gnu@hoptoad.uucp (John Gilmore) (05/05/87)
[I have added some American comments of my own at the end. -- hoptoad!gnu]
FidoNews 4-17 Page 11 4 May 1987
Steve Townsley
Opus / SEAdog 510/17
CCITT V21,V23,V22,V22bis
A Word About Standards
This is the first time I have written and article for FidoNews
although in England I will write around 3000 words every couple
of weeks or so. The main problem is not the material, it is
rather the problem of incompatibilty which prevents us in
Europe getting our message to you in the States.
For many of you reading this sending an article to FidoNews is
a simple affair, file attach to 1/1 at 2400 using your USR
modem or "real" Hayes. The FCC cares very little as to whether
your modem is a Taiwanise Hayes clone, runs CCITT tones or BELL
103/212a.
Over in the UK things are very different. Firstly, in order not
to receive a visit from the authorities I have to use a modem
which has gone through a series of approval tests. Secondly,
few of these modems use BELL tones. Thirdly, all must adhere to
the recommedations of the CCITT. Fourthly, approval is a long,
complicated and expensive process for all modem manufacturers.
Converting into dollars, the cheapest useable, approved modem
which could be described as Hayes is the WS4000. It costs
around $300 and will auto-answer at V21 or V23. It is a
variation of Hayes 1200 as it will auto-answer at 300 bps or
1200/75, but mail can only be done at 300 bps (V21).
The V23 standard is a popular in Europe because of Viewdata
services, where it is a standard speed, and for downloading
over crummy phone lines, of which there are many. A V23 modem
can be picked-up for $10 by anyone. So for the user of systems
in Europe a V23 modem is the cheapest way into comms. The 75bps
channel allows the user to type in messages at typing speed and
the 1200 bps channel allows cheap downloading.
The crunch for UK Sysops is that providing access to V23
callers means eiher spending $600 on a modem with V22 as well (to
talk to the States), or buying a $300 Hayes and sending mail at
300 bps. Hayes modems in the Uk which use V21/23 are 1200 bps for
our users but only 300 bps for Sysops.
Obviously we should buy modems that use V21,V23 and V22.
However now you are lokking at prices of around $600 dollars. A
real "Hayes" 1200/1200 modem costs around $750. Moreover, the
real "Hayes" does not use V21 or V23, and combined modems use
the CCITT recommendations on V22. Yes, the tones are compatible
with Bell 212a but, V22 modems will often wait for a V25 answer
tone before sending any data. So the UK sysop, even if he buys
a V22 modem still may not be able to send data to the US
because a BELL 212a modem does not send an answer tone
compatible with the CCITT V25 recommendation.
Perhaps we should buy a V22bis modem. Well prices in Europe
start at around $1100 dollars although the Dowty Quattro with
such things as BELL 212a compatiblity comes in at just $1300.
Remember we still have to offer V23 to our callers whose modems
change hands for around $10.
However, let me suppose that we in Europe suddenly solved all
the problems of the modems, our next problem would be getting
the modems to send CONNECT 1200 everytime a CONNECT 1275
happened because both Fido and Opus don't understand about V23.
SEAdog (Version 4.0) solves the problem by accepting 1275 as a
proper connect message.
Let me assume that the IFNA Committees insist that all Fido
compatible software must accept CONNECT 1275 as a valid 1200 bps
message, not unreasonable since it is merely an ASCII string.
The problems of Europe would still not be totally over as
currently we are fast loosing our position in the nodelist. The
1200 node limit of Fido is most accutely a problem in Europe
and Austrailia. If you cannot mail someone without an address
and you cannot keep the addresses of all nodes in the system we
will add just one one more problem into the cumulative problems
of international links that we had from day one.
As we run Opus, the other week, in an effort to find out more
about aspects of these problems, I logged onto the Dallas Opus
Help BB run by David Finster. In a sense I was pleased to read
the questions YOU, the US Sysops, are now asking about 9600 bps
modems. For, in your own way, you are now experiencing first hand
the frustration of non-standards that have plagued Europe from
day one of running Fido.
European BBS's, and other suppliers of data services, are
governed by the international standards of the CCITT. Non of us
really like the idea that we cannot use BELL tones, or cannot
just plug a USR 2400 bps straight online legally. We don't like
the idea that we cannot participate fully in net activity until
we can communicate at a common standard.
I would now argue that with the growth of nets outside the US
and the large number of systems that need to use the standards
imposed by the CCITT that the net should have a policy on
standards.
Up until now it has been the cry of Europe, unable to afford
the high price of approved CCITT equipment, that has wanted an
agreed standard. I would argue that "standards" is now a
net-wide issue. If US Sysops are to go 9600 bps, and they have
an influence on how modems are to be designed, they should insist
that modems are capable of communicating at CCITT recommended
data rates as well as BELL or one manufacturer's own new
standard.
Many US Sysops must have noticed that more and more European
BBS's are now offering 2400 bps. Little by little European
manufacturers are offering some compatability with US BELL tones.
It has taken the best part of two years to get UK manufacturers
to adopt the Hayes standard. Only some offer BELL tones. There
should be no reason why a modem cannot listen to a phone line,
determine whether an incoming call corresponds to BELL or CCITT
standards and answer with the correct tone. We in Europe are
constantly campaigning for such a modem. Whilst debating the use
of 9600 bps, all of you in the US should demand a modem that
talks to the world and not just Joe Public in the next town!
The moral of this tale... Europe has suffered because the CCITT
standards are not universal. In the next speed jump to 9600 we
must adopt a standard that can be approved for connection to
the phone line in any country in the world. This may mean that
that whatever modem is choosen it must use CCITT tones or yet
again we could face years of incompatibility.
--------
Comments by John Gilmore (hoptoad!gnu on Usenet):
I think the best solution involves the Europeans fixing their
phone/government bureacracies so that they can use whatever modems they
want. We in the US had the same problem through the early 1970s and we
fixed it. (I must say I had no part in the fixing of it -- I was a bit
young.) We've been plugging random modems into the phone wires for
years now and the phone system has not melted down. Europe's turn.
I suspect the first step is widespread disregard for the rules, e.g.
just get a modem and plug it in. They won't be able to hire enough
people to come by and hassle you over it if everyone does it.
If a modem user can buy and attach whatever modem they choose, they
won't be stuck with CCITT standards which don't follow the market, or
with expensive modems you can only get from The Phone Company.
Nobody can predict where the market will go or which modems will end
up cheap and widespread.
On the technical end, we can certainly press for modems which implement
a variety of standards, but in many cases the standards are incompatible.
For example, you can't determine whether an INCOMING call corresponds to
Bell or CCITT tones, because the ANSWERING modem is the first to produce
tones. I think this is backward, but it has been backward for years.
Kludges like first answering with one tone, then switching to another,
work after a fashion, but not if you have a list of 10 or 15 different
protocols to test for -- the caller will hang up before you get to the
right protocol!
Certainly there is economic incentive for manufacturers to support as
many standards as they can, but there is also incentive to keep their
modems cheap. More standards mean more design time, more components,
more testing, more upgrades in the field due to more bugs -- in short,
higher prices. YOU can support the companies that handle many standards
by buying a high priced modem. Well? What are you waiting for?
Another option is for the European sysop to simply buy several modems --
a $10 modem for the Teletext callers and a $300 modem for the 1200/1200
callers. This requires two phone lines, or fancy juggling, but does
not seem as bad as buying a $600 modem that does both. Oh, your software
doesn't support two modems? Well, you shouldn't have bought an MSDOS
machine anyway. You get what you pay for.
I think it's also time for the death knell of software with wired-in
modem control strings and responses. Unix uucp, tip, and getty are
notorious for this, as are many other programs inherited from the days
before inline serial dialers and multi-speed modems. A simple interpreted
language would suffice for sending strings to the modem, checking the
results, and providing for different alternatives depending on the
response or lack thereof. Having to get a new release of a program to
cope with "CONNECT 1275" rather than "CONNECT 1200" (as FIDO sysops do,
or as I've had to do to use USR 2400's on my Sun Unix 3.0 system) is
ridiculous given the overall somewhat advanced state of software.
Let us not let Unix System V's removal of split-speed (e.g. 1200/75)
support not go unremarked either!
--
Copyright 1987 John Gilmore; you may redistribute only if your recipients may.
(This is an effort to bend Stargate to work with Usenet, not against it.)
{sun,ptsfa,lll-crg,ihnp4,ucbvax}!hoptoad!gnu gnu@ingres.berkeley.edu