is a normal
    byte either way. Confirmed byte-for-byte -- a 4 MB send still puts
    4,311,816 bytes on the wire, 256 MB still receives at 484.99 MB/s and
    sends at 206.24 MB/s, and the 3e-6 corruption gate still passes 5 of 5 receiving and 3 of 3 sending.

    With the fix, ESCCTL interoperates in every direction: sexyz to lrz,
    lrz to sexyz, and sexyz to itself all verify byte-identical. The
    strongest evidence it is now correct rather than merely accepted is
    the wire size -- sexyz puts 5,245,034 bytes on the wire where lsz puts 5,245,030 for the same file, so the two escape the same byte set.
    Random data exercises every one of the 256 values, so nothing is
    escaped by luck.

    Under injected errors the escaped path is a little less robust than
    the plain one: 4 of 5 at 3e-6 and 5 of 5 at 1e-6, against 5 of 5 for
    the unescaped path at 3e-6. It carries 25 % more bytes and so takes
    25 % more corruption at a given per-byte rate. No lrzsz baseline
    exists for that comparison -- lsz -e to lrz -e does not survive the
    harness's error injection at all, dying inside 0.04 s on every
    attempt.

    ---
    þ Synchronet þ Vertrauen þ Home of Synchronet þ [vert/cvs/bbs].synchro.net
  • From Rob Swindell (on Windows@VERT to Git commit to main/sbbs/m on Sat Sep 5 15:45:00 2026
    https://gitlab.synchro.net/main/sbbs/-/commit/fb3a6fff74c45ea1da0f8fdc
    Modified Files:
    docs/v322_new.md src/sbbs3/logon.cpp
    Log Message:
    Fix "Multiple Nodes" (G) user exemption being ignored at logon

    The logon-time check for a user already being on another node tests the
    user's G exemption ("Multiple Nodes"), but 02792b752c (2025-09-08)
    mechanically replaced that test with the new useron_is_guest() helper,
    which reads the G *restriction* (Guest) instead. Every other conversion
    in that commit was of a restriction test, so this lone exemption test
    changed behavior despite the commit's "No functional change" claim.

    The result, shipped in v3.21e: a user granted the G exemption was still disconnected with "You can't possibly be in two places at the same
    time", while Guest accounts gained an exemption they never had. Stock
    Guest accounts set both the G restriction and the G exemption (see exec/makeguest.js), so the swap was invisible there - only non-guest
    users holding the exemption saw the breakage.

    Fixes issue #1234.

    Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>

    ---
    þ Synchronet þ Vertrauen þ Home of Synchronet þ [vert/cvs/bbs].synchro.net
  • From Rob Swindell (on Windows@VERT to Git commit to main/sbbs/m on Sat Sep 5 18:40:00 2026
    https://gitlab.synchro.net/main/sbbs/-/commit/64a15d67481a164b1db24e14
    Modified Files:
    docs/v322_new.md src/sbbs3/scansubs.cpp
    Log Message:
    Bug-fix: don't ask "Your Messages Only" during scan configuration for QWK nodes

    1f8a468b15 (2005-06-24) suppressed this prompt for QWK network nodes at
    the per-sub site but left the "all subs in group" branch testing
    useron.misc instead of useron.rest. FLAG('Q') is bit 16, which is
    unassigned in user.misc, so that test was always false and QWK nodes
    were still asked.

    Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>

    ---
    þ Synchronet þ Vertrauen þ Home of Synchronet þ [vert/cvs/bbs].synchro.net
  • From MRO@VERT/BBSESINF to Rob Swindell (on Windows on Sun Sep 6 02:35:00 2026
    Re: docs/v322_new.md src/sbbs3/scansubs.cpp
    By: Rob Swindell (on Windows 11) to Git commit to main/sbbs/master on Sat Sep 05 2026 06:40 pm

    https://gitlab.synchro.net/main/sbbs/-/commit/64a15d67481a164b1db24e14 Modified Files:
    docs/v322_new.md src/sbbs3/scansubs.cpp
    Log Message:
    Bug-fix: don't ask "Your Messages Only" during scan configuration for QWK nodes


    Why is it reporting your os version in your from: line?
    Is this a new feature?

    --
    https://i.imgur.com/MNKA5U0.png
    President of BBS Sysop's Union +++ https://bbses.info/union
    ---
    þ Synchronet þ ::: BBSES.info - free BBS services :::
  • From Digital Man@VERT to MRO on Mon Sep 7 14:41:00 2026
    Re: docs/v322_new.md src/sbbs3/scansubs.cpp
    By: MRO to Rob Swindell (on Windows on Sun Sep 06 2026 02:35 am

    Why is it reporting your os version in your from: line?

    My git commit messages have the OS of the system I made the commit from.

    Is this a new feature?

    No, it's been that way for years:
    https://git.synchro.net/gitpushlog.ssjs
    --
    digital man (rob)

    This Is Spinal Tap quote #40:
    Morty the Mime: Come on, don't talk back, mime is money, come on, move it. Norco, CA WX: 85.9øF, 60.0% humidity, 1 mph NNW wind, 0.04 inches rain/24hrs ---
    þ Synchronet þ Vertrauen þ Home of Synchronet þ [vert/cvs/bbs].synchro.net
  • From MRO@VERT/BBSESINF to Digital Man on Mon Sep 7 19:58:00 2026
    Re: docs/v322_new.md src/sbbs3/scansubs.cpp
    By: Digital Man to MRO on Mon Sep 07 2026 02:41 pm

    Re: docs/v322_new.md src/sbbs3/scansubs.cpp
    By: MRO to Rob Swindell (on Windows on Sun Sep 06 2026 02:35 am

    Why is it reporting your os version in your from: line?

    My git commit messages have the OS of the system I made the commit from.

    Is this a new feature?

    No, it's been that way for years:
    https://git.synchro.net/gitpushlog.ssjs


    Oh, okay. I've never noticed it.

    --
    https://i.imgur.com/MNKA5U0.png
    President of BBS Sysop's Union +++ https://bbses.info/union
    ---
    þ Synchronet þ ::: BBSES.info - free BBS services :::
  • From Rob Swindell (on Debian L@VERT to Git commit to main/sbbs/m on Tue Sep 15 20:47:00 2026
    https://gitlab.synchro.net/main/sbbs/-/commit/b7f7d2b3f506c3e449640733
    Modified Files:
    docs/v322_new.md src/sbbs3/chksmb.c writemsg.cpp src/smblib/smbadd.c smballoc.c smbdefs.h smblib.h smbtxt.c
    Log Message:
    smblib: raise the message data storage ceiling from 2GB to 4GB

    smb_allocdat() and smb_fallocdat() validated a data offset by testing (int)offset < 0, truncating a 64-bit off_t to 32 bits. Every allocation
    at or above 2GB read as negative and was refused with SMB_ERR_DAT_OFFSET,
    so a .sdt file could not grow past 2GB even though the format stores that offset in a uint32_t and so reaches 4GB - 1.

    The same truncation failed open at the top: at exactly 4GB, (int)offset
    is 0 and passes, and the header is then written with
    (uint32_t)4294967296 == 0, leaving the message's data at 4GB and its
    header pointing at the start of the file.

    Bound the end of the allocation against the format's own limit instead,
    named SMB_MAX_DAT_OFFSET/SMB_MAX_HDR_OFFSET, and repeat the check in smb_addmsg() where the offset is narrowed to uint32_t, which is the only
    guard for a hyper-allocated base. The data-length limit smb_addmsg()
    already enforced becomes SMB_MAX_DAT_LEN.

    Offsets on the way there were themselves 32-bit: ftell() is a 32-bit long
    on Windows, chsize() arguments were cast to int (one an int * int
    multiply that overflows at 2GB while truncating the data file), and the
    message text was read with a bare fseek() of hdr.offset + dfield.offset computed in 32-bit unsigned arithmetic.

    issue #12

    Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>

    ---
    þ Synchronet þ Vertrauen þ Home of Synchronet þ [vert/cvs/bbs].synchro.net
  • From Rob Swindell (on Debian L@VERT to Git commit to main/sbbs/m on Tue Sep 15 20:47:00 2026
    https://gitlab.synchro.net/main/sbbs/-/commit/aee485f478d18bf1aa97cb59
    Modified Files:
    docs/v322_new.md src/sbbs3/smbutil.c
    Log Message:
    smbutil: make pack error-safe

    packmsgs() renames the live .shd/.sdt/.sid aside before building their replacements, and three failure paths afterwards took a bare return with
    no message at all. A single unreadable message therefore left the base
    with no .shd, which a message server then treats as a base to create from scratch.

    Log those failures with the offending message number, skip a message
    whose data cannot be read rather than abandoning the pack, and restore
    the originals on any failure after the rename. The allocation tables are
    moved aside rather than truncated in place so that rollback is complete: against an emptied .sda, the next message written would land on top of
    live data. The replacements are flushed and checked before the swap, so a
    full disk no longer installs a truncated base, and the swap is now the
    last thing done rather than interleaved with the removals.

    Also stop discarding messages whose data exceeds 16MB. That limit was
    arbitrary and dropped the message from the base rather than leaving it uncompacted; the bound is now the format's own SMB_MAX_DAT_LEN.

    issue #1171

    Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>

    ---
    þ Synchronet þ Vertrauen þ Home of Synchronet þ [vert/cvs/bbs].synchro.net
  • From Rob Swindell (on Windows@VERT to Git commit to main/sbbs/m on Sun Sep 20 04:19:00 2026
    https://gitlab.synchro.net/main/sbbs/-/commit/b1771282ea73a88f08df07d0
    Modified Files:
    docs/v322_new.md src/sbbs3/sbbs_ini.c
    Log Message:
    Default POP3Interface/TLSInterface to their own server's Interface

    [Mail] POP3Interface and [Web] TLSInterface fell back to [Global]
    Interface when unset, rather than to the Interface of the server
    section they belong to. A [Mail] Interface of "0.0.0.0,::" with a
    [Global] Interface of "0.0.0.0" left SMTP, submission and SMTPS
    listening on both stacks while POP3 bound IPv4 only, so POP3 clients
    resolving the host to its AAAA record got connection refused.

    Also add the missing POP3Interface write block: the setting was read
    but never written back, and compare TLSInterface against the web
    server's own interfaces so the writer's omit-when-default rule matches
    the reader.

    Introduced in dbbfabf1b1 (2015-08-20).

    Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>

    ---
    þ Synchronet þ Vertrauen þ Home of Synchronet þ [vert/cvs/bbs].synchro.net
  • From Rob Swindell (on Windows@VERT to Git commit to main/sbbs/m on Sun Sep 20 04:19:00 2026
    https://gitlab.synchro.net/main/sbbs/-/commit/71736412815af6bc2f071367
    Modified Files:
    docs/v322_new.md src/sbbs3/filedat.c sbbsdefs.h src/sbbs3/scfg/scfgxfr1.c src/sbbs3/xtrn.cpp
    Log Message:
    Don't let a filename defeat the quoting of an external command line

    cmdstr() double-quotes the values it substitutes for %a, %o, %f and %~,
    but only when the value contained a space, and sh expands $ and a
    back-tick even inside double quotes. So a file-base filename could
    still reach a command shell carrying characters the shell interprets: ILLEGAL_FILENAME_CHARS already excludes " and \, but not $, `, &, (,
    ), ' or the glob characters. File viewers are the exposed path --
    their ex_mode defaults to EX_STDIO|EX_SH in scfglib2.c, so a
    sysop-configured viewer command containing %f runs through sh -c or
    cmd /c.

    Widen quoted_string()'s trigger to SHELL_QUOTE_CHARS, the characters a
    shell interprets outside double quotes but not inside them, and keep SHELL_EXPANSION_CHARS ($ and a back-tick) out of uploaded filenames in allowed_filename(), since no amount of quoting protects those.

    The restriction is deliberately in allowed_filename() rather than ILLEGAL_FILENAME_CHARS: illegal_filename() hacklogs its hits and is
    applied to names already stored in the file base and to names arriving
    over FTP, in message attachments and in QWK packets, so widening it
    would turn existing legal filenames into logged attack attempts.
    $ and a back-tick are legal on Windows and are not cmd.exe
    metacharacters, but file bases are shared between platforms, so a name
    uploaded here eventually reaches a POSIX host.

    Also teach the *nix argv splitter about those quotes. It splits on
    spaces only and never stripped a quote, so ever since cmdstr() began auto-quoting in 58bfc11aa7 (2005-09-02) a %f filename containing a space
    has reached a non-shell program as two arguments with literal "
    characters embedded. The splitter itself dates to e3a47718311
    (2003-02-28) and predates the quoting by two years. Windows is
    unaffected: its external() hands the whole command line to
    CreateProcess() and the child's CRT parses the quotes.

    The SCFG help text spells the two characters out rather than
    interpolating SHELL_EXPANSION_CHARS, because a back-tick in a uifc
    helpbuf toggles highlighting instead of being displayed.

    Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>

    ---
    þ Synchronet þ Vertrauen þ Home of Synchronet þ [vert/cvs/bbs].synchro.net
  • From Rob Swindell (on Debian L@VERT to Git commit to main/sbbs/m on Tue Sep 29 15:57:00 2026
    https://gitlab.synchro.net/main/sbbs/-/commit/86ff63dd979c2efb0b58dd88
    Modified Files:
    docs/v322_new.md src/sbbs3/date_str.c getnode.cpp main.cpp
    Log Message:
    Dates, times, and node activity follow the user's language

    A terminal session's copy of the configuration pointed at the system's
    default text strings, so the weekday and month names, "Never", and the
    node status/activity text formatted through it ignored the user's text.<lang>.ini translations. Point it at the session's own text
    strings.

    Displayed times also take their am/pm from the NScanAmQ and NScanPmQ text strings (" am"/" pm", whose leading space is the separator) instead of hard-coded English, as does the node list's download ETA. (#1265)

    Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

    ---
    þ Synchronet þ Vertrauen þ Home of Synchronet þ [vert/cvs/bbs].synchro.net
  • From Rob Swindell (on Debian L@VERT to Git commit to main/sbbs/m on Tue Sep 29 15:57:00 2026
    https://gitlab.synchro.net/main/sbbs/-/commit/778acdf7ae7fe34d37077bd7
    Modified Files:
    docs/v322_new.md src/sbbs3/str.cpp
    Log Message:
    New-scan time entry: re-prompt for an hour of 1-12 without am/pm

    On a 12-hour-clock system, a bare hour of 1-12 is ambiguous: after
    seeing "12 am", typing "12" means midnight to the user but noon as a
    24-hour value. Reject it and re-prompt with the hour and the original
    time's am/pm pre-filled (e.g. "3" becomes "3 pm"), so Enter accepts it or
    the suffix can be changed. Other invalid input re-prompts with the
    original value. A bare 0 or 13-23 is still accepted as a 24-hour value; 24-hour-clock systems are unchanged. (#1265)

    Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

    ---
    þ Synchronet þ Vertrauen þ Home of Synchronet þ [vert/cvs/bbs].synchro.net
  • From Rob Swindell (on Debian L@VERT to Git commit to main/sbbs/m on Wed Sep 30 16:22:00 2026
    https://gitlab.synchro.net/main/sbbs/-/commit/3a602f2c3d109d110579ec35
    Modified Files:
    docs/v322_new.md src/sbbs3/getmsg.cpp readmail.cpp readmsgs.cpp sbbs.h src/smblib/smbhash.c smblib.h
    Log Message:
    Verify the title before acting on a subject-CRC match in the readers (#1208)

    The "search title forward/backward" commands (> and <) of the sub-board
    and mail readers decided that two messages share a title by comparing the 16-bit subject CRC stored in the index, and jumped to the first CRC match without ever reading its header. A CRC-16 collision therefore silently
    landed the user on a message with an unrelated title, and the real next
    match was skipped. The same omission gated a non-sub-op's view of deleted
    posts on the raw from-name CRC.

    Treat the index CRC as the candidate filter it is, the way the rest of
    these files already use the name CRCs: on a CRC hit, read the header and compare the actual strings.

    - New smb_subject_match() in smblib compares two subjects with the same
    normalization smb_subject_crc() applies (leading "RE:" prefixes, letter
    case and trailing white-space ignored); the RE:-skipping is now shared
    by both functions
    - New sbbs_t::msg_subject_matches() does the CRC pre-check, then loads the
    candidate header and compares subjects; all four title-search loops use
    it
    - The deleted-post visibility check for non-sub-ops confirms the author by
    name after the CRC match, mirroring the private-post check below it

    Reproduced on a scratch terminal server with three posts whose first and
    second titles share a CRC-16 (only the third truly matches the first):
    before, > from message 1 landed on message 2; after, it lands on message 3
    and < returns to message 1, in both the sub-board and the mail reader. smb_subject_match() unit-checked against RE:, case, white-space and NULL inputs. The deleted-post change was not exercised.

    Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>

    ---
    þ Synchronet þ Vertrauen þ Home of Synchronet þ [vert/cvs/bbs].synchro.net
  • From Rob Swindell (on Debian L@VERT to Git commit to main/sbbs/m on Wed Sep 30 16:22:00 2026
    https://gitlab.synchro.net/main/sbbs/-/commit/04dfc61739938207f3f302df
    Modified Files:
    docs/v322_new.md src/sbbs3/sbbsecho.c
    Log Message:
    SBBSecho: don't write a stored-message header into an inbound packet

    import_netmail() marks a netmail message as received by writing its stored-message (*.msg) header back to the file it was read from. Since e7b93ae3f3 (types-26-ease, 2025-11-01) that write-back ran for packets
    too: update_fmsghdr() seeks to offset 0 and writes the 190-byte fmsghdr_t, which lands on the packet header and the first message. Any message that followed the imported netmail in the same packet was then read through the overwritten bytes (its type word became 256, from the destzone/origzone
    bytes of the stored header), reported as "Grunged message", and the packet
    was set aside as bad, losing the rest of its netmail and echomail. Before
    that commit the write-back was conditional on a *.msg path.

    Only write the header back for a *.msg file, at both of the sites that
    were made unconditional. The robot (uses_msg) site already branched on it.

    Reproduced with a hand-built Type-2+ packet carrying two netmail messages:
    the second was grunged at offset 177 with type 256 before the fix, and both import cleanly after it. Shipped in SBBSecho 3.37 with v3.21c and v3.21e.

    Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>

    ---
    þ Synchronet þ Vertrauen þ Home of Synchronet þ [vert/cvs/bbs].synchro.net
  • From Rob Swindell (on Debian L@VERT to Git commit to main/sbbs/m on Wed Sep 30 16:22:00 2026
    https://gitlab.synchro.net/main/sbbs/-/commit/d7d1b58fd16bf8c7a9cab1fe
    Modified Files:
    docs/v322_new.md src/sbbs3/sbbsecho.c
    Log Message:
    SBBSecho: generate a Message-ID for imported netmail too (#1167)

    fmsgtosmsg() generated an RFC822 Message-ID only for messages going into
    a sub-board (added in 0eccdade66, 2005-09-01, for echomail cross-
    referencing); the mail-base branch never did. Every netmail SBBSecho
    imports was therefore stored without one, and SBBSecho's own locally
    originated notifications (e.g. "FAILED Area Management Request from ...",
    which have no net origin) are exactly what chksmb flags as "No Message-ID",
    as reported in the issue.

    Generate the default Message-ID after the branch, for netmail and echomail alike, the same way the BBS does for every message it stores.

    Verified on a scratch install by importing a packet holding a plain netmail
    and an AreaFix request from an unconfigured node: both the imported netmail
    and the resulting failure notification now carry a Message-ID and chksmb reports the mail base OK.

    Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>

    ---
    þ Synchronet þ Vertrauen þ Home of Synchronet þ [vert/cvs/bbs].synchro.net
  • From Rob Swindell (on Debian L@VERT to Git commit to main/sbbs/m on Wed Sep 30 19:08:00 2026
    https://gitlab.synchro.net/main/sbbs/-/commit/aadcf82e0660f29d0b7afc10
    Modified Files:
    docs/v322_new.md src/sbbs3/websrvr.cpp
    Log Message:
    Web Server: log script errors, don't send them to the client

    Fixes #1203

    js_ErrorReporter() wrote every uncaught JavaScript error, with the script's absolute path and line number, into the HTTP response body as well as the
    log, and the request still completed with a 200 and the page's content
    type. That disclosed the server's layout to any visitor who could make a
    script throw, and any error message that interpolated request data (a query parameter, header or cookie) was a reflected script-injection vector in the BBS's own origin.

    Log the error only. A reported error (a warning does not count, nor does exit(), which reports nothing) marks the session, and exec_ssjs() then
    fails the request instead of sending what the script produced before
    failing, so the normal 500 error path answers it. An error after the
    script has already sent its headers can only be logged, as before.

    Verified on a scratch web server: a page that throws with a query
    parameter in the message returns 500 with an empty body and the error is
    in the log; a page that ends with exit() still returns 200.

    Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>

    ---
    þ Synchronet þ Vertrauen þ Home of Synchronet þ [vert/cvs/bbs].synchro.net
  • From Rob Swindell (on Debian L@VERT to Git commit to main/sbbs/m on Wed Sep 30 19:08:00 2026
    https://gitlab.synchro.net/main/sbbs/-/commit/cf6d568170a047dbb6647208
    Modified Files:
    docs/v322_new.md src/sbbs3/atcodes.cpp exec.cpp execmisc.cpp inkey.cpp sbbs.h xtrn.cpp
    Log Message:
    cmdstr(): bound the output by the caller's buffer size

    Fixes #1191

    cmdstr() limited its output to the size of its own member buffer even when writing into a caller-supplied one whose size it was never told, so a
    caller with a smaller buffer could be overflowed. @TYPE: and @INCLUDE: did exactly that, expanding into atcode()'s output buffer with a bound of 511.

    Add an overload that takes the output buffer's size (with the mode as a required argument, so a call passing a NULL buffer and a mode can never
    resolve to it by accident) and route the unsized form through it with the member buffer's size. Every caller that passes its own buffer now passes
    its size too: atcode() hands over the maxlen it already had, and the Baja interpreter, the hot-key handler and execmisc pass sizeof their arrays.
    An expansion that does not fit is now logged as truncated instead of
    silently cut.

    Verified on a scratch terminal server: a short @INCLUDE: still displays the file through the sized path, and a 400-character one is handled within
    the caller's buffer.

    Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>

    ---
    þ Synchronet þ Vertrauen þ Home of Synchronet þ [vert/cvs/bbs].synchro.net
  • From Rob Swindell (on Debian L@VERT to Git commit to main/sbbs/m on Wed Sep 30 19:08:00 2026
    https://gitlab.synchro.net/main/sbbs/-/commit/03fac2eefaccc428d7822bb5
    Modified Files:
    docs/v322_new.md src/sbbs3/smbutil.c src/xpdev/conwrap.c conwrap.h
    Log Message:
    xp_getch(): return 0 at end of input instead of an uninitialized byte

    Fixes #1256

    xp_getch() only checked read() for -1. At end of input read() returns 0,
    so the function fell through and returned whatever happened to be in its uninitialized stack byte. smbutil's message reader took that garbage as a keypress, hit its default case and prompted again, forever, at full CPU: "smbutil r1 base < /dev/null" printed 433,382 prompts in three seconds.

    Initialize the byte and treat read() <= 0 as the failure path, so end of
    input returns 0 like a read error (the header now says so), and make
    smbutil's reader stop on a 0 when its input is not a terminal. On a
    console a 0 can be the first of an extended key's two reads, so there the reader keeps looping as before. The other callers (chksmb, sbbsecho) only
    pause for a keypress and were never looping, so they need no change.

    Verified with the same command on a scratch message base: one prompt,
    "End of input", exit status 0, and the same through a pipe.

    Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>

    ---
    þ Synchronet þ Vertrauen þ Home of Synchronet þ [vert/cvs/bbs].synchro.net
  • From Rob Swindell (on Debian L@VERT to Git commit to main/sbbs/m on Wed Sep 30 19:08:00 2026
    https://gitlab.synchro.net/main/sbbs/-/commit/85e04a4262d8b4afb0ef5bf2
    Modified Files:
    docs/v322_new.md src/sbbs3/js_socket.cpp services.cpp
    Log Message:
    Services: end a script's blocking socket read when it is terminated

    Fixes #1255

    On shutdown the services thread closes the listening sockets and then
    waits, with no limit, for every dynamic service client to disconnect. A
    client blocked in Socket.recvline() only noticed the shutdown when the
    read's own timeout expired, since the read loop never looked at the
    termination flag. nntpservice.js reads with a 300-second timeout once
    its message base has closed, so one idle NNTP session held shutdown for
    up to five minutes, well past systemd's 90-second stop timeout, and the
    process was SIGKILLed. Backtraces from a stuck instance show exactly that:
    the client thread in js_recvline() with timeout=300, the services thread
    in its wait loop.

    The socket object's private data has a pointer to the script's callback
    struct, but it was only filled in when a script installed a socket event.
    Set it when every socket object is created (through a quiet lookup that
    returns NULL for a context without a "js" object), check it in the
    recvline read-check loop, and replace the single blocking wait inside js_socket_recv() with one-second steps that stop when the script is
    terminated, so recv() with a long timeout ends early too. Meanwhile the services thread logs, every ten seconds, which services still have
    clients connected and how many.

    Verified on a scratch services server with one idle NNTP client past its message-base close: SIGTERM to exit took more than 60 seconds before (the
    test gave up) and 2 seconds after.

    Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>

    ---
    þ Synchronet þ Vertrauen þ Home of Synchronet þ [vert/cvs/bbs].synchro.net
  • From Rob Swindell (on Debian L@VERT to Git commit to main/sbbs/m on Wed Sep 30 19:08:00 2026
    https://gitlab.synchro.net/main/sbbs/-/commit/e97a1efe3645baaa65ae9f5c
    Modified Files:
    docs/v322_new.md src/sbbs3/js_bbs.cpp qwk.cpp
    Log Message:
    QWK: restore the resume pointer and clean up after a failed download

    Fixes #1261
    Fixes #1260

    pack_qwk() advances both of a sub-board's pointers for every message it
    packs: the new-scan pointer and the on-line reader's resume position. When
    the download then failed (aborted transfer, quit at the protocol prompt,
    not enough time), qwk_download() put back only the new-scan pointers, so
    the next on-line read of that sub-board started at the last packed
    message rather than where the user had left off. Save and restore both.

    bbs.qwk_download() and bbs.qwk_upload() also depended on setup that only
    the QWK section stub does around them: refreshing the user's directory
    list (and sub-board list for QWK network nodes), and clearing the node's temporary directory afterwards. qwk_download() removed only the .qwk
    file, leaving MESSAGES.DAT, the index files, CONTROL.DAT and the rest in
    temp for the remainder of the session. Both methods now do that setup themselves, and the download clears the temp directory when done, so a
    custom shell or module can call them from anywhere.

    Verified on a scratch terminal server, quitting at the protocol prompt:
    before, the resume pointer had advanced from 0 to 1 and seven packet files remained in temp; after, both pointers are back at 0 and temp is empty.

    Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>

    ---
    þ Synchronet þ Vertrauen þ Home of Synchronet þ [vert/cvs/bbs].synchro.net
  • From Rob Swindell (on Debian L@VERT to Git commit to main/sbbs/m on Wed Sep 30 19:08:00 2026
    https://gitlab.synchro.net/main/sbbs/-/commit/173a186e70b43fde72488d35
    Modified Files:
    docs/v322_new.md src/sbbs3/xtrn_sec.cpp
    Log Message:
    DOOR.SYS: derive the caller-alarm field as PCBOARD.SYS does

    Fixes #1251

    The DOOR.SYS writer hardcoded field 9 ("caller alarm") to Y, while the PCBOARD.SYS writer sets its alarm from whether an answer sound is
    configured and ctrl/sound.mute is absent. The same sysop setting therefore reached a door or not depending only on which drop-file format the door
    was configured for. Use the PCBOARD.SYS expression for DOOR.SYS too.

    Verified with a scratch external that copies its DOOR.SYS out: field 9 is
    Y with an answer sound configured and N once sound.mute exists.

    Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>

    ---
    þ Synchronet þ Vertrauen þ Home of Synchronet þ [vert/cvs/bbs].synchro.net
  • From Rob Swindell (on Debian L@VERT to Git commit to main/sbbs/m on Wed Sep 30 19:08:00 2026
    https://gitlab.synchro.net/main/sbbs/-/commit/7506ebb01d65da5952a58c26
    Modified Files:
    docs/v322_new.md src/sbbs3/scfglib.h scfglib1.c scfglib2.c
    Log Message:
    Config loading: warn when an .ini string value doesn't fit its field

    Every string read from ctrl/*.ini into a fixed-size scfg_t field went
    through SAFECOPY(), which cuts silently. A value longer than its field
    (a 101-character external program command line, say) was simply lost
    past the limit, with nothing in any log, and the resulting behavior had
    to be reverse-engineered: a scratch external configured with a 103-char
    command ran without its trailing %n argument. Values reach those files
    by many paths (SCFG, scripts, hand editing, other tools), so SCFG's own
    input limit is not the only one in play.

    Add scfg_ini_get_str() and its INI_GET_STR() macro, which do the
    iniGetString() lookup and the copy into the field, and log a warning
    naming the key, the limit and the whole value when it had to be
    truncated, and move the 137 copy sites in scfglib1.c and scfglib2.c
    onto it (each now names its key once). Every program that links the
    config readers already supplies the lprintf() that load_cfg.c and
    readtext.c require, so there is no new link requirement.

    Verified by loading a scratch config with that 103-character command:
    "!Config value 'cmd' truncated from 103 to 100 chars: ..." is logged
    once, and the build is warning-free.

    Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>

    ---
    þ Synchronet þ Vertrauen þ Home of Synchronet þ [vert/cvs/bbs].synchro.net
  • From Rob Swindell (on Debian L@VERT to Git commit to main/sbbs/m on Wed Sep 30 19:08:00 2026
    https://gitlab.synchro.net/main/sbbs/-/commit/fd16298c62e1a5ee1e410937
    Modified Files:
    docs/v322_new.md src/sbbs3/sbbs_ini.c
    Log Message:
    sbbs.ini: warn when a server-setting string value doesn't fit its field

    The same silent truncation as the previous commit, for the server
    settings sbbs_read_ini() copies into the *_startup_t structs. The config-library helper could not simply be reused here: sbbs.ini is read
    by the host program (sbbscon, sbbsctrl, ...) before any server exists,
    and libsbbs's lprintf() drops output until a server has installed its
    startup struct, as the first test showed (the xtrn.ini warning logged,
    the sbbs.ini one vanished). Two of the programs that compile this file,
    ntsvcs and sbbsctrl, provide no lprintf() of their own either.

    So the reader reports through the log callback (lputs/cbdata) of the
    startup struct it is filling, which every *_startup_t carries in its
    common leading members and which sbbscon installs before it reads the
    ini; a program that installed none stays silent, as before. A local ini_get_str() does the lookup and the checked copy, and the 51 copy
    sites move onto its INI_GET_STR() macro.

    Verified on a scratch install with a 265-character [BBS]
    LoginRequirements: sbbscon logs "!Config value 'LoginRequirements'
    truncated from 265 to 40 chars: ..." at startup, before the servers'
    own messages. Only the Linux build is verified.

    Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>

    ---
    þ Synchronet þ Vertrauen þ Home of Synchronet þ [vert/cvs/bbs].synchro.net
  • From Rob Swindell (on Debian L@VERT to Git commit to main/sbbs/m on Thu Oct 1 15:27:00 2026
    https://gitlab.synchro.net/main/sbbs/-/commit/8a99194387a88c85b801de3d
    Modified Files:
    docs/v322_new.md src/sbbs3/sbbsecho.c
    Log Message:
    sbbsecho: don't cut UTF-8 header fields mid-character on export

    FTS-0001 packed-message headers hold 35 bytes of To/From and 71 bytes of Subject. SMB header fields are unbounded, so export truncates them, and it
    did so by byte count: a UTF-8 field over the limit (e.g. a Cyrillic name,
    2 bytes per letter) could end in a partial multi-byte sequence, which
    readers show as a replacement character.

    Echomail export and create_netmail() now copy these fields with
    utf8_strlcpy() when the message's header fields are UTF-8 and valid, so the
    cut falls on a character boundary. On ASCII-only (SUB_ASCII) sub-boards
    the field is converted from UTF-8 to CP437/ASCII before truncation rather
    than after, so it no longer converts an already-split sequence and keeps
    up to twice as many characters.

    Fixes #1276

    Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

    ---
    þ Synchronet þ Vertrauen þ Home of Synchronet þ [vert/cvs/bbs].synchro.net
  • From Rob Swindell (on Debian L@VERT to Git commit to main/sbbs/m on Thu Oct 1 15:27:00 2026
    https://gitlab.synchro.net/main/sbbs/-/commit/9dec26e6d38d842bfa9f0228
    Modified Files:
    docs/v322_new.md src/sbbs3/sbbsecho.c
    Log Message:
    sbbsecho: support FSP-1030 UCSFROM/UCSTO/UCSSUBJ control lines

    FTS-0001 packet headers hold 35 bytes of To/From and 71 of Subject, which
    in UTF-8 can be as few as 8 and 17 characters. FSP-1030 (preserved as
    FRL-1021) carries the full field in a ^AUCSFROM:, ^AUCSTO: or ^AUCSSUBJ: control line. GoldED+ and AmberEdit now write them.

    Export: when a UTF-8 message's To, From or Subject does not fit its packet header field, echomail export and create_netmail() add the matching UCS*
    line with the full value (not on ASCII-only sub-boards, and not when the message already carries one).

    Import: for a UTF-8 message (per CHRS/CHARSET, or auto-detection), a UCS*
    line replaces the header field only if it extends what the packet header already says (allowing for a trailing partial UTF-8 sequence left by the sender's tosser), so it cannot substitute a different name or subject. An applied line is consumed; any other is kept as an unrecognized control
    line, as before.

    The full fields are worked out before anything matches on them:

    - twit list and subject filtering check the full fields as well as the
    packet's, so a filtered name or word past the cut is still caught, and
    entries written against cut names keep matching
    - netmail recipient matching (robots, AreaFix/ping, sysop aliases, user
    lookup) and the echomail recipient notification use the full To; user
    name matching ignores non-ASCII bytes, so a cut To could otherwise
    match a different local user than the full one

    Refs #1276

    Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

    ---
    þ Synchronet þ Vertrauen þ Home of Synchronet þ [vert/cvs/bbs].synchro.net
  • From Rob Swindell (on Debian L@VERT to Git commit to main/sbbs/m on Fri Oct 23 20:52:00 2026
    https://gitlab.synchro.net/main/sbbs/-/commit/b4ff53364840117fe90ee585
    Modified Files:
    docs/v322_new.md src/sbbs3/ars.c ars_defs.h chk_ar.cpp userdat.c
    Log Message:
    ARS: add SOUND and MOUSE terminal capability keywords

    SOUND is true when the terminal can play audio files (the same gate the
    @SOUND: family of @-codes uses) and the user has not turned off
    BBS-initiated sound. MOUSE is true when the terminal supports mouse
    hot-spots. This lets a display file gate content around a sound clip,
    e.g. @SHOW:SOUND@@WAIT:100@@SHOW@ to hold the connection open only for
    callers who will hear it.

    Without a live terminal (sessionless evaluation, e.g. in the web and FTP servers), SOUND follows only the user's sound setting and MOUSE the
    user's stored mouse setting. Failing either keyword reports the
    existing NoAccessTerminal text.

    Also adds the v3.22 release-note entry for the SCFG template sub-board
    fix (#1280).

    Fixes #1292

    Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

    ---
    þ Synchronet þ Vertrauen þ Home of Synchronet þ [vert/cvs/bbs].synchro.net
  • From Rob Swindell (on Debian L@VERT to Git commit to main/sbbs/m on Fri Oct 23 20:52:00 2026
    https://gitlab.synchro.net/main/sbbs/-/commit/50a5a194e66000493f55a783
    Modified Files:
    docs/v322_new.md src/sbbs3/answer.cpp
    Log Message:
    answer: detect UTF-8 with U+2060 Word Joiner, not obsolete ZWNBSP

    Terminal auto-detection has sent U+FEFF (ZWNBSP) after a CR and taken
    the second cursor-position report as the verdict since 11d3f59f5a (error-3-cancel, 2019-06-28): column 1 or 2 means UTF-8. Since Unicode
    3.2, U+2060 is the in-text zero-width no-break character and U+FEFF is
    just the byte-order mark, which a terminal may render as an unsupported character. A UTF-8 terminal drawing a full-width replacement glyph
    lands on column 3 and goes undetected.

    Probe with U+2060 (E2 81 A0) instead. glibc wcwidth() gives both 0, and
    in tmux both leave the cursor where it was, so conforming terminals see
    no change. A terminal too old to know U+2060 and drawing it one column
    wide still lands on column 2 and is detected, with unicode_zerowidth=1.

    The "x < 3" threshold stays: 0x81 is a C1 control in ISO-8859-x, so a
    Latin-1 terminal that swallows it prints two glyphs and lands on
    column 3, which must not read as UTF-8.

    Fixes #1284

    Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

    ---
    þ Synchronet þ Vertrauen þ Home of Synchronet þ [vert/cvs/bbs].synchro.net
  • From Rob Swindell (on Windows@VERT to Git commit to main/sbbs/m on Sun Oct 25 14:40:00 2026
    https://gitlab.synchro.net/main/sbbs/-/commit/eb6e30af8af755f35dc494bc
    Modified Files:
    docs/v322_new.md src/sbbs3/sbbsexec.c vdd_func.h xtrn.cpp
    Log Message:
    Allow concurrent DOS doors on the same node number from two BBS hosts

    The mailslots and events the Windows VDD (sbbsexec.dll) and its host
    share were named by node number alone since 7e3e47141a (2000-10-10),
    so a second BBS instance on the same system (another Synchronet install
    or a different BBS package using our FOSSIL/UART driver) running a DOS
    program for the same node number failed its launch with an ERR_CREATE
    on the rd mailslot.

    The host now creates the legacy names first, the mailslot before the
    events since its name is machine-global and CreateMailslot() fails
    atomically when it exists. On ERROR_ALREADY_EXISTS it retries once
    with a ".<pid>" suffix and launches the VDM with SBBSEXEC_INSTANCE=<pid>
    in its environment; the VDD appends the suffix to every name it opens.
    With no variable set the VDD uses the exact legacy names, so an older sbbsexec.dll keeps working wherever a launch worked before.

    Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>

    ---
    þ Synchronet þ Vertrauen þ Home of Synchronet þ [vert/cvs/bbs].synchro.net
  • From Rob Swindell (on Windows@VERT to Git commit to main/sbbs/m on Sun Oct 25 18:47:00 2026
    https://gitlab.synchro.net/main/sbbs/-/commit/45782f780df71c073a2beb92
    Modified Files:
    docs/v322_new.md src/sbbs3/js_global.cpp js_socket.cpp jsexec.cpp mailsrvr.cpp main.cpp sbbs.h services.cpp websrvr.cpp
    Log Message:
    Route JS Socket log output to the script host's logger, not the terminal server's

    js_socket.cpp logged everything through the file-global lprintf(), which
    inside sbbs.dll/libsbbs.so resolves to main.cpp's terminal-server logger.
    The mail, web and services servers each keep their own static lprintf(),
    so a Socket's debug trace, TLS errors and listening-socket failures from
    their scripts appeared in the Terminal Server's log, were tagged "term"
    in error.log and published under the terminal server's MQTT topic.
    Under jsexec on Windows the terminal startup is NULL, so the output was
    dropped entirely and errors were written to an error.log in the current directory through the DLL's zeroed scfg. This dates from the Socket
    class's introduction in a8352397c7 (2001-06-12); the TODO noting it was
    added in 34587421f7 (2023-11-21).

    Add an lprintf function pointer to js_callback_t, set by every host when
    it initializes its callback struct and inherited by background load()
    threads. js_socket.cpp resolves its logger through the socket's js_cb,
    falling back to the terminal server's when a host set none, and passes
    the same pointer to ssl_sync(), add_private_key(), destroy_session() and xpms_create(). The cryptlib attribute helpers take the socket private
    struct instead of a bare session so they can reach it, and their TLS
    messages now carry the socket number like the rest.

    Resolves #700

    Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>

    ---
    þ Synchronet þ Vertrauen þ Home of Synchronet þ [vert/cvs/bbs].synchro.net
  • From Rob Swindell (on Debian L@VERT to Git commit to main/sbbs/m on Mon Oct 26 00:09:00 2026
    https://gitlab.synchro.net/main/sbbs/-/commit/5cfbf97f3d802ff7918ecfe8
    Modified Files:
    docs/v322_new.md src/sbbs3/qwk.cpp
    Log Message:
    qwk: take the QWK job lock before a REP upload, not after it

    qwk_upload() took the user's QWK job lock only once the protocol had
    received the REP, so when another session (FTP, Web, a script or another
    host) held it, the user's uploaded packet was refused and deleted, and
    had to be sent again.

    The lock is now taken before the upload prompt, as qwk_download() does,
    and held through unpack_rep(); a busy lock shows QWKTransferInProgress
    before any transfer starts.

    Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

    ---
    þ Synchronet þ Vertrauen þ Home of Synchronet þ [vert/cvs/bbs].synchro.net
  • From Rob Swindell (on Windows@VERT to Git commit to main/sbbs/m on Mon Oct 26 20:38:00 2026
    https://gitlab.synchro.net/main/sbbs/-/commit/45147ca28ee23dd4ef43c81f
    Modified Files:
    docs/v322_new.md src/sbbs3/data_ovl.cpp
    Log Message:
    Defer the "Loading message pointers" meter the same way

    Share the deferred-progress callback 5ea61816cf (2026-10-26) added
    for the duplicate-user search with getmsgptrs(), so the logon-time
    meter also draws only once loading has outlasted the progress
    interval and erases only if it was drawn.

    Add the release note for both.

    Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>

    ---
    þ Synchronet þ Vertrauen þ Home of Synchronet þ [vert/cvs/bbs].synchro.net
  • From Rob Swindell (on Debian L@VERT to Git commit to main/sbbs/m on Sun Nov 1 00:20:00 2026
    https://gitlab.synchro.net/main/sbbs/-/commit/442843d7aae805670e59d103
    Modified Files:
    docs/v322_new.md src/sbbs3/sbbs.h sbbsdefs.h src/sbbs3/scfg/scfgxtrn.c src/sbbs3/writemsg.cpp xtrn.cpp xtrn_sec.cpp
    Log Message:
    Offer BBSDEV.DRP and DOORDROP.INI drop files to external editors

    Both types are now in SCFG's drop file list for external message editors,
    not only online programs. When an editor uses one, the message editor
    launch writes the file to the node directory, hands the editor its path
    in the BBSDEV_DRP or DOORDROP_INI environment variable (the %f specifier
    keeps naming the message text file) and, for DOORDROP.INI, removes the
    file after the editor exits. A failure to write the file aborts the
    editor instead of launching it with a dangling path.

    The DOS-side path of a drop file (the DOSEMU drive on Linux, the doscmd
    drive on FreeBSD, the short path on Windows) moves from exec_xtrn() into dos_drop_path(), so the editor launch shares it. The doscmd drive
    letters, until now string literals in the .doscmdrc writer, get DOSCMD_*
    macros alongside the DOSEMU_* ones.

    Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>

    ---
    þ Synchronet þ Vertrauen þ Home of Synchronet þ [vert/cvs/bbs].synchro.net
  • From Rob Swindell (on Debian L@VERT to Git commit to main/sbbs/m on Fri Nov 6 20:31:00 2026
    https://gitlab.synchro.net/main/sbbs/-/commit/fc12511c119b01a78909deb7
    Modified Files:
    docs/v322_new.md src/sbbs3/telgate.cpp
    Log Message:
    Don't drop remote output in telnet_gate() when the output buffer is full

    RingBufWrite() stores only what fits in the ring buffer and returns that
    count. telnet_gate() ignored the return value, so once the 20000-byte
    outbuf filled (any time the remote host produced output faster than the
    user's connection drained it), whatever didn't fit was silently discarded.
    A door sending Sixel graphics over an rlogin gate to a SyncTERM user lost chunks mid-image.

    Now the gateway receives from the remote only as much as the output buffer
    can hold (half that when expanding LFs, since the expansion can double the data) and idles, still forwarding the user's keystrokes, when there is no
    room; TCP back-pressure then throttles the remote host. The local echo
    path waits for the output thread to make room rather than dropping too.

    Verified with a fake rlogin host blasting 240 KB at a deliberately slow
    telnet client: before, the stream diverged at byte 19999 and 147 KB never arrived; after, all 240 KB arrive intact, including with TG_ECHO (keys sent mid-stream) and TG_EXPANDLF.

    Fixes #1325

    Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>

    ---
    þ Synchronet þ Vertrauen þ Home of Synchronet þ [vert/cvs/bbs].synchro.net
  • From Rob Swindell (on Debian L@VERT to Git commit to main/sbbs/m on Sat Nov 7 20:38:00 2026
    https://gitlab.synchro.net/main/sbbs/-/commit/61a343a6418a6cb9aba1a82c
    Modified Files:
    docs/v322_new.md src/sbbs3/scfg/scfg.c scfg.h scfgsub.c scfgxfr2.c src/uifc/uifc.h
    Log Message:
    scfg: make long Actual File Paths viewable

    A directory with a blank or relative Actual File Path shows the resolved
    full path in [brackets] on the directory menu, where it is clipped at the window edge, and pressing Enter edited the (blank) stored value, so there
    was no way to read the whole path. Fixes #1327

    - Pressing Enter on a blank Actual File Path now seeds the input box with
    the derived path, so it can be scrolled. Accepting it unchanged keeps the
    setting blank (derived); any edit is stored as typed.
    - The directory and sub-board configuration windows now widen to fit their
    longest line (minimum 72 columns, so an 80-column screen is unchanged)
    via a new opt_width() helper, so wider terminals show the full path.
    - MAX_OPLN raised from 75 to 256 (the screen-width limit); menu lines
    built with it were truncated to 47 characters of value after the 27
    character label, which clipped the path before the window did.
    - Help text for Actual File Path, FidoNet Area Tag (dirs and subs), and
    Newsgroup Name now explains that a [bracketed] value is derived, not
    stored.

    Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>

    ---
    þ Synchronet þ Vertrauen þ Home of Synchronet þ [vert/cvs/bbs].synchro.net
  • From Rob Swindell (on Debian L@VERT to Git commit to main/sbbs/m on Sun Nov 8 15:46:00 2026
    https://gitlab.synchro.net/main/sbbs/-/commit/16d0248e3cc9fa1e833acdf0
    Modified Files:
    docs/v322_new.md src/sbbs3/con_hi.cpp getstr.cpp writemsg.cpp
    Log Message:
    Accept Unicode (UTF-8) text in the internal message editor

    On a UTF-8 terminal, the internal editor translated every typed
    character to CP437 and dropped any with no CP437 equivalent (e.g.
    "¨", "¨", "¨"), because its getstr() calls lacked K_UTF8. The subject
    prompt has passed K_UTF8 since f68a24d61b (registry-20-rail,
    2024-02-20), so a subject could hold emoji while the body could not.

    getstr() now handles multi-byte UTF-8 input as characters: the bytes
    of a sequence are collected per keystroke, byte-offsets and display
    columns are tracked separately (via str_width/prev_char/next_char),
    and all editing operations (word-wrap, insert/overwrite, backspace,
    delete, Ctrl-W/Ctrl-D, Home/End, cursor keys, tab stops, centering)
    work on characters and columns. The line-length limit remains a byte
    limit (the caller's buffer size) while the wrap point is the terminal
    width in columns. Edits go through two shared routines (insert_bytes/ delete_range) that print the same screen state as before for CP437
    input.

    redrwstr() moved to getstr.cpp (its only caller) and takes byte
    offsets for both the cursor position and the length, converting to
    columns itself. It previously compared a column count against a byte
    count and so, after a multi-byte character was typed into a K_UTF8
    field (e.g. an emoji in a subject), moved the cursor back to the start
    of the field; text typed next overwrote the character on screen.

    A CP437-translated high-bit character (e.g. "­") typed on a UTF-8
    terminal was stored but never echoed, since the echo branch required
    the buffer to be valid UTF-8. It is now echoed via outchar(), which
    performs the CP437 to UTF-8 translation.

    The editor's per-line byte budget (MAX_LINE_LEN) grows from cols+1 to (cols-1)*UTF8_MAX_LEN+3 to hold multi-byte lines. The old value also
    left no room for the CR/LF of a full-width line: a 79-column line on
    an 80-column terminal lost its LF when the lines were rejoined.

    The "Export ASCII Characters Only" sub-board setting is an export-time conversion and does not restrict local input.

    Fixes #1317

    Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>

    ---
    þ Synchronet þ Vertrauen þ Home of Synchronet þ [vert/cvs/bbs].synchro.net
  • From Rob Swindell (on Windows@VERT to Git commit to main/sbbs/m on Tue Nov 10 22:21:00 2026
    https://gitlab.synchro.net/main/sbbs/-/commit/66b4a82b68da0e86c2c07abb
    Modified Files:
    docs/v322_new.md src/sbbs3/jsexec.cpp
    Log Message:
    jsexec: add -T to prefix log and status messages with the date and time

    A long-running script that logs to a file (-A) had no way to say when
    anything happened: working out the order and duration of events in an ircd
    log meant inferring the time from protocol messages that happen to carry
    one.

    With -T each message written by lprintf() or mfprintf() (script log()
    calls, JavaScript errors and warnings, and the per-run status lines) starts with the local date and time to the millisecond:

    2026-11-10 22:03:30.494 info line

    Output written by print(), write() and the like is the script's own and is
    not prefixed, and syslog output (-d) already carries a time.

    Co-Authored-By: Claude Fable 5.2 <noreply@anthropic.com>

    ---
    þ Synchronet þ Vertrauen þ Home of Synchronet þ [vert/cvs/bbs].synchro.net
  • From Rob Swindell (on Debian L@VERT to Git commit to main/sbbs/m on Tue Nov 10 22:33:00 2026
    https://gitlab.synchro.net/main/sbbs/-/commit/3592004a31ca096b15faeafd
    Modified Files:
    docs/v322_new.md src/sbbs3/ftpsrvr.cpp mailsrvr.cpp main.cpp websrvr.cpp src/xpdev/multisock.c multisock.h
    Log Message:
    Close server listening sockets as soon as shutdown begins

    The terminal, mail, FTP, and web servers kept their listening sockets open until cleanup(), which runs only after the wait for active clients and
    child threads. Nothing calls accept() during that wait, but the kernel
    still completes TCP handshakes into the listen backlog, so every client
    that connected during a shutdown or recycle sat on an unanswered
    connection until the sockets finally closed, then got a reset. On a busy
    web server that drain can last two minutes.

    Add xpms_close() to xpdev, which closes a set's sockets but leaves the set allocated, and call it in each server right after its accept loop exits. xpms_destroy() now uses it. The set must stay allocated rather than being destroyed early because the FTP transfer threads treat ftp_set == NULL as "server torn down": clearing it at the start of a recycle would abort in-progress transfers. The services server already closed its sockets at
    this point.

    Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

    ---
    þ Synchronet þ Vertrauen þ Home of Synchronet þ [vert/cvs/bbs].synchro.net
  • From Rob Swindell (on Debian L@VERT to Git commit to main/sbbs/m on Wed Nov 11 13:30:00 2026
    https://gitlab.synchro.net/main/sbbs/-/commit/525b70ed664b4fdb3da26c25
    Modified Files:
    docs/v322_new.md src/sbbs3/atcodes.cpp
    Log Message:
    @TUSED@/@TLEFT@: don't drop the first hours digit at 10+ hours

    The truncated HH:MM:SS duration format (the default for @TUSED@ and
    @TLEFT@) skipped the first character of sectostr()'s output to drop a
    leading zero, but did so unconditionally: once a duration reached 10 hours
    it removed the first digit of the hours instead, so a 14h44m session was
    shown as "4:44:30" (e.g. "Time used:" in the stock logoff.msg) and 100
    hours as "00:00:00". Skip the leading character only when it is a '0'.

    Fixes #1336

    Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

    ---
    þ Synchronet þ Vertrauen þ Home of Synchronet þ [vert/cvs/bbs].synchro.net
  • From Rob Swindell (on Windows@VERT to Git commit to main/sbbs/m on Wed Nov 11 13:56:00 2026
    https://gitlab.synchro.net/main/sbbs/-/commit/38c93be5ba8d0cd42049e8d6
    Modified Files:
    docs/v322_new.md src/sbbs3/js_internal.cpp js_socket.cpp js_socket.h Log Message:
    js: don't block a non-blocking TLS socket's send on an idle peer

    cryptlib's write timeout defaults to 5 seconds and Socket never changed
    it, so send() on a TLS socket whose peer had stopped reading held the
    script for 5 seconds in the push and again in the flush, even with
    nonblocking set. In an event-driven script (the IRC daemon) that froze
    every other connection and timer for as long as one peer was slow.

    A non-blocking TLS socket now gets a zero write timeout, and the
    session's own timeout back if the script makes it blocking again. A
    flush that cannot finish is no longer an error: js_handle_events()
    retries it, for sockets with a read event, and wakes every 20 ms while
    any output is waiting. cryptlib waits 5 ms or more on the descriptor
    even with a zero timeout, so send() and the retry first check that the
    socket is writable.

    Socket.sendfile() on a non-blocking TLS socket would now give up as
    soon as the peer's window filled, so it waits for room and retries, as
    it does for a plain socket.

    Not changed: a script that sends on a non-blocking TLS socket without
    a read event keeps unsent output until its next send, and close()
    discards what the peer has not taken rather than waiting for it.

    Built with MSVC and GCC. On Windows, 600 KB sent to a peer idle for
    20 seconds: longest send() 10.02 s before, 0.019 s after, with every
    byte delivered.

    Co-Authored-By: Claude Fable 5.2 <noreply@anthropic.com>

    ---
    þ Synchronet þ Vertrauen þ Home of Synchronet þ [vert/cvs/bbs].synchro.net
  • From Rob Swindell (on Windows@VERT to Git commit to main/sbbs/m on Wed Nov 11 23:15:00 2026
    https://gitlab.synchro.net/main/sbbs/-/commit/87e5d349ce31d2419bea793e
    Modified Files:
    docs/v322_new.md src/sbbs3/js_user.cpp userdat.c
    Log Message:
    GUEST ARS keyword and user.is_guest now match a not-logged-in user

    user_is_guest() returns true for user number 0 as well as for an account
    with the 'G' restriction. A NULL user (no user context, e.g. jsexec)
    is still not a guest.

    An unauthenticated web server request is evaluated as user number 0,
    which has no restrictions, so an access requirement of "NOT GUEST" on a
    file library or directory was satisfied without a login: the directory
    was listed and served under the web file vpath (e.g. /files/) and no
    HTTP authentication was requested. Such a request now fails "NOT GUEST"
    and gets a 401.

    Co-Authored-By: Claude Fable 5.2 <noreply@anthropic.com>

    ---
    þ Synchronet þ Vertrauen þ Home of Synchronet þ [vert/cvs/bbs].synchro.net
  • From Rob Swindell (on Debian L@VERT to Git commit to main/sbbs/m on Fri Nov 13 20:36:00 2026
    https://gitlab.synchro.net/main/sbbs/-/commit/fdf8426de7555abd40f9eec5
    Modified Files:
    docs/v322_new.md src/sbbs3/js_console.cpp
    Log Message:
    js: rename console.ansi() to console.attrstr(), keep ansi() as an alias

    The method returns the attribute sequence for the user's terminal type
    (ANSI, PETSCII, or an empty string for terminals without attribute
    support), not only ANSI. The new name matches the underlying Terminal::attrstr() and the documentation now describes the actual
    behavior.

    Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

    ---
    þ Synchronet þ Vertrauen þ Home of Synchronet þ [vert/cvs/bbs].synchro.net
  • From Rob Swindell (on Debian L@VERT to Git commit to main/sbbs/m on Fri Nov 13 21:49:00 2026
    https://gitlab.synchro.net/main/sbbs/-/commit/e11b783b6a7846f48cd175c8
    Modified Files:
    docs/v322_new.md src/sbbs3/atcodes.cpp
    Log Message:
    Let @CPS@ take a byte-count format parameter

    @CPS@ now accepts the same optional byte-count format letter as
    @FILE_BYTES@ and @XFER_CPS@ (B, K, M, G or V), e.g. @CPS:V@ for "15.2K". Without a parameter it is unchanged: the rate in bytes per second.

    Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

    ---
    þ Synchronet þ Vertrauen þ Home of Synchronet þ [vert/cvs/bbs].synchro.net
  • From Rob Swindell (on Debian L@VERT to Git commit to main/sbbs/m on Sat Nov 14 01:20:00 2026
    https://gitlab.synchro.net/main/sbbs/-/commit/55845c1ea968da4b0aed10a7
    Modified Files:
    docs/v322_new.md src/sbbs3/atcodes.cpp
    Log Message:
    Let @FREESPACE@ take a byte-count format parameter

    @FREESPACE@ (free space in the temp directory) now accepts the same
    optional byte-count format letter as @CPS@ and @FILE_BYTES@ (B, K, M, G or
    V), e.g. @FREESPACE:V@ for "27.8G". Without a parameter it is unchanged:
    the free space in bytes. The FREESPACEK, FREESPACEM, FREESPACEG and
    FREESPACET codes are unaffected.

    Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

    ---
    þ Synchronet þ Vertrauen þ Home of Synchronet þ [vert/cvs/bbs].synchro.net
  • From Rob Swindell (on Windows@VERT to Git commit to main/sbbs/m on Mon Nov 16 21:58:00 2026
    https://gitlab.synchro.net/main/sbbs/-/commit/6e8fd1d1ed5a73b9a1542431
    Modified Files:
    docs/v322_new.md src/sbbs3/js_file.cpp
    Log Message:
    JS File.position: support offsets beyond 4GB

    The position setter converted the value to a 32-bit unsigned integer (370b79dd7a (2021-04-22)) and both accessors used fseek()/ftell(), so
    a script could neither seek past 4GB in a larger file nor read back
    its position there: an assigned offset silently wrapped modulo 2^32.
    Example: a script walking the atoms of a 37GB MP4 file could not reach
    the "moov" atom stored at the end of the file.

    Take the value as a number and use fseeko()/ftello(). A negative or
    non-numeric value now seeks to the start of the file.

    Co-Authored-By: Claude Fable 5.2 <noreply@anthropic.com>

    ---
    þ Synchronet þ Vertrauen þ Home of Synchronet þ [vert/cvs/bbs].synchro.net
  • From Rob Swindell (on Debian L@VERT to Git commit to main/sbbs/m on Thu Nov 19 09:38:00 2026
    https://gitlab.synchro.net/main/sbbs/-/commit/b17526d9ed681dec30a638aa
    Modified Files:
    docs/v322_new.md src/sbbs3/smbutil.c
    Log Message:
    smbutil: test pack's header allocation result before offsetting it

    Pack added the base's header offset to smb_fallochdr()'s return value
    before testing it for an error. With the usual header offset (32) the
    sum stayed negative, so the message was still skipped but the error
    was logged with the wrong code (e.g. -70 for SMB_ERR_HDR_OFFSET,
    -102). A base with a header offset larger than the error code's
    magnitude would have had the failure go unnoticed and the header
    written at a bogus offset. The header offset is now added only to a
    successful allocation.

    Also note the allocation and chksmb fixes in the v3.22 release notes.

    Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

    ---
    þ Synchronet þ Vertrauen þ Home of Synchronet þ [vert/cvs/bbs].synchro.net
  • From Rob Swindell (on Debian L@VERT to Git commit to main/sbbs/m on Thu Nov 19 11:51:00 2026
    https://gitlab.synchro.net/main/sbbs/-/commit/a324eb90f5f5fd77fadded77
    Modified Files:
    docs/v322_new.md src/sbbs3/listfile.cpp
    Log Message:
    listfiles(): skip directories with no new files before a List Files module

    In a new-file scan, listfiles() invoked a configured List Files loadable
    module for every directory, before its own (fast) check for new files.
    Starting a JavaScript module costs about 25 ms per directory, so a scan
    of hundreds of directories with few new files took many seconds longer
    with a module than with the built-in lister, for no output.

    The new-file check (and the update of the user's new-file scan time) now
    comes first, and a directory with no new files returns 0 without
    invoking the module. A module is still invoked, with the same arguments,
    for every directory that has new files and for every other kind of
    listing.

    Note for DDFileLister (Nightfox) users: installed as the List Files
    module, it printed "No new files found in <library>: <directory>" for
    each directory without new files during a new-file scan; it is no
    longer invoked for those directories, so those lines no longer appear,
    matching the built-in lister. Its use as the Scan Dirs module is
    unaffected.

    Fixes #1359

    Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

    ---
    þ Synchronet þ Vertrauen þ Home of Synchronet þ [vert/cvs/bbs].synchro.net
  • From Rob Swindell (on Debian L@VERT to Git commit to main/sbbs/m on Thu Nov 19 11:51:00 2026
    https://gitlab.synchro.net/main/sbbs/-/commit/e78f01685b5886283df63c12
    Modified Files:
    docs/v322_new.md src/sbbs3/answer.cpp
    Log Message:
    SSH: range-check the client's terminal width and height

    The width and height from an SSH client's pty-req were accepted as long
    as they were greater than zero, so an SSH client could set dimensions
    (e.g. 1 column, or billions) that the Telnet window-size, ANSI
    auto-detection, JavaScript console and user-setting paths all reject.
    Apply the same TERM_COLS_MIN/MAX and TERM_ROWS_MIN/MAX limits and log
    and ignore an out-of-range value, leaving the default or auto-detected dimension in effect, as for Telnet.

    Also log the width as "width" rather than "height".

    Fixes #1368

    Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

    ---
    þ Synchronet þ Vertrauen þ Home of Synchronet þ [vert/cvs/bbs].synchro.net
  • From Rob Swindell (on Debian L@VERT to Git commit to main/sbbs/m on Thu Nov 19 11:51:00 2026
    https://gitlab.synchro.net/main/sbbs/-/commit/17f2d9a42922f0c2604629e9
    Modified Files:
    docs/v322_new.md src/sbbs3/sbbsdefs.h
    Log Message:
    Lower the minimum terminal width from 40 to 32 columns

    The Tandy Color Computer 1 and 2 have a 32x16 text screen, and some
    callers use terminal programs at that size. TERM_COLS_MIN gates every
    source of terminal width (Telnet window size, SSH, ANSI auto-detection,
    the user's setting, console.screen_columns), the message editor's
    too-narrow check, and the recorded width putmsg() honors when
    re-wrapping a message, so lowering it lets all of those work at 32-39
    columns.

    Michael Pittsley (JustMike, CoCoNet BBS) ran v3.21e with this change at
    32x16 on a CoCo 2 and found nothing broken: reading and posting
    messages work, and messages posted at 32 columns re-wrap to full width
    when read on wider terminals. Remaining issues are cosmetic (file
    listings, fixed-width lists, and menus without a .c32 variant).

    Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

    ---
    þ Synchronet þ Vertrauen þ Home of Synchronet þ [vert/cvs/bbs].synchro.net
  • From Rob Swindell (on Debian L@VERT to Git commit to main/sbbs/m on Fri Nov 20 15:25:00 2026
    https://gitlab.synchro.net/main/sbbs/-/commit/30e82b9600a154489f0fef85
    Modified Files:
    docs/v322_new.md src/sbbs3/websrvr.cpp
    Log Message:
    websrvr: file-area downloads are sent sandboxed and not sniffed

    A file requested from a file area (FileVPathPrefix) was sent with a Content-Type taken from its extension and nothing else. An uploaded
    .html or .svg file, opened through its link, was a page of whatever
    host name the link used: script in it ran with that site's origin and
    could act as the signed-in visitor there.

    Every file-area download now carries

    X-Content-Type-Options: nosniff
    Content-Security-Policy: sandbox

    The sandbox makes the browser treat the file as a document of its own:
    no script, and no access to the site's cookies, storage or pages. The
    file still displays. PDFs get nosniff only, because a browser's built-in
    PDF viewer does not start inside a sandbox. Listings (the index script),
    the rest of the web root and scripts are unchanged.

    Checked on a scratch web server on loopback with an .html, .svg, .pdf
    and .txt file in a file area: without the change none had either
    header; with it the first, second and fourth have both and the PDF has
    nosniff. In headless Firefox the test page's script did not run and the document's origin was "null". Built with GCC only.

    Fixes #1363

    Co-Authored-By: Claude Fable 5.2 <noreply@anthropic.com>

    ---
    þ Synchronet þ Vertrauen þ Home of Synchronet þ [vert/cvs/bbs].synchro.net
  • From Rob Swindell (on Debian L@VERT to Git commit to main/sbbs/m on Sat Nov 21 12:52:00 2026
    https://gitlab.synchro.net/main/sbbs/-/commit/faada4f5bad39e5eaf7e4b74
    Modified Files:
    docs/v322_new.md src/sbbs3/scfg/scfgindex.h scfgmsg.c
    Log Message:
    SCFG: "Maximum Messages Waiting" is "Maximum E-mail Messages Waiting"

    The option under Message Options did not say that it is about mail.
    The menu line, its help texts, the input prompt and the search index
    use the new name; the msgs.ini key (max_msgs_waiting) and system.max_msgs_waiting are unchanged. The help paragraphs that name it
    are wrapped again to stay within 72 columns.

    scfgmsg.c passes a syntax-only GCC compile; SCFG was not built or run.

    Co-Authored-By: Claude Fable 5.2 <noreply@anthropic.com>

    ---
    þ Synchronet þ Vertrauen þ Home of Synchronet þ [vert/cvs/bbs].synchro.net
  • From Rob Swindell (on Debian L@VERT to Git commit to main/sbbs/m on Sat Nov 21 18:12:00 2026
    https://gitlab.synchro.net/main/sbbs/-/commit/c7ca0b49db60aa086c15eb8b
    Modified Files:
    docs/v322_new.md src/sbbs3/js_user.cpp
    Log Message:
    JS: new User.mailbox_bytes() method

    Returns the size, in bytes, of the user's mail as counted against the
    mailbox quota (system.mailbox_quota): the Inbox, spam included, plus the folders, Sent copies and Trash kept by the mailbox library. It is the
    total User.mailbox_full() compares with the quota.

    This is the remaining half of #22, which asked for user.stats
    mail_waiting (added in v3.12) and mail_bytes_waiting. It is a method,
    not a user.stats property, because it is not cheap: each call scans the
    mail base index and reads a header per Inbox message (about 2 ms for a
    user with no mail, 10 ms for 450 messages, measured over SMB), so a loop
    over every user takes seconds.

    Fixes #22

    Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

    ---
    þ Synchronet þ Vertrauen þ Home of Synchronet þ [vert/cvs/bbs].synchro.net
  • From Rob Swindell (on Debian L@VERT to Git commit to main/sbbs/m on Sat Nov 21 18:12:00 2026
    https://gitlab.synchro.net/main/sbbs/-/commit/6de0bf87c59a00c09f226ad2
    Modified Files:
    docs/v322_new.md src/sbbs3/filterfile.hpp
    Log Message:
    filterFile: keep the loaded list when a changed filter file can't be read

    filterFile (ip.can, ip-silent.can, host.can, ipfilter_exempt.cfg, spamblock.cfg, dnsbl_exempt.cfg, ...) freed its cached list before
    reading the changed file, and recorded the new date whether or not the
    read succeeded. A file that could not be opened, even after fnopen()'s
    retries, was therefore treated as empty until its date changed again:
    for ipfilter_exempt.cfg, a file rarely edited, every exemption could
    vanish until the next edit.

    Now:
    - An unreadable file keeps the list already loaded until the file
    changes again. It is not retried at each check: each attempt repeats
    fnopen()'s lock retries (about 28 seconds for a permission error) with
    the mutex held, stalling every thread that consults the filter.
    - A deleted file lists nothing. It used to keep its last contents,
    because fdate()'s -1 was not newer than the recorded date.
    - Any change of date reloads the file, so one restored from a backup
    with an older date is read.
    - A file changed within the last second is read again at the next
    check, in case it was read while still being written.

    Tested on Linux with the web server and ip.can: an entry blocks; the
    file deleted, the entry no longer blocks; restored with a date two days
    older, it blocks again; rewritten without the entry but unreadable, the
    entry still blocks (the first check waited out fnopen()'s retries, 28
    seconds, the next did not); readable and changed, it no longer blocks.

    Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

    ---
    þ Synchronet þ Vertrauen þ Home of Synchronet þ [vert/cvs/bbs].synchro.net
  • From Rob Swindell (on Debian L@VERT to Git commit to main/sbbs/m on Sat Nov 21 19:18:00 2026
    https://gitlab.synchro.net/main/sbbs/-/commit/8093a5954e39adc1e4fbe8ca
    Modified Files:
    docs/v322_new.md src/sbbs3/services.cpp websrvr.cpp
    Log Message:
    Web Server: count logins with an unknown or deleted user as failed

    Every other server that authenticates (terminal, RLogin, SSH, FTP, POP3,
    SMTP AUTH, Services) counts a login with an unknown user name as a
    failed attempt. The web server did not, on either login path, so
    through it a client could guess user names without reaching the
    temporary-ban or auto-filter thresholds or the hack log.

    Now, against the request's client (the forwarded address behind a
    trusted reverse proxy):
    - HTTP authentication counts an unknown user name, with the Basic
    password ("<digest>" for Digest), as a wrong password already was.
    Its !UNKNOWN USER line now names the protocol and client address.
    - The JavaScript login() method counts an unknown user name, now logged
    as !UNKNOWN USER, and a deleted or inactive account. A user record that
    cannot be read is logged as an error and not counted.
    - The Services login() method also counts a deleted or inactive
    account, as the terminal server does on Telnet auto-login. An unknown
    name is logged as !UNKNOWN USER; a user record that cannot be read is
    logged as an error and no longer counted.

    A lookup that could not run (find_login_id() < 0) is still a system
    error and not counted.

    Tested on Linux against an isolated test server, with a temporary-ban
    threshold of 3, through a trusted proxy: three HTTP Basic logins as an
    unknown user, three login() calls for an unknown user, and three for a
    deleted account each led to a temporary ban (403) of that client, while
    a real user's login() from another address succeeded. The Services
    changes were built, not exercised.

    Fixes #1390

    Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

    ---
    þ Synchronet þ Vertrauen þ Home of Synchronet þ [vert/cvs/bbs].synchro.net
  • From Rob Swindell (on Debian L@VERT to Git commit to main/sbbs/m on Sat Nov 21 19:18:00 2026
    https://gitlab.synchro.net/main/sbbs/-/commit/380c33adbf3f1a62e24ab3b9
    Modified Files:
    docs/v322_new.md src/sbbs3/websrvr.cpp
    Log Message:
    Web Server: refuse HTTP authentication for a deleted or inactive account

    check_ars() looked up the user named in an HTTP Basic or Digest
    Authorization header and checked the password, but never the account's
    status, so the right password logged in a deleted or inactive account.
    An inactive account keeps its name, so it could log in by name; a
    deleted one has its name removed from the name index, but with logins by
    user number allowed (LOGIN_USERNUM) it could still log in by number.
    Every other login path refuses such an account: the terminal server,
    FTP, POP3, SMTP AUTH, and the JavaScript login() of the web server and Services.

    Now such a login is refused like one with an unknown user, logged as
    !DELETED or INACTIVE user, and counted as a failed login attempt against
    the request's client. The password recorded with a failed HTTP
    authentication (the Basic password, or "<digest>") comes from one
    helper.

    Tested on Linux against an isolated test server: HTTP Basic logins with
    the right password for a deleted account and for an inactive one were
    refused (401) and logged, and an active user's login still succeeded.

    Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

    ---
    þ Synchronet þ Vertrauen þ Home of Synchronet þ [vert/cvs/bbs].synchro.net
  • From Rob Swindell (on Debian L@VERT to Git commit to main/sbbs/m on Sat Nov 21 23:54:00 2026
    https://gitlab.synchro.net/main/sbbs/-/commit/aaf3d2395407e9e30d3f54e1
    Modified Files:
    docs/v322_new.md src/sbbs3/ftpsrvr.cpp mailsrvr.cpp main.cpp services.cpp userdat.c userdat.h websrvr.cpp
    Log Message:
    PROXY protocol: check temporary login bans against the client's address

    With HAPROXY_PROTO, the Terminal and Web Servers record a failed login
    under the client's address from the PROXY protocol header, but
    loginBanned() looked up the socket's peer, the proxy, so a client's
    temporary ban never took effect. With the proxy on the same system it
    also matched the server's own address, which is never banned.

    loginBanned()'s unused host_name parameter is now client_addr: the
    client's address when it is not the socket's peer, or NULL to use the
    peer as before. The Terminal and Web Servers pass the address the
    connection was accepted with (from the PROXY header when one was read);
    the FTP and Mail Servers and Services pass NULL. The rule that the
    server connecting to itself is never banned now compares that address.

    Tested on Linux with the Web Server behind a trusted PROXY protocol
    peer and a temporary-ban threshold of 3: after three wrong passwords
    from one client address, its next connection was refused with
    !TEMPORARY BAN, while another client address logged in. The Terminal
    Server change was built, not exercised.

    Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

    ---
    þ Synchronet þ Vertrauen þ Home of Synchronet þ [vert/cvs/bbs].synchro.net
  • From Rob Swindell (on Debian L@VERT to Git commit to main/sbbs/m on Sat Nov 21 23:54:00 2026
    https://gitlab.synchro.net/main/sbbs/-/commit/c2cb2a487863524951f4e2ef
    Modified Files:
    docs/v322_new.md src/sbbs3/websrvr.cpp
    Log Message:
    Web Server: exempt ipfilter_exempt.cfg clients from the connection limiter

    The connection rate limiter (MaxConnectsPerPeriod, SCFG Limit Rate of Connections), applied when a connection is accepted, did not check ipfilter_exempt.cfg, unlike the request rate limiter and the concurrent-connection limit. Behind a reverse proxy that passes the
    client's address in a request header, every connection comes from the
    proxy, so once the proxy exceeded the limit, all visitors were refused
    for the rest of the period, even with the proxy listed as exempt. The
    limiter's auto-filter already skipped an exempt address (filter_ip()
    checks the list), but the connection was still refused.

    Tested on Linux against an isolated Web Server with a limit of 3
    connections per 10 minutes: five connections from the exempt loopback
    proxy were all served; with loopback not exempt, the fourth was refused.

    Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

    ---
    þ Synchronet þ Vertrauen þ Home of Synchronet þ [vert/cvs/bbs].synchro.net