El 04/08/2012, a les 12:38, Sampsa Laine <sampsa at mac.com> va escriure: For various reasons to do with my general mental stability etc, I decided to learn about IBM mainframes, so got one running.. I used to be a MVS systems programmer :) Unfortunately, my specialisation was not networking/communications, but perhaps I can help a little bit... Any way to connect this to DECnet? Maybe over over DECNet-OSI since the thing natively does VTAM? Are you using the "public domain" MVS3.8J available in the net? That's quite old. No TCP/IP support. Just plain old SNA/VTAM. At those days you needed special hardware to talk to IBM machines. They just didn't speak ethernet. I don't think even if you had a working 3745 you could attach it to an Hercules emulator. So I'm afraid there is no joy. You _could_ be able to use part of DECNET/SNA to do open a 3270 session from your VAX, if it can be directed to use a telnet port. Hercules exposes its virtual 3270 controllers thru telnet, so that could be a possibility. Of course, there are... hrrr... alternatives to MVS3.8J. But they are not legal to run in Hercules. Technically, you can fire up a modern z/OS system, with full TCP/IP support on Hercules. But don't tell IBM about that ;) Details: http://sampsa.com/2012/08/04/mvs-on-an-s390-because-openvms-is-just-too-eas… That distribution comes without the butter and bread of MVS. You get just native TSO (yuck!) and a replacement for the "real thing" (that is, ISPF/PDF) called RPF. If you go thru the books you'll be lost, since they talk about the IBM tools which come with ISPF. Specifically, the editor. RPF comes close to it, but it is different enough so the TSO/ISPF books won't help you. The best way to edit/manage files in that 3.8J emulated system would be using the card reader / card punch emulation to put in and get out files. As for "datasets", here comes a crash course: - MVS does not have a "proper" filesystem. Each disk ("volume", in IBM-esse) has its own table of contents (VTOC), which contains the files resident at that volume. The files in a disk or tape are called "data sets". When you open a dataset from a program, then it is a "file". The "files" and "data sets" are related using JCL thru "DDNAMES" or "Data Set Names". Think about them as some kind of logical names. - Each "dataset" has a name which can be up to 43 characters long. The names are formed attaching parts of up to 8 characters long, separated by dots. The characters can be any alphanumeric byte (excepting for the first one in each 8-byte part, which has to be an alphabetic character). All of this in EBCDIC, of course. Exammples: SYS1.PROCLIB MYFILES.SOURCES.COBOL MYFILES.LOAD THIS.IS.AN.ABSURDEL.Y.LONG.NAME And so on... - There is an indexed file (actually, a set of logically linked VSAM KSDS files) which _could_ contain all the datasets known to the system, correlated to the volumes where they reside (including tapes), called CATALOG. In more modern versions of MVS the system programmer (system manager) can enforce that every permanent file MUST be referenced in that catalog. - There is a special kind of data set, named "PDS", or "Partitioned Data Set". That dataset can have "members" inside, and has a directory to list its members. The member name goes after the dataset name inside parenthesis: MYFILES.SOURCES.ASM(MYPROG) This is the way source and load (executable files) are usually managed in the mainframe environment. You use the ISPF utilities to allocate and create the library (as PDS are also known) and then use the editor to create/modify the members. The ISPF panels (screens) to edit, compile, linkedit and run programs are prepared to manage libraries and allow you to browse the library directory using the usual wildcards. Of course, you probably already know all this stuff :) Anyway, if you need help, email me and I'll try to assist you. Jordi Guillaumes i Pons jg at jordi.guillaumes.name HECnet: BITXOV::JGUILLAUMES PS: I have lost connectivity to UPDATE again. And this time rebooting my router didn't fix it... Any other one in the same situation?