CHANGES

Version 0.1: original version.

Version 0.2: fixed bug when several reads are necessary
on one end or the other before a write flushes them.
Fixed bug which threw away data not yet sent to the
other side on close, when running under Linux. Fixed
associated bugs that probably affected other operating
systems as well. Fixed bug causing long, perhaps
indefinite pauses when a possible connection to a
server socket went away before the accept() call,
resulting in a blocking call.

Version 0.3: fixed additional bugs relating to
the code previously used only by non-Linux OSes.
This should fix problems such as connections not
going away when they should or connections being
mysteriously closed. Most of that code is now used by 
Linux also, so it is likely that rinetd is much closer 
to bug-free on non-Linux platforms. Of course, I don't 
actually have any to play with it on.

Version 0.4: added support for kill -1 (SIGHUP)
and specification of service names instead of
port numbers. Removed calls to realloc(), replacing
them with code that should fail gracefully without
crashing the program or breaking existing connections
when another application is hogging memory.

Version 0.5: added logging in both tab-delimited
and web-server-style formats. No longer exits if
an individual configuration file line generates
an error. Added allow and deny rules. Added 
-c command line option to specify a configuration file.

Version 0.51: fixed failure to check for an open
log file before writing log entries.

Version 0.52: documentation added regarding the
ability to bind to all IP addresses, if desired,
using the special address 0.0.0.0.

Version 0.6: ported to Win32. Various compatibility
fixes were made and some minor oversights without
functional consequences were corrected.

Version 0.61: fixed a bug in 0.6 which completely
broke rinetd under Linux. Oops.

Version 0.62: fixed a potential buffer overrun;
prior versions failed to reallocate one of the
arrays correctly when reallocating memory to
accommodate more connections.  Thanks to 
Sam Hocevar.