Risc OS 6 upgrade startup error

I am trying to set up a Pipod on my A7000.
It requires !VNCsvr which is optimised for Risc OS 6 so I have upgraded my 4.39 to 6.2 softload.
I have already failed to get it to work with 5.22 ROMs.
On startup it reports “Bad Domain Name”, but this error message overwrites what is probably the line containing the problem too quickly to see it.
The OS then loads and generally appears to run OK.
I test the network with Sunfish to connect to my ShareNFS on my NAS drive – working OK before and the network settings have carried across OK.
Sunfish now reports “No route to host” which may/not be related to the start up error.
Anyone able to help?
Thanks,
Ray

On RISC OS 6 if you type *dmesg in a TaskWindow it will show you the boot logs.

The line before the error is:
!Boot.Choices.Hardware.Disabled.Boot.PreDesk.SetupNet
That file has one line:
Run BootResources:!Internet
That applications !Boot is:
| !Boot file for !Internet
|

IF “<Inet$Path>” = "" THEN Set Run$Path <Run$Path>,<Obey$Dir>.bin.
IF “<Inet$Path>” = "" THEN Set Inet$Path <Obey$Dir>.
IF “<InetDBase$Path>” = "" THEN Set InetDBase$Write <Choices$Write>.Internet.Files
IF “<InetDBase$Path>” = "" THEN Set InetDBase$Path Choices:Internet.Files.
IconSprites <Obey$Dir>.!Sprites

There is a !Run with lots of tasks.
Since the network is set up just as it was previously, I’ve no idea what is triggering the error – Repeat: Bad domain name

You may perhaps start by checking the state of your interfaces with
*ifconfig -a

which will list the interfaces avalailable, then type the name of the interface (depends on the card used), for example:

*ea0

this will perform and report some checks on the card.

My card use the Ethernet 3 driver, and I remember that the ether3-16 module failed to initialise due to the absence of MEMC related stuff in RISC OS 6.
I got it working by patching the module to skip the checks.

Here are the results:
*ifconfig -a
lo0: flags=8049<UP,LOOPBACK,RUNNING,MULTICAST> mtu 16384
inet 127.0.0.1 netmask 0xff000000
eh0: flags=8863<UP,BROADCAST,NOTRAILERS,RUNNING,SIMPLEX,MULTICAST> mtu 1500
inet 1.104.38.65 netmask 0xff000000 broadcast 1.255.255.255
ether 00:c0:32:00:26:41
*eh0
File ‘eh0’ not found
*lo0
File ‘lo0’ not found
I guess two responses since the network card has 10baseT (eh0) and 10base2 (lo0).
eh0 reports RUNNING but I am not really any the wiser.
Ray

The command you want is likely to be *EHInfo

Try *Help eth.

lo0 is the loopback address, nothing to do with your hardware.

Note that 1.×.y.z is not a private Class A address – they are of the form 10.×.y.z. Even when you get the networking working you may be inadvertently masking out 1/256th of the internet from your network.

1.×.y.z is the default address and 255.0.0.0 the default mask for RISC OS 2, 3 & 4 (Also possibly 6).¹

Many RISC OS users have/do use it. The only problem I’ve heard of from using such a range locally is the recipient of one email from such a machine saying their system had flagged it as incorrect. It did still get through though!

¹ I’m not sure if RISC OS 5 has always used 10,x,y,z

1.×.y.z is the default address and 255.0.0.0 the default mask for RISC OS 2, 3 & 4 (Also possibly 6).1

It was the default on the AUN config which was not expected to communicate outside the local network. Addresses like 1.1.1.1 were commonly used (slightly irresponsibly) by various manufacturers as 1.×.×.x was not global allocated IP space. Cisco actually had (may still have) documents that specify the use of 1.1.1.1.

This became a real pain in recent years when the 1.1.1.1 block was allocated and then global routers had to pass traffic to it. Plans were amended to deal with the global abuse of the IP.
1.×.×.x is a routable IP set, 192.168.×.x, 172.16.×.x and 10.×.×.x are not publicly routed and as such are “safe” to use in home networks

I’m not sure if RISC OS 5 has always used 10,x,y,z

The default was changed after I pointed out the issue the other year.¹
The issue remaining in RO 4.3x etc is a consequence of the lack of development.

¹ Network Nazi? Not so much. More a case of avoiding problems early.

… you may be inadvertently masking out 1/256th of the internet from your network.

Wouldn’t that only mask out when accessing a site by number?

Every site is accessed by number.

The local network is configured to 192.168.×.x the other addresses are nothing to do with how I have configured the network.
Help, see below, shows MEMC1a is required.
I seem to remember this being mentioned in the install instructions for 4.39 or 5.22 but the i-cubed interface was listed as supported and always worked.

*Help eth.
==> Help on keyword EtherH (Module)
Module is: EtherH 4.33 (27 Nov 1997) © i-cubed ltd. 1997

Commands provided:
EHInfo EHTest

Configuration keywords:
EHVirtual
*EHInfo
EtherH Version 4.33 © i-cubed ltd. 1997

Card Info:- i-cubed, EtherLan 600 Ethernet interface
Unit 0, slot 8, 16-bit driver.
Ethernet address=00:c0:32:00:26:41
10BaseT (twisted pair) interface in use.

I/O Stats:- Rxframes=87, Rxerrs=0, Txframes=33, Txerrs=0, DMAframes=0
Collisions=0, Rejects=1, NoMbuf=0, NoDMA=0
RXPending=0, DMABusy=0, TXActive=0, IRQActive=0

Interface:- Twisted pair interface:- link is good
Frame Info:- Type=8035 FrmLvl=1 AddrLvl=1 ErrLvl=0 Handler=3c7d454
Type= 806 FrmLvl=1 AddrLvl=1 ErrLvl=0 Handler=3c7d454
Type= 800 FrmLvl=1 AddrLvl=2 ErrLvl=0 Handler=3c7d454
Type=IEEE FrmLvl=4 AddrLvl=2 ErrLvl=0 Handler=3d0359c

Controller Mode:- Multicast, reject frames with errors

Driver Error:- This driver requires MEMC1a

*EHTest
Diagnostic tests started
Diagnostic tests complete

EtherH Version 4.33 © i-cubed ltd. 1997

Card Info:- i-cubed, EtherLan 600 Ethernet interface
Unit 0, slot 8, 16-bit driver.
Ethernet address=00:c0:32:00:26:41
10BaseT (twisted pair) interface in use.

I/O Stats:- Rxframes=112, Rxerrs=0, Txframes=35, Txerrs=0, DMAframes=0
Collisions=0, Rejects=2, NoMbuf=0, NoDMA=0
RXPending=0, DMABusy=0, TXActive=0, IRQActive=0

Interface:- Twisted pair interface:- link is good
Frame Info:- Type=8035 FrmLvl=1 AddrLvl=1 ErrLvl=0 Handler=3c7d454
Type= 806 FrmLvl=1 AddrLvl=1 ErrLvl=0 Handler=3c7d454
Type= 800 FrmLvl=1 AddrLvl=2 ErrLvl=0 Handler=3c7d454
Type=IEEE FrmLvl=4 AddrLvl=2 ErrLvl=0 Handler=3d0359c

Controller Mode:- Multicast, reject frames with errors

*

You’d need a 32-bit EtherH to work with RISC OS 5.22. Try the one in NIC Drivers in https://www.riscosopen.org/content/downloads/riscpc

I have a 32-bit EtherH that works fine on both 4.02 and 5.22 on my SA RISC PC.

Have you tried re-entering the networking details under RISC OS Six?

I had re-entered the network details under ROS6 but had another look.
The only place where ‘Domain’ is mentioned is when setting the Local Domain.
Following the instructions on RISCOS.org for 4.39 Select, this was set to .nom and worked under 4.39
However, I have reset it to a name with just characters and the start up error has gone.
Thanks for your help with this.
Ray

You’d need a 32-bit EtherH to work with RISC OS 5.22. Try the one in NIC Drivers in https://www.riscosopen.org/content/downloads/riscpc

Reminder:
Read the install instructions fully. The flash has a copy of the SharedCLibrary which needs updating for the updated version of driver to work.

Chris – as Frank says, every site is accessed by number. We use names because it is a lot simpler to remember riscosopen.org than 91.203.57.172; plus the use of names allows virtual hosts where one machine serves several different websites (quite a common hosting arrangement).

A quick way to get an IP address of something is to ping it.

*ping riscosopen.org
PING riscosopen.org (91.203.57.172): 56 data bytes
64 bytes from 91.203.57.172: icmp_seq=0 ttl=54 time=779.865 ms
64 bytes from 91.203.57.172: icmp_seq=1 ttl=54 time=290.407 ms

--- riscosopen.org ping statistics ---
2 packets transmitted, 2 packets received, 0% packet loss
round-trip min/avg/max = 290.407/535.136/779.865 ms

Try http://91.203.57.172/

On the other hand, my website (and several others) accessed by IP address gives a default page (http://193.33.179.237/) because they are virtual hosts, so they need the domain name to be given in the request. Without that, it just reverts to a default does-nothing setup.

Your own site ( http://193.35.57.242 ) takes you to a default “oops” page for the same reason.

BTW, why does cje.co.uk give a plain “We’ve moved” message? If you set up the server to respond with a 301 Moved Permenantly response, browsers will automatically redirect and smart browsers may also offer to update bookmarks.

We use names because it is a lot simpler to remember riscosopen.org than 91.203.57.172;

Ah, but in the local subnet L2 communication is MAC to MAC. So when you ping something local the stack at each end translates it to the MACs. Better bouncing through a router means your device sees the other device as a L2 connection via the MAC of the router interface facing the device.

So communication is from something like 00:c0:32:00:26:41 to something like 56:26:0A:3B:07:2E via 00:24:17:e2:03:40 and 00:24:17:e2:03:41
Somehow MyPeeCee.home and Router.home seem a bit easier.

Chris – as Frank says, every site is accessed by number.

Yes I do know that, but doesn’t the router NAT take care of that?
Tens of thousands of RISC OS users have used 1.×.y.z for many years and I’ve never heard of a single site that wasn’t accessible!

Tens of thousands of RISC OS users have used 1.×.y.z for many years and I’ve never heard of a single site that wasn’t accessible!

and for years the IP block 1.×.y.z was not allocated and the global routers of the internet were programmed to drop the packets aimed at the block into the proverbial bit bucket.
That is part of the reason that the oiks in Cisco recommending it for “circuitless IP” use (or indeed Acorn using it for a “local” networking protocol) could do it and not be bombarded with complaints from the internet authorities.

A quick check on the simplest IP in the block for obsessive/compulsives

Tens of thousands of RISC OS users

Tens of thousands ? :-P

have used 1.x.y.z for many years

They also used MOVS PC,R14 for many years. Or rotated loads. Or…

As Steve said, the 1.x.x.x block was unallocated back then, and some idiot vendors decided that “unallocated” was a euphemism for “available for use” with no concern as to what would happen if those machines ended up in the public internet with those addresses being, you know, allocated.
Really, it would not have been hard for Econet to be carried on a proper local address. Econet topology is for up to 126 “networks” (zero means "my network) and up to 254 stations (0 & 255 have special meaning), so it could fit into an 192.168.net.stn pattern, and careful use of network numbers and station allocation could arrive at a situation where Econet and normal TCP/IP packets share the same interface (so both Ethernet and Econet could be active at the same time).
That would surely have been better than spamming 1.something addresses.

Tens of thousands ? :-P

Well I think Acorn made well over 100,000 RISC OS 3 computers! A significant number will have been connected via ADSL to the internet. Maybe drop the plural and say ‘ten thousand’ odd!

As Steve said, the 1.×.×.x block was unallocated

If NAT doesn’t sort it I’m surprised someone running 1.×.y.z hasn’t come across a site on 1.×.y.z!

I’ve tried unsuccessfully to find a site that uses 1.×.y.z
Anyone know of one?

I’ve tried unsuccessfully to find a site that uses 1.×.y.z
Anyone know of one?

Doesn’t have to be running a site. It just needs to be allocated to something.

https://db-ip.com/1.2.3.4

If NAT doesn’t sort it I’m surprised someone running 1.×.y.z hasn’t come across a site on 1.×.y.z!

Not all routers do what a home router does and NAT by default. Worse, you possibly have noticed that Cisco are a big firm and might just possibly have sold a few items that are in use ‘out there’ on the net.

Oh, and NAT can cause problems. It isn’t a magic cure. It most certainly doesn’t allow the user of a 1.×.×.x address to communicate with a real world instance of the same or adjacent addresses.
Entertaining idea for you: Change your internal address range on your router to 91.203.57.1 – 254 (91.203.57.1 router gateway IP and subnet mask of 255.255.255.0) and then browse the net. Google will be fine, as will most of the net. You may hit difficulties communicating with us guys…

I’ve tried unsuccessfully to find a site that uses 1.×.y.z

Really? I find that simply putting an IP into the search box of a search engine like google tends to offer various links that refer to web sites with suitable lookup services.

Anyone know of one?

Well since you ask I’ve decided to put another number sequence into the google search page and come up with this information page

I wouldn’t put this down as a complex query.

https://db-ip.com/1.2.3.4

That’ll do, French based lookup site, offered early in the listed links?

offered early in the listed links?

It’s been my lookup site of choice for years…