On 11/17/19 4:32 PM, alan--- via cctalk wrote: sounds painful do you mean over a twinax port? I was just thinking I don't think I've seen the equivalent of the CUT spec for twinax There was firmware and cards that existed for the National DP8344 that supported CUT and twinax but I've never been able to find a copy.

On 11/18/19 2:56 AM, Dave Wade via cctalk wrote: Are they working on Cisco's SDLC line protocol and / or physical (serial) interface? So something like this: [3270]---(coax)---[3174]===(SDLC)===??? ... ???---[Hercules] What is the physical / logical connection to Hercules and the machine it's running on? Or is this more like this: [3270]---(coax)---[3174]===(SDLC)===[Cisco]---Ethernet---[Hercules] In that the Cisco translates from the physical SDLC circuit to Ethernet and maybe does a protocol conversion to send things into Hercules across standard Ethernet? Meaning that the Cisco becomes a fancy remote SDLC interface that Hercules communicates with? I've been pontificating DLSw coming out of OS/390 running in Hercules / on my P/390-E to a Cisco and leveraging DLSw support to connect SNA (802.2 SNAP) to clients downstream. This seems similar but is integrating an SDLC link to a 3174. Hum. I thought the TN5250 (and TN3270) was the 5250 (3270) protocol over TelNet. As such, I don't think a 5250 (3270) terminal can speak the TN5250 (TN3720) protocol at all. Instead, they need some sort of physical device that they can connect to that will gateway from <whatever> physical w/ 5250 (3270) protocol to any IP network with TN5250 (TN3270). So, do any Cisco devices have twinax (?) interfaces for 5250 (or coax for 3270)? I know that older & larger Cisco gear has the Channel Interface Processor that can attach to Bus & Tag channels from a mainframe and convert from TN3270 (I presume TN5250) to the physical B&T interface and necessary protocols there over for the mainframe. But that's decidedly going the opposite direction. Perhaps ~> hopefully I'm just ignorant of other solutions from Cisco, or anyone else, that will do what is desired. I believe the *Remote* 3174 (et al.) is one of these devices from IBM. IBM's "Enterprise Extender" comes to mind too. But I don't know what sort of down stream equipment that connects to. -- Grant. . . . unix || die

On 11/18/19 9:42 AM, Dave Wade via cctalk wrote: ACK I don't think DLSw being SNA only is a problem. I say this because I think DLSw requires TCP/IP connectivity. Thus you already have TCP/IP connectivity to the mainframe. ACK [3270]---(coax)---[3174]===(SDLC)===[Cisco]---(Ethernet)---[Hercules] [3270]---(3720)---[3174]===(????)===[Cisco]---(DLSw/IP)----[Hercules] Is it proper to say that SNA is the protocol that is carried across the DLSw over IP? Is it also what's carried over the SDLC? Okay.... I'm lost. If there's no Cisco box with Twinax and the host doesn't have TCP/IP, how are you going to connect things? SNA across Token Ring and / or Ethernet? #confused pr0n :-) -- Grant. . . . unix || die

On 11/18/19 10:12 AM, Dave Wade wrote: Okay. Do you know what the proper, or any, name is for said channel attached protocol? Ah. I hadn't contemplated a system with an SDLC interface but no Twinax / Token Ring / Ethernet. Given your previous comment, this makes more sense. Thank you for enlightening me. ~chuckle~ "Yes dear" usually works well for me. YMMV -- Grant. . . . unix || die

I started with a token ring card in a Windows/2000 box so I could connect directly. I have also gone via a Cisco router with a Token Ring interface NATed onto my normal LAN. I now have a 3174 with an Ethernet card but only floppies. I was planning to try moving the card to a bigger 3174 with a Hard Disk (actually an MFM emulator) but the PSU in that has dies.. ... just as I got the MFM emulator working :-( The documents presented seem to have lots of info. Perhaps I should get the scope out and see what happens on the co-ax... Dave

TN3270 into Hercules so they look local to the OS... TH3270. Cisco merely converts Token Ring to ethernet as my current PC only has PCI-Express and you can't get PCI-Express Token ring cards. 3270 to Hercules.. Dave

On 11/18/19 9:45 AM, Al Kossow via cctalk wrote: Please confirm if it is DLSw between the Cisco and OS running in Hercules? Sorry for being obtuse. There are a LOT of different things in IBM networking, compounded by other vendors, and I want to make sure I'm tracking properly. Particularly as this is something I'd like to reproduce for similar reasons. Would you please clarify if you have the "Local" or "Remote" 3174? Can "Local" and / or "Remote" 3174s be converted to the other end of the link? I ask because I think most "Local" 3174s expect the host to be channel attached. Don't they? -- Grant. . . . unix || die

Grant, One thing which complicates things is that we have (roughly) layered protocols in place between the 3174 and the host. So at the physical level we have channel attached, SDLC, Token Ring and bi-sync. Above that we have the channel protocols, HDLC/SDLC and bi-sync polling. Then we may have SNA (which encompasses many options) to transport the data through the network. At the top the protocols and frame formats in the Data Streams manual which are broadly the same but with tweaks for SNA, Local and Bi-Sync. Dave

On 11/18/19 2:22 PM, Chuck Guzis via cctalk wrote: "?HASP?" MVS / OS/390 / z/OS term. "?on our 11-750?" that sounds like a VAX term (11/750). Brain: Wait a minute. These are different ecosystems. Yes, they are supposed to be able to communicate, but is that what we're really talking about here? "?Berkeley Systems Group?" that's decidedly BSD, like on a VAX (11/750). "?leased line?" ecosystem agnostic, but common in IBM. "?Bell 209 modems?" fits with my understanding of leased lines. Brain: Yep, it looks like we're crossing ecosystems. Chuck, I think you are the first person that I've heard / read about with direct exposure to HASP on non-IBM platforms. I've heard / read that it's supposed to be possible. But I've not had first hand accounts before. Now I have. Thank you. -- Grant. . . . unix || die

On 11/18/19 1:36 PM, Grant Taylor via cctalk wrote: You've got that right--we leased our VAX on the condition from the vendor that it could run HASP. The BSD folks had a package from a third party that claimed to do that. We never figured out that it could. The leased line was a 9600 bps sync from St. Paul to Santa Clara. IIRC, it was about $5K/month. IBM hardware on the far end, feeding an ETA 10 system. How's that for a bizarre configuration? Would have been nice had it worked. By 1990, 9600 bps over voice-grade lines would be abysmally slow. --Chuck

On 11/18/19 2:12 PM, Ethan Dicks via cctalk wrote: does that code still exist? I have two stand-alone boxes IBM 3708 and Interlynx/3278 that essentially do that, SDLC -> multi-port VT100

On 11/18/19 3:15 PM, Chris Zach via cctalk wrote: yes, it would be good to get this archived on bitsavers should talk to ethan about his stuff as well

On Mon, Nov 18, 2019 at 7:42 PM Al Kossow via cctalk <cctalk at classiccmp.org> wrote: The golden years of Software Results and COMBOARDs were 1982-1986, perhaps a little later, but not much. By early 1989, we really didn't have many board sales, just maintenance contracts for the existing customer base and an occasional upgrade sale (Unibus->VAXBI or Unibus-Qbus). By 1994, there weren't that many customers left on maintenance. The last contract expired around February 1995. We did have a last breath of demand in the early 90s that garnered a tiny handful of sales - when EDI began to take hold, one of the standard transport models was 3780 to an IBM service that essentially took care of delivery in the fashion of an ISP. If you wanted to use that network, you needed _a_ product that would move files using sync modems and the 3780 protocol. There were a couple of PC products that could do it, and we were one of the last companies still in the Bisync space for minicomputers. After EDI moved to TCP/IP, that was all over. But in the early 80s, we made a few million dollars getting PDP-11s and VAXen to interoperate with IBM mainframes. At that time, lots of large companies needed it, then fairly quickly nobody needed it. -ethan

I find it interesting that one of the few remaining mainframe third parties is called Micro Focus Will

I don't think its really "the same" anymore since it took on much of HPEs software. But essentially the great grandchild of the business which I believe started Don Higgins 370 emulator. https://www.mail-archive.com/ibm-main at listserv.ua.edu/msg52038.html I would like be able to use some of the sync comms stuff, but I don't have either the hardware or the software. I have used TN3270 to go from my VAX to my P390 but that?s about it. Neither my P390 or my VAXen have sync cards, nor I think does my CISCO router. I think I could use a 3174 as a gateway but that?s a whole new can of worms... Dave

On 11/19/19 1:36 AM, Dave Wade via cctalk wrote: Couple of their folks used to belong to the same bicycling club that I was a member of. Sometime in the 1980s, I believe. One of them gave me a copy of their COBOL. --Chuck

On 11/18/19 7:34 PM, Frank McConnell wrote: yes, both, 'Rumba' which supported several different boards/drivers they bought the Apple Nubus coax/twinax board product line and snaPS software product line as well (I've been trying to find the hardware, I found the software on eBay)

On 2019-11-18 19:42, Al Kossow via cctalk wrote: In attempting to bridge the twinax world to something else, I guess I was naive coming from the ASCII/UNIX realm thinking 'how hard could it be?' The electrical-wire and line framing and coding of the TwinAx data cable is pretty straight forward. 1 MSps differential Manchester coding with a fixed frame sync pattern. Then up to 256 bytes of payload per frame to 7 logical addresses. I was under a simplified-thinking impression the 5250'ish terminal and printer data protocols would live directly on top of that. Now it seems from my research, it's SNA/LU7 that sits on-top of TwinAx for early S/3x, correct? Then a 5250 data stream? Then 5250 display station specific commands? Are there any ideal resources I should be looking at? I've browsed through the GC30-3073-1 SNA Technical Overview but it doesn't cover differences in LU types, for example... No wonder Big Blue was always so expensive. -Alan

On 11/18/2019 6:42 PM, Al Kossow via cctech wrote: At first other TCP implementations were superior to IBMs. We did a bid back in the day (don't remember the exact year). Jupiter (a division of Intel by then) won it with their channel attached multi-bus based implementation. It had tn3270 access from our Unix workstations and VAXen. Before that (in 1986) we also had an Interlink 3711 channel attached DECnet interface that ran on RSX-11 so that our VAXen could communicate with our mainframes. I don't recall it doing 3270 access from the VAXen, but I think we could log on to the VAXen from our 3270 in line-by-line mode. JRJ

On Mon, Nov 18, 2019 at 5:50 PM Al Kossow via cctalk <cctalk at classiccmp.org> wrote: I have all the code. Among other places it's in VMS saveset files on one of my Linux laptops I use for project hacking. We licensed an existing product and rewrote it extensively for part of the application code to run on a COMBOARD (MC68000, 128K/512K local RAM, COM5025 USART, DMA interface to Unibus/Qbus) and the other part on the VAX, with a lot of shared memory blocks to split the work across both CPUs (the DMA engine mapped the entire Qbus/Unibus space into 68000 space so on the COMBOARD you could treat VAX memory like local memory but it was just slower). The actual technique could be illuminating to examine, but to run the code in its present form would require our custom front-end board. I do remember that it was a multi-person-multi-year effort to get that complicated wad of code working split across the bus. 99% of my coding work there was with the Bisync products so I am not the foremost authority on our SNA product. I had to handle support calls but I never fixed bugs on it. -ethan

Yes that?s in there.... BiSync polling gets in the way at the top.... Sadly no. If we are talking VTAM rather than TCAM then the layers are somewhat ill defined. From http://vm.marist.edu/~vmshare/browse.cgi?fn=HUMOR91&ft=MEMO&args=sn… April Foolishness ============== We present here fanciful ideas and interpretations related to VM, IBM, and computing in general. o Networking breakthrough: for those days when your cooperative applications won't, IBM introduces APPS (Advanced Program-to-Program Squabbling). This brings the confusion and disruption of incompatible computers and unspecified interfaces to the farthest reaches of an enterprise's network. Append on 03/30/88 at 17:40 by Jean Rall Grumman Data Systems 516-575-7140: Thought I'd pass this on; Conversation between a caller and John Akers: CALLER: Hello, John? AKERS: Yes? CALLER: John Akers? AKERS: Yes? CALLER: John F. Akers? AKERS: Yeah? CALLER: John F. Akers, CEO for IBM? AKERS: Yes, what is it? CALLER: John F. Akers, CEO for IBM in Armonk? AKERS: Yes, that's right, what do you want? CALLER: John F. Akers, CEO for IBM in Armonk, ...? (etc. for a few more exchanges) AKERS: All right, enough! What is going on here?! CALLER: Oh, not much John. Just thought you would like to know what it is like to talk over SNA. *** APPENDED 03/30/88 17:40:44 BY GDS/JEAN *** And from http://vm.marist.edu/~vmshare/browse.cgi?fn=HUMOR88&ft=MEMO&args=sn… Append on 03/30/88 at 17:40 by Jean Rall Grumman Data Systems 516-575-7140: Thought I'd pass this on; Conversation between a caller and John Akers: CALLER: Hello, John? AKERS: Yes? CALLER: John Akers? AKERS: Yes? CALLER: John F. Akers? AKERS: Yeah? CALLER: John F. Akers, CEO for IBM? AKERS: Yes, what is it? CALLER: John F. Akers, CEO for IBM in Armonk? AKERS: Yes, that's right, what do you want? CALLER: John F. Akers, CEO for IBM in Armonk, ...? (etc. for a few more exchanges) AKERS: All right, enough! What is going on here?! CALLER: Oh, not much John. Just thought you would like to know what it is like to talk over SNA. *** APPENDED 03/30/88 17:40:44 BY GDS/JEAN *** Seriously the SNA Bind is an animal to reckoned with. I haven't included the SNA manuals, but if you are suffering from insomnia you may wish to browse... http://bitsavers.org/pdf/ibm/sna/GC20-1869-2_SNA_Introduction_to_Sessions_B… Well sort of. APPN/APPC is deeply integrated in SNA. See above... Dave

My interest is I have a S/36 with both TwinAx and SDLC. SDLC is a lot easier to interface to. But I have built a electrical interface board to go from a balun'd RJ11 twin-ax interface to an FPGA with the intent on eventually building an open-source TwinAx <--> USB converter. Sounds like there's a lot of interest. If anyone is wanting to help or test, I'll move this up the priority list. -Alan On 2019-11-18 12:31, Al Kossow via cctalk wrote:

On 11/18/19 10:38 AM, Dave Wade wrote: I forgot about the fact that Hercules, not the mainframe OS, is doing some translation. Do you know what protocol name would be used to describe the communications between the mainframe OS and what it thinks are locally attached terminals? ACK Hum. ... Very intriguing. Am I understanding you correctly to mean that Hercules itself has some DLSw functionality? Part of me wants to ask if the 3174 support can be used for SNA out of the mainframe, as in the below diagram. But my reading of various pieces of Hercules documentation indicate that it can't be done (yet). [Mainframe OS]---(SDLC)---[Hercules 3174]---(DLSw)---[Cisco]---(SNA)---[client] Thank you for understanding. ACK "most", in this context, is more to allow for my loose understanding and zero exposure. I'm not tracking "when". Do you have the model numbers handy? (I assume the all the type numbers are 3174.) That sounds like Remote 3174s. At least based on my limited and likely flawed understanding. My understanding is that: ? Local 3174s were channel (?) attached to the host. ? Remote 3174s were SDLC / X.25 / Token Ring / Ethernet connected to Local 3174s. Where "local" is physically close to and connected to the host. Similarly "remote" is not physically close nor connected to the host. -- Grant. . . . unix || die

On 11/18/19 11:02 AM, Al Kossow via cctalk wrote: $ReadingList++ Thank you. $ReadingList++ Hum. I don't know much about CICS. I had naively thought that it was also about transactions. But I definitely see the benefit of both. LOL Joy. -- Grant. . . . unix || die

Yes, I have my 3174 communicating with Hercules (as well as my MP3000). TTFN - Guy

On 2019-11-18 11:54, Al Kossow via cctalk wrote: Yes, that hack did occur to me as well. But finding suitable working 3708s or equivalents has been challenging. -Alan

On 11/17/19 10:05 PM, jim stephens via cctalk wrote: tried sending email there, it didn't bounce but I didn't hear anything back

On 11/18/2019 12:05 AM, jim stephens via cctalk wrote: Though I have worked with real 3270's in the past, I am no expert on the CUT or DFT protocols. However, it would not surprise me at all if, for a 3278, that information is provided by the controller -- but not as part of the data stream that the host computer sees. I found these documents, but they don't address that status line, because they are for the older 3271/3272 controllers. http://www.bitsavers.org/pdf/ibm/3270/3271-3272_to_Device_Interface_Spec_Ma… http://www.bitsavers.org/pdf/ibm/3270/fe/ANR-specification-1975.pdf JRJ