Deepthaw (DS94):
This was broken in revision 1.584 by deuce, Thu Aug 20 05:19:45 2015 UTC:
The startup->socket_open callback is optional (may be NULL), and in the case
of sbbsNTsvcs.exe, it is NULL, thus the crash when it was called
unconditionally during socket creation. I guess not many people are running
sbbsNTsvcs these days. :-/

- SBBSecho will now maintain a file called the Bad Area File
  (default name is data/badarea.lst), in BACKBONE.NA format
- This file is suitable for importing in SCFG->Message Areas now, but expect
  some automation in the near future
- To get area descriptions (not just echo tags) in the Bad Area File, be sure
  to configure at least one "Additional EchoList" to point a comprehensive echo
  list (e.g. BACKBONE.NA or similar)
- SBBSecho will automatically add new areas (when new messages are received for
  msg areas you don't carry) or remove areas that you start to carry - no sysop
  maintenance of this file should be needed. And the file does not have to be
  deleted after being imported (e.g. with SCFG).
- SBBSecho does not (and may never) actually modify your Synchronet Message
  Areas configuration directly, that is yet to be determined

Another new feature: Area File (areas.bbs) backups:
- Whenever the Area File is modified via AreaFix, the current file is backed
  up  (e.g. areas.0.bbs, etc.)
- This feature can be tweaked or disabled via the AreaFileBackups key in
  sbbsecho.ini (not yet in echocfg), default backup level is 100

* Numerous EchoCfg help text updates.
* Some minor terminology updates (e.g. AREAS.BBS->areas.bbs, Areafix->AreaFix).
* Incremented SBBSecho version from 3.00 to 3.01.

reached (connections with a successful login aren't counted).

1. Randomize the last 3 characters of the internal code to generate a unique
   internal code.
2. Attempt to skip sub-directory names when generating long/short names from
   imported path, if no sub-dirs are in the path, then just right-justify the
   truncated path - trying to generate meaninful long/short names from long
   paths.
3. Do not allow duplicate paths to be imported into the same library.
4. Skip (do not import) directory entries that are > 63 chars in length.
5. Stop importing if/when total dirs >= MAX_DIRS (65534).

If anybody actually hit these limits I'd be impressed, but it's not impossible.
65535 (-1 as an int16_t) is a special value used to indicate an "invalid" item.

1. added option to import empty sub-directories (previously empty subdirs were
    always ignored).
2. attempt to generate a unique internal code when the auto-generated internal
    code (based on the characters in path) is not unique.
3. display "Scannnig Directories..." popup since this can take a while
    (especially on network drives and especially recurisvely).

Note: if you're importing lots of deeply nested subdirs, it's likely that a
unique internal code can't be generated (from the path) for all subdirs.
A randomly generated internal code might be an option if this is really
a problem for any sysops.

Also updated the brace style of this .c file, finally.

globally (using a new macro in sbbbsdefs.h: STR_UNKNOWN_USER). "<unknown>"
in some contexts was too vague (e.g. the 'c' command output from the sbbs
console).

   not daemonized: log output would be repeated (once with a timestamp in
   the log string, once without). Now direct calls to lputs/lprintf from
   sbbscon.c won't go to syslog (in this mode), but I think that's okay.
2. Change the prompt when terminating the servers: replace the ?=help and
   colon implying you can enter commands with just "Termianting..." since
   you can't actually type commands while a terminate is in progress.
3. Output the inactivity timeout values when waiting for servers to
   terminate (e.g. because they'res till in use).
4. Report IP addresses in square-brackes in the client ('c' command) output
   to differentiate from hostnames.

other things to detect/display the actual filename (when different than
what's stored in the database).

This should fix KK4QBN's problem with Windows->Linux migration of file bases
of Nighowl CDs (where the files.bbs contain lowercase filenames, but the files
themselves are uppercase).

Added (back) the '-d' command-line option to disable the deletion of processed
(e.g. imported) inbound netmail messages.
- This option has no effect if DeleteNetmail is set to "false" in sbbsecho.ini.
- This option *only* disables deletion of incoming netmail (no effect on
  outbound netmail deletion)
- In SBBSecho v2, the '-d' option disabled all netmail deletion, even though
  the help output only mentioned "netmail after import" (it was wrong).

update the file path in the message subject to point to the inbound directory
where the packet was found.

name(s) are zero-length strings.

attribute for netmail (both *.msg and in the mail base). This option should
*not* be used when the DeleteNetmail option is enabled (set to 'Yes').
A warning will be logged in this configuration as dupe netmail msgs may be
sent in that case.

by Bill McGarrity.

number is particularly interesting to me given the error's he reported seeing.

sbbsecho.h:186:7: warning: array æpktTypeStringListÆ assumed to have one element
Warning: size of symbol `pktTypeStringList' changed from 8 in .../rechocfg.o

For packet creation, the only difference (from Type-2+) is with packets created
from point nodes: the origNet field won't be set to 0xffff (as is the case with
FSC-48, Type-2+ packets). The auxNet field is now always set to the originating
net (ala Squish) when configured for Type-2+ packet generation.
When importing Type-2e (FSC-39) packets, the detected/logged packet type may be
reported as "2e" for valid "2+" packets. This is normal and expected when the
originating node is not a point and the auxNet field is 0.
The second set of orig/dest zone fields in a type 2e/2+ pkt header will
now be used when importing if the old/QM orig/dest zone fields are 0.
Also, the originating point field may now be imported even if the auxNet field
is not used (origNet != 0xffff).

New umask sbbsecho.ini configuration option (global), maybe used to over-ride
the default *nix file creation mode mask (077), similar to sbbs.ini.

Use more formal terminology (e.g. "Type-2", not "type 2.0").

- %RESCAN returns a different failure result if the requested echo is a
pass-through area (cannot be rescanned)
- %UNLINKED will return only the unlinked echoes even when an "additional
echolist" is utliized and the linked areas are all pass-through (for Bill).

If what appears to be a QWKnet addressed was passed, but the QWK-ID portion was
exactly 9 characters, a 1-byte stack corruption would occur as the 10-char
local variable (node) could be written with 11 chars (9-char QWK-ID, colon,
null). The reported issue happend when attempting to send netmail to
"hemo@ujointbbs" which is neither a valid Internet email address or a valid
QWKnet address.

findstr() and findstr_in_list().
This means that rather than listing an IP address range using the traditional
wildcard characters (e.g. "192.168.1.*"), you could specify a subnet range
using CIDR notation (e.g. "192.168.1.0/24").
The '!' prefix is supported for CIDR match lines (for negated match results),
but no other wildcard/match characters are (e.g. '^', '~', or '*').
You can mix and match traditional match lines and CIDR match lines in the same
filter file.
IPv6 CIDR notation is not yet supported.

- When importing echomail packets and tossing to linked-nodes, don't loop
messages back to originating nodes, even if their address is not in the PATH or
SEEN-BYs.
- To get the full originating address of an echomail message (for loop-back
prevention), parse the Origin Line before tossing to links. If the Origin Line
cannot be parsed, log a warning message.

how "frequent" the progress meter/percentage displayed is updated.
Most uses of this function are now using a interval of 10 so that fast-running
searches aren't now "slow" due to the progress display.

than sprinkling the mapping around the source files.
Also, disable the message thread nav keys (arrow keys, home, end) when not
in Thread View Mode - too confusing.

Removed some unused constant definitions/macros.
Updated copyright date (to 2016).

incoming mail): it turns out, stripping tabs (ctrl-i) is breaks many folded
header fields (e.g. MIME-encoded messages).
So, just strip bare carriage returns instead, as that was the original problem.

While "fixing" compiler warnings, a bug was introduced that would cause an
infinite loop while inserting File Libraries. This is possibly the same
problem reported by KK4QBN.

or Areas.BBS will all be found/opened successfully if "areas.bbs" is used.

Make sure pause ([Hit a key]) prompt doesn't appear on the same line as the
Reading prompt.
The '?' command now turns off "Thread View Mode".
Backspace is now treated the same as '-' (move backwards through sub).
When in Thread View mode, moving backwards now first searches for a thread
with a lower thread ID and if that fails, look for any thread with a different ID.
New Thread Nav command: Home (move to first msg in thread) and End (move
to last message in thread). Last in this context means the last reply to last
reply to the original message (not necessarily the newest reply). This is most
obvious how this works when in Thread View mode.
Moving forward through the sub (+ or <CR>) while in Thread View Mode now
searches first for a thread with a higher thread ID, and if that fails, then
any thread with a different ID.

I experimented with not clearing the screen (homing the cursor and using clear
to end of screen to remove old data), to avoid the "flash" while browing a
thread, but both Netrunner and mTelnet appear to not support the <ESC>[J
terminal escape sequence (clear to end of screen) correctly.

no longer ignores Ctrl-Z characters (CP/M EOF) by default. If you want the
old behavior (Ctrl-Z characters not displayed, just skipped), specify the
P_CPM_EOF mode flag.

HOME - homes cursor (to upper left, 1:1)
CLRLINE - clears current line (cursor to column 1)
CLR2EOL - clears to end-of-current-line (doesn't move cursor)
CLR2EOS - clears to end of screen (doesn't move cursor)

'J' - Clear to End-of-Screen, for ANSI users sends <ESC>[J, nothing otherwise
'`' - Home Cursor, for ANSI users sends <ESC>[H, form feed otherwise