Econet changes

This probably won’t affect many (any?) people, but I notice with the recent change to Econet to resync with the allocations database that Port_TeletextServerPage has changed from &B3 to &B4.

Given that this function is listed on MDFS.net as port &B3, what is the rationale for this change?

I put in a request prompted by work on !Machines after the recent Econet day at TNMoC. There was a bit of head scratching from the allocations manager, presumably because nobody has asked for anything Econet related in 20+ years (before ROOL took on administering the allocations).

Turns out the Econet allocations very closely matched those in the already published header, so the simplest thing to do was copy the allocation manager’s ones into the published headers (minus any confidential ones). As you spotted this revealed a discrepancy.

So…there’s no rationale: they were out of sync, and I assume the allocation manager’s copy is king because nothing in the ROOL sources referenced the published ones.

I assume the allocation manager’s copy is king

I’d be careful saying that about oldey-timey legacy stuff from the 6502 era. ;)

That being said, the MDFS document does mention that one of the sources was my ancient Econet stuff, so maybe it came from there and has always been wrong?

Does anybody have / know / remember the teletext server to confirm what the correct ports are? Unfortunately my various searches didn’t turn up any documentation, and the AUG doesn’t even list ports, only “this block is user, this block is Acorn, etc”.

Later: Jeez, one could be forgiven for thinking that the Teletext Server never actually existed. Was it Acorn? Was it a ROM? Or something that loaded alongside ATS/TFS? Google isn’t much help, Bing is worse…

Checking through my archives, it looks like I did some reverse engineering and added Teletext server support to aund (never released, though…). I think that Port_TeletextServerPage is &B4.

There were at least two teletext servers, TServ and TeleView: https://mdfs.net/System/Teletext/

I remember being involved in discussions to merge networked teletext server protocols and update code to match, and checking the actual code shows the ports are:
&B2 TeletextServerReply – reply from server to client
&B3 TeletextServerCommand – command from client to server
&B4 TeletextPageData – page data from server to client
&B5 TeletextHeader – header data from server to client

The protocol is described in TServ.zip→Manual (but there’s 256 bytes missing due to a bad disk block)

☝️
Authoritative ×2

So, &B4 it is (and has been otherwise wrong all these years).

Thanks, guys.

Interesting, is that originally my !Machines (It would say !Machines, List Econet stations, A.Williams, 2.03 in the info box) or is there another implementation with the same name?

I was looking at its just the other week and considering hitting rool up for some Econet allocations so they could be in for quite a flurry of activity.

I will be after some new machine type allocations, hence the reason I had !Machines on the blocks.

I am thinking from the pi point of view where there are presently about three different Econet implementations.

Personally I would like !Machines to be able to tell me if its a Pi 2,2a,3,4,5,w,w2 or CM4 and if its RISC OS or Linux etc. Not all those combinations are possible, the Linux solution I have is not good below pi4 obviously the RISC OS solution not above 4. Though if I wrote Linux or FreeBSD drivers for my hardware there is no reason it wouldn’t be happy on a 5.

What I have been wondering is if its important to know what the interface is in addition to the pi type and the OS. Presently meaning either of the two ADF10 adapters, Chris Royal/Ken Low PI hat, James Rayner’s PICO USB adapter or my adaptation of Andrew Gordon’s PICO Econet line interface to RISC OS on a PI hat.

I am likely to raise the same discussion over on stardot as its where most of the Econet activity seems to be at the moment, before compiling a list and getting everything new that’s floating about at the moment an allocation.

Alan

What I have been wondering is if its important to know what the interface is

Strikes me that this is hardware level stuff, and for the user it probably wouldn’t matter beyond “the OS type” (in case of subtle differences or, if anybody still does it, the remote peeking and code execution ¹).

¹ How innocent those halcyon days were that remotely poking memory and then executing it was by design…and seemingly nobody stopped to think that such a thing with bored teenage boys around might be a massive bucket of fail.

1 How innocent those halcyon days were that remotely poking memory and then executing it was by design…and seemingly nobody stopped to think that such a thing with bored teenage boys around might be a massive bucket of fail.

Totally, I remember one of my first experiments into code with adjustable mark/space ratios had calls to remote machine halt and continue in it.

Later when I was working for Barsons in Melbourne I heard of kids who would stick staples into the DIN sockets, and then colour them in with back texta (that’s Australian for felt tip pen)

Interesting, is that originally my !Machines (It would say !Machines, List Econet stations, A.Williams, 2.03 in the info box) or is there another implementation with the same name?

It was musings about updating or rewriting in this thread.

Personally I would like !Machines to be able to tell me if its a Pi 2,2a,3,4,5,w,w2 or CM4 and if its RISC OS or Linux etc. Not all those combinations are possible, the Linux solution I have is not good below pi4 obviously the RISC OS solution not above 4. Though if I wrote Linux or FreeBSD drivers for my hardware there is no reason it wouldn’t be happy on a 5.

What I have been wondering is if its important to know what the interface is in addition to the pi type and the OS. Presently meaning either of the two ADF10 adapters, Chris Royal/Ken Low PI hat, James Rayner’s PICO USB adapter or my adaptation of Andrew Gordon’s PICO Econet line interface to RISC OS on a PI hat.

Like Rick, I’m not convinced proliferating machine numbers helps much. While it might be fun to show them in !Machines, and keep someone busy designing icons, in most instances the actual distinction isn’t relevant.

I think that’s why Acorn just has “Archimedes” as a machine type rather than one for every model – from the client at the other end’s point of view that’s accurate enough. When doing a machine peek you get back the machine type and version of N(et)FS, and again as a client just knowing “it’s a Pi with a tonne of RAM and a really fast processor” is probably the level you want to compare with “I’m a Z80 please don’t talk too fast”.

“I’m a Z80 please don’t talk too fast”.

I’ll have you know that the Z80 in an MDFS was more than capable of keeping up with an A5000, even with the clock tweaked for extra speed.

The FileStore, on the other hand….. ;)

> While it might be fun to show them in !Machines, and keep someone busy designing icons, in most instances the actual distinction isn’t relevant.

Agreed, the reason everything 32 bit Acorn returns Archimedes is so the file servers can decided if the client supports extended handles or not, and I don’t think the modern implementations are even on board with that. I will update !awServer to include an extensible list of what client types do understand extended handles.

“it might be fun to show them in !Machines” I have to admit that was pretty much exactly what I had in mind :-)

Alan, for info:
Single para quotes can be done by starting the line with bq. as in the formatting help (bottom left) and the whole list of possible formatting is covered in the web pages linked to by the “Textile reference here” link at the very bottom.