			XROUTER - RECENT CHANGES
			========================

21/11/01	v172a	Test version for G6TJZ's "special" PC120. It allows
			the card to be specified as two separate PC100
			cards at different IOADDR's but shared IRQ.

24/11/01	v172b	Netrom Ping now displays resolved target as in:
			"pinging KIDDER:G8PZT <Return> to quit".

			Added RTT to repeating NPING display, and fixed the
			average RTT which was prone to overflow at RTT > 64
			sec.

			If a "learned" netrom route was obsoleted, further
			frames wouldn't reactivate it - only a node bcast
			would do so. Fixed - it will now reactivate both
			the obscnt and quality upon rcpt of a frame.

27/11/01	v172c	Added PMS and PMS command. This adds PMSCALL
			PMSALIAS and PMSQUAL to the config.

28/11/01	v172c	On extended N comands, RTT was displayed as 0.00
			if not known.  It is instead no longer displayed
			unless it has been measured.

			RTT figures were starting high and taking a long
			time to come down. They also had a tendency towards
			the high side. They should be more accurate now.
			Note they will always be inflated by the other end's
			delayed ack timer.

			Unprotos sent by appls now have FIRST appl's call
			as the source call. If first appl doesn't have a
			call, (e.g. only available from cmd line),
			sourcecall is portcall. BPQ's host API spec doesn't
			allow me to do anything better than this!

			At last, found the bug, introduced in v167b, which
			would occasionally cause the router to crash upon
			closure of a netrom session.

30/11/01	v172d	If no msgs read from disk, first msg in PMS was
			numbered 0 instead of 1 - fixed.

6/12/01		v172d	Bugfix: L4 connects to appls were going to the
			wrong appl!

			Bugfix: Suppressed "Trying hostxx" on L4 connects
			to appls.

			<probably other stuff here?>

18/12/01	v172e	"Hidden" nodes (alias beginning with #) were not
			being entered into nodes table, and L2 downlinks to
			the alias were not possible. This was due to the
			callsign verification, and caused (bad len) to
			be displayed on the incoming and outgoing trace.
			AX25L2v2.0 states that callsigns should be upper
			case alpha and numeric characters only (which
			means even spaces are illegal!), but I've made a
			special exception for #.  Tracing will now say
			"bad call" if the call isn't legal.

			Added "IP ROUTE LOAD" command, to reload IP/ARP
			tables from file after making changes.

			Removed the duplicate IRQ check because it was
			preventing the use of shared IRQ async cards.


21/12/01	v172e	VC routing fixed - it was a problem with the
			reserved SSID bits.

			If port paclen was set to 256 it was causing 1
			byte frames - fixed.

			Since someone seemed concerned about the variable
			paclens, I've disabled adaptive parameters - they
			are now enabled by setting port paclen to 0.


27/12/01	v172f	Bugfix: Node qualities of adjacent nodes was being
			reset to global minqual when a l3 frame rcvd with
			that node as source call.

5/1/02		v172g	APRS static position report will now be recognised
			if it starts within the first 40 chars of a beacon.
			All other reports must start on the first char.

			Fixed code for G6TJZ's pc120 card.
			(Release to G6TJZ only)

6/1/02		v172h	Rewrote SCC initialisation to save bytes and make
			it easier to understand.

                        A simple typo meant one of the DRSI 8536 registers
			wasn't being initialised - fixed.

			Added PC120 type, specifically for G6TJZ.  As with
			other card types, all PC120 interfaces should use
			the same IOADDR and INTNUM, which should be lower
			address.


9/1/02		v172i	ARP ADD was reporting Error! for ethernet addresses
			(but was OK for other types.  Traced to an error
			introduced on 3/8/01 into ethernet address
			conversion function. Fixed.

13/1/02		v172j	I was working to fast to be bothered dcumenting
			it, and now I've forgotten what I did!

20/1/02		v172k	Since v172h, SCC driver wouldn't work if SPEED was
			set to 0 - fixed.

			PC120 being initialised as a BAYCOM, which may
			have been wrong. Fixed.

			Added SCC type ITACARD.  Currently only the first
			chip is recognised because there's some doubt over
			the register layouts in the second chip.  Released
			to Piero Trani for evaluation.  If the first chip
			works, I can enable the second.

			SCC interfaces can now use CHANNEL=[A|B|C|D] or
			COM=[1|2|3|4] interchangeably.  This makes it
			compatible with previous versions, but alllows
			those who can't get to grips with "COM=" to use
			the more familiar "CHANNEL=" instead.

			If a ping interval > 32 secs was specified, it
			generated a pingstorm - fixed.

21/1/02		v172l	I think this one fixed SCC code which I'd seriously
			broken in 171k.

27/1/02		v172m	Rinse Pasma wanted "upper half" IBM graphics to be
			displayed - done.  They were originally filtered
			out to keep the screen tidy when monitoring FBB
			compressed forwarding traffic.

3/2/02		v172m	The PC120 channels were apparently in the wrong
			order. I had assumed that the 8530 at the lower
			address was for channels a/b, and the higher
			address was for channels c/d, as with all other
			makes of card.  Having never seen the card, and
			being unable to get any definitive information on
			it, even from PAC-COM, I can only take other
			peoples' word for the channel layout.  I've now
			swapped them over, so the IOADDR parameter must
			now specify the higher of the chip addresses.

6/2/02		v172n	Ip route lookup was returning FIRST match not
			BEST match - fixed.


7/2/02		v173a	Added rudimentary static NAT.  This adds a new
			set of configuration commands:
			IP NAT <ADD|DROP|LIST>...  can be used at the
			command line, and NAT ADD... is used in IPROUTE.SYS.

			IP datagrams received on hardware multicast address
			were being routed instead of passed up to higher
			layers - fixed.

			Traceroute wouldn't show the IP address of any
			Xrouter in the route - fixed.

			I've incremented the version number because this
			one has new commands and the routing may behave
			differently due to the bugfixes and the code mods.
			This may require some changes to the routing
			entries in IPROUTE.SYS if it was previously
			configured wrongly.  Any change has the potential
			to introduce more bugs, so 173a should be treated
			as an alpha version, even though it's probably
			100 percent OK.


8/2/02		v173a	Added static PAT (Port Address Translation).  Each
			entry in the NAT table includes optional port and
			protocol information which modify the translations.


9/2/02		v173a	Added dynamic PAT.

			I have done my best to test NAT and PAT, but it
			is unfeasible for me to test them in every possible
			configuration.  Further development will depend
			on user feedback.
