Add 3.12a CVS dist.
Add cvs-mirror for CVS dists.

Move X_HEADERS and X_PATH macros into Common.gmake rather than GNUmakefile
for no good reason.

link with sbbs.dll).

problem of the JS global function random() not being seeded by non-sbbs.dll's
(e.g. websrvr.dll) - causing predictable random numbers to be returned.

single-threaded Unix apps. :-)

name or it confuses the MS linker), segregating single and multi-threaded
intermediate files (solves sexyz linking problem).

actually used for SSJS, not CGI).
The web server verifies the existence of the temp dir during initialization.

(not SyncTerm's fault!).

Clear screen first and wait for a keypress before exiting (Win32 may not be
started from a console, and some curses implemetations clear screen on exit)

responding to options as a server would (rather than a client).
It now responds to terminal type requests and sends window size, however the
window size is currently incorrect (always 80x24?) - I'll let Deuce fix that.
Eliminated some bcc-reported warnings.
Added CVS ID keyword.

execute after cleanup() is called, bad things happen (e.g. segfault).

There's still a bug here (in Win32 build at least), when running
"syncterm somebogusparam", it prints the usage all on one line.

as the height of the listbox rather than calculating it.
Allow list title to be NULL.

initiated with the -l (loop) option.

with no borders... (No jokes please!)

themselves now.

with large buffers.
Fix post_offset calculationduring CGI execution.
Handle CGI exit better.

respectively in non curses/ANSI modes.  This allows broken programs which
use CTRL-Q (Xon) and CTRL-S (Xoff) to "work" as well as they do for other
broken terminals in those broken programs.

NOTE: By passing these through regardless of binary mode, I'm currently
breaking the RLogin spec... TODO!

If the screen is 24-rows, disable the status line automatically.
Allow ALT-H to hang-up as well as CTRL-Q (Not documented)

Always prompt for User/PW now that ALT-L exists.

separate file in the format:
(prefix)vhostname-date.log

*always* free the log entry... regardless of the file being open
(Memory leak)