The following bugs have been squashed.
- Tscript mailing once again works
The  d and  dr commands didn't mail their output in release 1.11. Now
fixed.
- Small TCPGATE buglet fixed
When a TCPGATE connection was made, the logfile tried to print an integer
as a string, causing garbage in TNOS/DOS logfiles, and causing occasional
program crashes under TNOS/Linux. Now fixed.
- The BBS 'RM' command fixed
A seemingly long standing bug with the 'RM' command has been fixed. (Can't
believe no one ever reported that one before).
- A long-standing trace buglet squashed
For quite some time (maybe forever) the hex dump tracing of AX25 interfaces
have had a leading zero byte in the display. In TNOS 1.00, I stripped the
first byte, only to find that the AX25 was fine, but others weren't.
TNOS 1.10 and 1.11 were released with the original buglet. Now the first
byte (the bogus zero byte) is stripped if the interface is an AX25, but
it is left alone for all other interface types.
- Fixed the alternate REQSVR "uploadblind" syntax
The "ub filename" alternate syntax for the "uploadblind" sub-command didn't
work due to a now squashed buglet.
- Fixed a minor security buglet in the BBS 'conf' command
There has been a minor buglet discovered in the mbox 'conf' command. If
you use the 'c' command, then permissions are properly checked for the
NO_CONVERS flag, and the attempt fails if the permissions have this flag.
Well, the 'conf' command WASN'T checking this permission.
- New mail buglet smashed
There was a buglet that occurred if you were in the BBS with unread mail
and a new message came in. What happened was that the current message was
set in the routine that checks for new mail to the FIRST unread message,
which is one MORE than what is just WAS. This now preserves the current
message number when new mail comes in.
- BBS buglet, causing unnecessary DNS lookups, is dead
There was a buglet in the isanIPhost() function, which is called when the
user sends a mail message. It had a piece of code, which looked at the
address, and if it was a common Internet address extension, then it would
bypass a DNS lookup.Well, the logic was wrong, and the result was to ALWAYS look up the
address with DNS.
- BBS Login 'ghost' now busted
Thanks to Rob Mayfield <mayfield@guest.adelaide.edu.au> for pointing me
to another problem. Every once in a while (in any xNOS) you will see that
you have a ghost user, that is 'LOGGING IN', but isn't really there. They
WERE, but the connection is gone and the tdisc timer didn't kick in.Well, this was because the tdisc timer wasn't started until AFTER the
login was complete. This has now been moved to before the login, so when
the 'mbox tdisc' timer kicks, the ghost is busted...
- Small filename buglet with tutorial files under Linux
While the intention was to allow a tutorial file to be *.tut, for Linux
this was fine but NOT if you wanted to name it *.*.tut. There could be
only one '.' in a file name.Thanks to Rob Mayfield <mayfield@guest.adelaide.edu.au> for catching this
one....
- Expiration bug found
One of the hardest ones I've had to find was this one....It seems that if the public areas are expiring AND a message comes in for
THAT area while THAT area is expiring, bad things happen. Sometimes the
file(s) get corrupted. Sometimes a lockup. Sometimes the SMTP process
would abort, leaving the ".CTL" file open. When it tries again later, if
the expiration is STILL going on (large file - short timers), then
ANOTHER ctl file COULD be left open. This SOMETIMES resulted in DOS
getting all available files hanging, making a crash eventual.
Well, this one should be good and squashed!
- Forwarding subchannels quirk fixed
If you were using the subchannels feature for forwarding, and not all BBSs
with traffic to pass could start (due to subchannel limiting), then a quirk
occured if the first (or an early BBS) fails, due to not being available.
If this happened, the next forwarding slot would go to them, NOT to the
next one waiting. This only became a BIG problem if the subchannel was
limited to one. In this case, the next ones NEVER did get a chance.This has
now been fixed.
- Statline command in Linux version fixed
I THINK (repeat, THINK) that I found what was goofing up the statline
functionality under Linux, and causing the instability.
The problem was two-fold. If you were NOT on the command session the
display would not update, though the routine WAS being called !?!?
Second, turning it on made you a ticking timebomb waiting for a place to
happen.Well, it seems that it WAS updating the screen, the COMMAND screen. And
after it had been sufficiently backlogged, the thing blew! So you could
stay on the command session all day with no problem, but start to do
something in another session and you would die after a SHORT while.
- BBS Mail scan at startup w/Linux
There was a bug with the mail scan at login in the Linux version. It is
fixed in release 1.12