This does not fix the underlying problem, just prevents twiddling outside
of the array bounds.

File directory storage paths that were left blank in SCFG were not correctly
initialized to <data_dir>\dirs\<code>\ as the dir's internal code was not yet
initialized at this point in the load_cfg() process of scrubbing the cfg data.
This bug was introduced with the internal-code prefix enhancement of 2003
(almost 11 years ago now).

lengths.  SAFECOPY() isn't anymore so don't.

Thanks Access Denied for sticking with it.  This one should be more workyer.

sbbsecho.cfg line lengths and addressess all "%compression none" AreaFix
command issues (according to Deuce).

Note that it still assumes that teach arctype uses one line only.  It will only add
the specified node to one line, but every time it comes across a line with
the old type, it will replace it with a new line containing every node which uses
the old type.  This is likely not what you want, so only use a single USEPACKER
line per archiver type is you're modifying this by hand.

I'm looking at you access_d.  Your welcome.

This does not mean you can actually *USE* lines longer than 256 characters
though because the areafix stuff will still take a dump on your config if you
do.

buffer size, or really do anything except fgetc() in a loop.  Do that then.

Also, use DLLEXPORT/DLLCALL/etc. macros.

DAMN YOU ALAN!

May fix a crash, may eat your cat.  If it breaks, you get to keep both
the pieces.

Clang warnings.

Some people just want to watch the world burn.

requested by Joe (1:249/303), Wilfred (2:280/464) and Michiel (2:280/5555) for
their little 3-node inter-zonal FTN:
"Zone Blind" option (enabled in echofg->Toggle Options or by adding a line
containing the string "ZONE_BLIND" to the sbbsecho.cfg file). This option
causes zones to be ignored (back to the 2D addresses of the 80s) when parsing
and generating SEEN-BY and PATH kludge lines in echomail messages.
Incremented version to 2.23.

aren't the same it's assumed to be an IPv6 address.

(accidentally bumped 2 versions in previous commit)

memcpy calls and explicit setting of fields to 0) - this code is just bad.
The only functional change should be that the SBBSecho product code (0x12FF)
and major/minor version number is now added to headers of type 2+ packets
generated by SBBSecho. Type 2 and 2.2 packets will have just 0xff for the
product code and the SBBSecho major version number in the product "serial
number" field (as is the common practice). I guess everyone's using type 2+
packets these days? You don't get 5D addresses with 2+, but SBBSecho doesn't
really do anything with 5D addresses anyway. <shrug>

Print wanring if type 2+ orig/dest zones don't match the zone fields in the
type-2 portion of the header.
Updated some pkt header field names in fidodefs.h.

system's address was being added to SEEN-BY list twice when destination zone
was different than the system's zone.
Another experimental change for Joe (1:249/303) and Wilfred: when parsing
SEEN-BYs from incoming packets, assume the zone of the nodes in the SEEN-BYs
is the same as the origin zone of the packet, rather than the destination zone
of the packet.

that MSVC RTL functions will not terminate the program (e.g. jsexec, sbbsctrl)
if an invalid parameter is detected (e.g. unsupported format specifier in call
to strftime()).

this one really does use the message number now.