gnu@hoptoad.uucp (John Gilmore) (08/19/87)
ron@topaz.rutgers.edu (Ron Natalie) wrote: > I doubt there will be an RS-232D, they moved on to the RS-4XX standards > for serial lines. And the RS-4xx standards (e.g. RS-422, RS-423, RS-449) are even more broken than the RS-232C standard. Would you believe a 39-pin plug with an additional optional 15-pin plug? This is not a joke. The reason we have no standards better than RS-232 is because the people who tried to produce one screwed it up royally, so nobody followed it, and each manufacturer who didn't do RS-232 (IBM, Apple, etc) invented something random that was convenient for them. -- {dasys1,ncoast,well,sun,ihnp4}!hoptoad!gnu gnu@postgres.berkeley.edu My name's in the header where it belongs.
phil@amdcad.AMD.COM (Phil Ngai) (08/19/87)
Gee, all this plaintive wishing for hardware flow control at a mere
9600 bps. Yet the people who run fast use V.35 or T1 interfaces which
don't have hardware flow control. Wonder how they do it?
(oh, I left out Ethernet. No RTS/CTS on that media either.)
--
I speak for myself, not the company.
Phil Ngai, {ucbvax,decwrl,allegra}!amdcad!phil or amdcad!phil@decwrl.dec.comhoward@COS.COM (Howard C. Berkowitz) (08/26/87)
In the discussion of the inadequacies of RS-232C and its followons, there seem to be some misunderstandings both of the design goals of these standards, and what the standards actually say. In article <2790@hoptoad.uucp>, gnu@hoptoad.uucp (John Gilmore) writes: > ron@topaz.rutgers.edu (Ron Natalie) wrote: > > I doubt there will be an RS-232D, they moved on to the RS-4XX standards > > for serial lines. > And the RS-4xx standards (e.g. RS-422, RS-423, RS-449) are even more broken > than the RS-232C standard. Would you believe a 39-pin plug with an > additional optional 15-pin plug? This is not a joke. I'm afraid there is a joke here, because RS-449 (which I argued against in standards groups when it was proposed) does not use the specific connectors John cites, and RS-422/423 are ELECTRICAL standards only, which do not prescribe connectors. I don't mind when a standard is trashed for what it is, but I do have some objections to attacks based on errors in quoting the standard, implementations which violate the standard, or applications of a standard for things it was not intended to do! Deficiencies in RS-232 had been recognized for some time, and replacements were planned. It is a reality that standards take a LONG time to develop, and technology sometimes jumps over the standards process. This explains the limited popularity of RS-449 (but not 422/423). The intention to replace RS-232 involved a family of four standards: MECHANICAL/LOGICAL ELECTRICAL ------------------ ---------- EIA RS-449 (37-pin plus EIA RS-423 (unbalanced optional 9-pin) electrical passively compatible with RS-232) CCITT X.21 (15 pin) EIA RS-422 (balanced digital, high speed) One could mix and match -- the more expensive RS-422 has up to 10MB capacity, yet could be used on either connector. The philosophical difference between RS-449 and the CCITT X.21 physical interface (there are other interfaces defined in X.21) is the intelligence expected in the higher-layer protocol. RS-449 assumed a dumb upper layer, with separate pins required for a variety of diagnostic and control functions. X.21 assumed these could be part of the data link and higher layers. Flow control is a basic feature of intelligent link- and higher-layer protocols, so extended flow control pins are strictly not necessary in the physical interface when HDLC or equivalent protocols are used. > > The reason we have no standards better than RS-232 is because the > people who tried to produce one screwed it up royally, so nobody > followed it, and each manufacturer who didn't do RS-232 (IBM, Apple, > etc) invented something random that was convenient for them. There clearly needs to be a faster way to develop standards. RS-449 does meet a certain family of applications, but most of those applications are obsolescent. X.21 physical works well for many more applications. The bottom line (from the standards perspective) is that flow control should not be implemented in the physical interface, but at data link or a higher layer. Yes, much development has used the alternative approach, but it can be economic to use higher-level flow control if economies of scale permit large-scale chip support for it. Those who condemn proprietary implementations might ask themselves how often they accept them anyway, and provide no market incentive for standard solutions to evolve and improve. A personal opinion: standards are necessary for interoperability of multivendor products, and they will have bugs in their early implementations if those implementations get done in any reasonable time. Independent testing of implementations can help clear these problems, but the bottom line is that users must demand standard solutions if manufacturers routinely are to implment them. -- -- howard(Howard C. Berkowitz) @cos.com {seismo!sundc, hadron, hqda-ai}!cos!howard (703) 883-2812 [ofc] (703) 998-5017 [home] DISCLAIMER: I explicitly identify COS official positions.
jhh@ihlpl.ATT.COM (Haller) (08/26/87)
In article <2790@hoptoad.uucp>, gnu@hoptoad.uucp (John Gilmore) writes: > ron@topaz.rutgers.edu (Ron Natalie) wrote: > > I doubt there will be an RS-232D, they moved on to the RS-4XX standards > > for serial lines. > And the RS-4xx standards (e.g. RS-422, RS-423, RS-449) are even more broken > than the RS-232C standard. Would you believe a 39-pin plug with an > additional optional 15-pin plug? This is not a joke. Actually RS-232D was approved by EIA either late last year, or early this year. It's significant new information was codifying subsets of RS-232C that are in common use. It has been generally realized that assigning pins to functions ala RS-4xx was the wrong to go about things, since microprocessors are so cheap. Multiplexing encoded signalling information is likely to be the next type of connector, if any new ones are produced. You may note that the ISDN connector is just a modular jack, not a 157 pin connector.