I finally reproduced this issue myself on a fresh install on Windows.

We needed explicit file closing here as going out of scope doesn't immediately
finalize (and close) a File object. That may not happen until garbage
collection and thus prevent subsequent re-opens of the same files, so always
close() your files.

init-fidonet.js calls install-binkit.js which uses cfglib.js, so that was
a source of these errors too.

DD MsgReader and SlyVote: Fixed poll voting for single-answer polls

Closes #642

See merge request !337

A web-based sbbsctrl implementation

See merge request main/sbbs!318

DD File Lister: If used as a loadable module and blanked out the NFilesListed string, then revert NFilesListed on exit of the script

See merge request main/sbbs!336

This should avoid/work-around issue #630

Nelgin's terminal server log indicated that the node's configuration was being
reloaded for a new client connection (which doesn't normally/always happen,
only when there's been a config file change or a node "rerun" requested),
while there was still a user online using that node. This would explain the
segfault (configuration fields being freed and reallocated).

The node_socket[] array indicates when a node thread is actively servicing
a client, so it should contain INVALID_SOCKET for the any node's that WFC.
If it doesn't, that means something/someone changed the node's status value
to 0 (WFC) in the node.dab file while the node was still in-use. Let's not
crash in this case and rather just log the message and give the client the
next available node (if any).

Some of the errors could be pretty vague since they looked the same and
we were depending on the path/file attempted to be opened in the log msg
to clue us into what exactly was being attempted. But some of these file
paths are sysop-configurable, and they can put unexpected things in there
that make it difficult to determine exactly what file was trying to be
opened and for what purpose. This should help a little bit.

DD File Lister: If used as a loadable module and blanked out the NFilesListed string, then revert NFilesListed on exit of the script

DD File Lister: Blank out the "# Files Listed" string if used as a loadable module (optional, defaults to true)

See merge request main/sbbs!334

DD File area chooser: Bug fix for searching with name collapsing

See merge request main/sbbs!335

DD File Lister: Blank out the "# Files Listed" string if used as a loadable module (optional, defaults to true)

This allows the calling script to control what the default item is (when the
user just hits enter), and not always the first item.

Fixes issue #637

DD File Lister 'empty directory' fix, help screen pause, filename searching, etc

See merge request main/sbbs!333

DD area choosers: Header line display fix

See merge request main/sbbs!332

Much of the logic ported from sbbs_t methods in con_out.cpp

For Nelgin's default_settings.js project

IIRC, the error was undefined has no properties or split() is not a function,
something like that.

As requested by Keyop in #synchrob0yyz

tbz is technically still supported by create_archive(), but without it
being included in the supported_archive_formats definition, isn't easily
chosen as an option and nobody seems to miss it.

Apparently one was copied from the other and used the same original
output file subdirectory name ("upgrade") oops.

I didn't notice that these projects were still being built with a toolset
that didn't target WinXP. I might need to update the SDK too.

Also, although not a bug (because we re-read/parse the user's record every
menu cycle), don't decrement user.xedit before calling uselect()
- just not a good practice to not modify variables unnecessarily.
See the corresponding change to exec.cpp, which was a bug.

This bug would leave the user's external editor setting decremented by one if
they aborted (with Ctrl-C).

Fixes issue #631

e.g. 'jsexec delmsgs mail from joe-blow'

This should be enough for Nelgin to make progress with his user defaults menu
script.

Not all sections are read and parsed and there will be some extra properties
(e.g. "code" in repeated sections that don't normally have that), so the
returned objects aren't (yet) suitable for re-writing back to the *.ini files.

I also cleaned up some of the exceptions (throwing strings rather than Errors)
and removed some empty functions.

Most logs sysops are looking at already have a date/timestamp for each
message, so this one ended up with 3 date/timestamps in a single line.

This fixes issue #629

I'll add detailed docs to http://wiki.synchro.net/custom:atcodes soon, but the
list is:

- TERMTYPE
- TERMROWS
- TERMCOLS
- AUTOTERM
- ANSI
- ASCII
- COLOR
- ICE
- RIP
- PETSCII
- SWAPDEL
- UTF8
- MOUSE
- UPAUSE
- SPIN
- PAUSESPIN
- EXPERT
- HOTKEYS
- MSGCLS
- REMSUBS
- FILEDESC
- FILEFLAG
- AUTOHANG
- AUTOLOGON
- QUIET
- ASKNSCAN
- ASKSSCAN
- ANFSCAN
- EDITOR
- SHELL

This should (at least mostly) address the issue #629

Apparently if you call mosquitto_publish_v5() with a non-NULL properties
pointer, it'll only publish via protocol v5 and doesn't work if you're using
protocol v3 connection.

Thanks Ree.

... regardless of the configured MQTT protocol version (v5 or v3.x).

Also include an MQTT v5 user-property ("time") with the message's origination
time/date stamp in ISO-8601 format in all log messages (to both topics).

This will at least double the amount of MQTT log traffic to the broker. When
MQTT v5 is used, the additional user-properties will increase it even a little
more.

Is this what you were suggesting Ree?

This should effectively turn script-terminating errors into just warnings and
thus at least working-around issue #397: the exceptions will be logged as
warnings intead of errors, but still logged.

The request from Nightfox and Accession via DOVE-Net was to be able to set
a user's external editor even if there's no user logged-in.

These 2 user class properties in the JS object model were a bit special in
that they *only* wrote to the user database and did not immediately modify
the in-memory copy of the user_t structure, depending on the re-reading of
the user.dat/tab file to re-populate the current user_t structure when needed.
This didn't work if the current user is user #0 (no user).

So, set the current user_t.xedit and user_t.shell accordingly whenever those
JS properties are assigned a value (a string, the appropriate internal code).

This should fix issue #626