a zero-length one).

This fixes the problem everyone but me has known about for some time now.

too complicated for me to be comfortable supporting. if anyone still would like a copy email me, but i think the better idea is for them to make their own, to their needs.

acknowledgement ("250 OK") can be sent. In this case, the sender is either a
spammer or it is a legitimate delivery attempt that has been aborted for some
reason (e.g. timeout due to throttling).

of two.

Essentially, any bits above the next lowest power of two are less likely
to be set.  For example, xp_random(6) would use 3 bits of randomness with
the following distribution:

000 = 0
001 = 1
010 = 2
011 = 3
100 = 4
101 = 5
110 = 0
111 = 1

As you can see, zero and one are twice as likely to occur than 2,3,4, or 5.

around C++ scoping issues (return from a macro?). :-)

Updated copyright dates.

make system.matchuserdata return only the next match (instead of all matches) from specified starting user if match_next==true. add system.matchallusers(field,data); to return an array containing all matches.

user.dat record offset.

descriptor (-1), this is part of the node.dab error recovery scheme.

make system.matchuserdata return an array of all matches if match_all==true, starting at usernumber if specified. (userdatdupe now takes an additional paremeter)

command, where end of article range was not provided (or provided as 0).
Patch by Ragnarok.

dm writting poor documentate and shitting code like microsoft bug poop

fixed drawFX method (no longer slow as mole asses)

fixed filename property

set js.branch_counter to 0 before execution
display error filename and error line number

fixed cut (when at the end of a file or the file is empty)

(highlight "function", do not highlight "myfunction")
cut and copy now clone a line instead of referencing it (oops)

update column on key up/down