logon.lst file format) and thought I came up with a cool new idea. Nope.
It's already a thing:
http://jsonlines.org/
http://ndjson.org/

Here's a library to append-to and read these "JSON Lines", "NDJSON" and
"LDJSON" files.

On underscores versus dashes in filenames:
Apparently my tendency is to use underscores in source/script filenames, while
other CVS contributors (e.g. mcmlxxix) prefer dashes (e.g. json-*.js), so the
filenames are starting to look a bit bipolar. Oh well. :-(

'maxlen' argument. E.g. passing 10 would result in a maximum read string length
of 9 characters.

optionally including the current (your) node.
New @-codes:
"ANODES" (active-nodes count), aliases: "ANODE" and "AN"
"ONODES" (other-active-nodes count), aliases: "ONODE" and "ON"

Make "NN" an alias for "NODE" and "TN" and alias for "TNODE".
Add "TNODES" alias for "TNODE" as well, for symmetry.

Version 1.69: Updated to only use console.inkey() for user input and not use console.getkey() anymore.  Also, SlyEdit will now write the editor style (ICE or DCT) to result.ed at the end when a message is saved.  Also, when editing a message, if the cursor is at the end of the last line and the user presses the DEL key, then treat it as a backspace.  Some terminals send a delete for backspace, particularly with keyboards that have a delete key but no backspace key.

maintained between lines, reported by Immortal@IDOMAIN:
Use P_SAVEATR to retain the new attributes after each call to putmsg() in
printfile() line-at-a-time mode. We are no setting the initial attribute to
LIGHTGRAY or restoring the original attribute at the end (like a call to
putmsg() withou the P_SAVEATR would) - so we may need to add that later
for line-at-a-time mode, as well.

Use adjusetuserrec() for credit adjustments (support full 4GB range).

a CR *and* it doesn't contain a 0xE3 (QWK newline). <sigh>

Use the proper method of detecting an ANSI terminal:
  console.term_supports(USER_ANSI), not (user.settings&USER_ANSI)

Use a user-property, [logon] makguest=false, rather than the 4G flag to
indicate that the sysop has opted-out of creating a 'Guest' account on their
BBS. If you don't have an Guest account and you already opted-out, sorry, this
will prompt you again. However, get the use of a security flag (4G) back for
your own definition/use, as originally intended.

Removed some commented-out debug output (clean-up).

(Sorry Deuce).
If/when HTMLterm is resurrected, this logic should probably go into a
subordinate module that's invoked when appropriate.

(ctrl_dir).
Use the CPM-EOF printfile mode to stop printing files when encountering a
SAUCE record.

This variation on fcntl() record-locks resolves multiple-thread / forking
issue with region locks. No longer falls back to using flock() (whole-file
locks), when OFD locks are available and used successfully. At the least,
this should provide a performance boost for shared files using region
locks (on Linux). No change for Windows (region locks already worked sanely
among multiple threads) and other *nixes (that use POSIX fcntl() locks
only).

(credits) to use the full 4 gibibyte range of a 32-bit unsigned integer:
max-out at 4,294,967,295 instead of 2,147,483,647.
Since these values are stored as 10 characters, we could theoretically extend
the range to 9,999,999,999 (credits/bytes) without changing the user.dat format
but is a bigger change (moving to 64-bit integers everywhere these values are
read or adjusted).

Only retain the last (most recent) 10 data/<id>.qwk.*.bad files and
data/file/<user-num>.rep.*.bad files.

(CID) variable, even when the client isn't SEXPOTS.

TERMINAL LOCATION NUMBER (RFC 946) in sbbs_t::telnet_location.

Some (older?) versions of fTelnet request to send this option to the BBS to
share the real IP address of the client. Accept this option (when using
"format 0") and store the first 32-bits (an IPv4 address) in the telnet
location storage / caller-ID information. The same place the telnet
SEND LOCATION value is stored if received.

Now, we're not requesting this option (e.g. with a Telnet "DO" command), but
fTelnet doesn't need that: It offers to send it, so we accept.

to put in the description (in strftime() format). Can also be used in
combination with '-t' if you want today's date/time instead.
It's not very clear from the help/usage output, but if you specify both -f
(or -F) and -t, you'll *only* get today's date in the description (not both
the file date and today's date).

"kludges ... should be ... terminated by either a carriage return, or the
(0xE3) QWK terminator" :-(

(not delete) from the end of the sorted list of files (normally, 0).
Created an sbbs_t::delfiles() wrapper which defaults to keep 0 files and also
generates a error.log message if any of the file-deletions fail.

(not delete) from the end of the sorted list of files (normally, 0).

deprecated

Elsewise if it's a number, treat it as a port.
The last "hostname" and "port" encountered on the commandline according
to these rules is what the websocket server done gonna connect to.

"Should" fix problem reported by Mortifis.

one.

Version 1.68: Each paragraph of text written in a message is now saved as one long line so that they can be wrapped appropriately according to the terminal/reader width (which should work online as well as in offline readers).  Also, message uploads are now supported by typing /UPLOAD (or /UL) on a line by itself (the uploaded message will replace any message written into the editor).  Also, now honors the SUB_ANON and SUB_AONLY flags for the sub-boards when cross-posting so that the "from" name is "Anonymous" if either of those flags enabled.

is being forwarded to an external email address. Introduced in rev 1.690,
user-forwarded emails wouldn't work because sending envelope would contain
the original forward-path. Thanks again to Alterego (ALTERANT) for testing
this stuff and reporting issues with detail!

(new). Reported by Alterego (ALTERANT)

A few *service.js scripts call this function without a password argument
(the second argument), e.g. login("guest");
If there was no guest account (or the guest account had a password assigned),
this would result in a failed login attempt as "guest" along with a garbage
password (e.g. a floating point number, like 3.7042561) and since it would
be a unique garbage password for each login() call without an actual password
specified, these login() calls would be counted as unique failed login attempts
and potentially cause the client's IP address to be added to the hack.log
and even ip.can (IP address filter).

As seen on Mortifis' system where VERT was filtereed due to
"SUSPECTED NNTP LOGIN HACK ATTEMPT", likely due to the daily sbbslist
verifications when just perform a TCP connection and no actual login attempt,
but nntpservice.js would still call login("guest") before the client (vert)
would be disconnected.

would reuse the same buffer where the message subject (list of filenames)
was stored causing havok. Reported by Alterego (ALTERANT).

P_OPENCLOSE mode flag is *not* specified, do not malloc()/read() the entire
file and then display. Instead, print one line at a time (so long as that
line is <= 1MB in length), calling putmsg() for each line.

This should allow the viewing of massive text files in SBBS again without
alloc/swap issues.

I left the calls to utf8_normalize_str() in here, but I'm not so sure about
them now. putmsg() will convert UTF-8 to CP437 through print_utf8_as_cp437(),
and that only lacks a couple of conversions that utf8_normalize_str() does
(e.g. ellipsis to "..."), so that might be a good candidate to remove in
the future.