<input type="image" src="/graphics/delete_messages.gif" name="Action" value="Delete" />
(Specifically the name= and value= )  Sheesh!

Should be set to true when many updates are being done and they don't need
to be shown as they happen.  After setting back to 0, any ciolib function
should do the full update.

Win32: The cursor doesn't move while hold_update is set.
Curses: No refresh() is called while hold_update is set.

This allows programs to be faster without compromising conio compatability.

updated immediately.

name.

notation (for Deuce).

filenames with spaces in them. <sigh>

(sizeof returns 4).

conflict with user-defined variables and more consistent with existing "errno"
vars.

Doesn't need to operate differentely across platforms, and you CAN provide
argumes in the .ini file.

NOTE: Looks like executable was never actually a requirement.

terminator silly)

cgi_env.ini
web_handler.ini
headerless CGI scrupts.

get_cgi_handler() modified to behave differently between Win32 and *nix.
The script still MUST have execute permissions.  I may relax this someday
if anyone points out a reason for it.

socket errors (like failed resolve_ip() or resolve_host() calls). On *nix, this
value will be the same as "errno", but on Win32, it's different (derived from
GetLastError()).

error).
If an exit_code was passed to the exit() function, enqueue that value instead.

initialized as such by the caller).

It's been determined that "null" is the better error return value for functions
(for one, it's easier to to enqueue/dequeue).
Eliminated unnecessary *rval = JSVAL_VOID assignments (this value is already
initialized as such by the caller).

"infinite".
Enhancements: Queue.peek() and Queue.read() now support an optional timeout
argument (in milliseconds).

cleared by waiting on it (in multi-threaded builds).
msgQueueRead() will now attempt to dequeue a message from the linked-list even
if the semaphore wait fails.

the result of the last evaluated statement if it was "undefined" (many
functions returned "undefined" on failure).

without converting to an empty object or segfaulting.

null on look-up failure).

understand correctly (it's a Win32 thing) should be set when the xpdev
stuff is being imported from a DLL.  So, the xpdev cuntions something
with WRAPPER_IMPORTS is linked against should be built with WRAPPER_EXPORTS
set.  Currently, I don't belive xpdev is EVERY built with WRAPPER_EXPORTS
set except when building a DLL... and Synchronet does not use the DLL
version of xpdev.

This SHOULD fix the build errors for BCC.