active_clients (protected integer) could be 0 at the time of recycle because
there can be delay at the beginning of the FTP ctrl thread (e.g. looking up
filtered IPs/hostnames) before the active_clients is incremented which could
be during/after it was destroyed by the cleanup() function. Now tracking
number of threads using protected integer which increments immediately upon
child thread creation and is not destroyed until the main thread terminates.
There is currently *not* timeout while waiting for child threads to terminate
upon shutdown/recycle, so this could result in a hung server thread if a child
thread is indefinitely hung - consider using a timeout.

Using C99 types (e.g. uint8_t instead of uchar).

did not work (since rev 1.96, Tue Sep 27 09:36:26 2005):  after selecting such
an editor available to a user solely because of their user number, their editor
setting would automatically revert to 0 (none/internal editor) because the user
number used for the ARS check in getuserdat() was 0. Fixed by setting the
user->number value before calling chk_ar().

Convert binary export pointer file (*.sfp) to a field of the <subcode>.ini.
Strip control character (e.g. new-line sequences) from log file output.
Paranoid use of parenthesis around bit-wise operations.
Resolve some 64-bit warnings/problems (e.g. ulong msgptrs).
When exporting a reply to an existing message, do not "generate" a REPLYID
field if the original (replied-to) message did not have a MSGID (bug reported
by Mark Lewis).
Convert CP437 chars used for creating sbbsecho.msg to hex to resolve viewcvs
UTF-8 python error:
UnicodeDecodeError: 'utf8' codec can't decode byte 0xdc in position 147782: invalid continuation byte

This should really get a new file with happy wrappers at some point - I'm not
sure where lprintf() will go if it happens from main() right now etc.

Fix Warnings by logging cryptFlushData() errors.

size_t type where ssize_t was formerly used.  Is the correct size and type
and should work on Windows where the Microsoft compiler and the API are
missing the ssize_t type.

Also, remove illegal assignments to casts.

Now when you can't allocate the second GB of RAM to show your billion line
scrollback, will simply not show you the 1GB of scrollback you have amassed.

You still won't be able to see it, but it won't crash anymore.

Reported by Digital Man.

was made multi-instance safe.

This *may* break any exit() calls in *nix OSs.

unmoderated message in a sub.

Per RFC3659, the correct response for "Good command, but no size for you" is
550, not 500 or 501 (these indicate an inability to parse the command).

This works with Chrome.

before continuing.

Use multiple buffers for Win32 to avoid start/stop related clicks
malloc()/free() buffers where possible.
Add a xptone_complete() function intended to wait for the current tone to stop playing.