Re: COBOL on HPUX
On Mon, Jan 6, 2020 at 10:45 PM R. Anthony Lomartire < opensourcek...@gmail.com> wrote: > So I recently landed a job working in COBOL on HP-UX. It's been a trip! > HPUX is "interesting". HP and IBM both made IT-friendly variants of Unix (previously it was just an engineering OS; named "HPUX" and "AIX" respectively) long before POSIX standardized the needed richer security/permissions features (e.g. ACLs), and of course the other brands refused to bless either HPUX or AIX's variations. So life is odd on either of them. I survived HPUX, and liked AIX, when I had projects on them. This stuff is from before my time but it's been really interesting to > learn. Have any of you folks worked with this stuff? > Well yes. I coached a couple of girlfriends through the COBOL assignment in their Survey of Languages courses in '79-'80, using an already obsolete IBM 1401 user manual, without having taken the course or studied COBOL more than casual reading. (They both passed, and I'm still married to one of them!) Then in 1981, i was paid to "write" (tweaked copy-pasta reuse) 2 lines of COBOL on the TOPS-10 PDP-10 at DOT VOLPE center, to add field 13 A to the processing for a form, after adding 13 A between 13 and 14 in the Screen Painter and the DBMS schema. (And dump and reload the data of course.) (We were the Fortran department, but our previous DB guru was bilingual and noticed that the Cobol/DP dept had gotten a Form-painter application that worked on glass terminals in block mode, to get away from literally keypunching data on cards, and it only supported COBOL -- would generate a DATASECT and an object to link to; and our DBMS System 1022 also generated a DATASECT for COBOL (and did similar for Fortran), so it was a small matter of (pseudocode cobol) CALL INPUT_FORM_ROUTINE IF <*validate input buffer field*> COPY input7 TO output7 ELSE SET ERROR_SEEN TO 1 *... lather rinse repeat 1 to 17 ... and then insert 13A between 13 and 14 after it's been in "production" for months.* CALL WRITE_OUTPUT_TO_DBMS The DEC PDP-10 had 36 bit words, so DEC TOPS COBOL had 6 x 6-bit ASCII UPPER CASE CHARACTERS PER WORD. (WHO NEEDS LOWER CASE?) DEC TOPS Fortran by contrast had discovered mixed case and had 5 x 7-bit ASCII per word. (But the bit left over was mantissa lsb, not sign, so was pretty much useless as a out of band marker.) We're looking to migrate away eventually, maybe anyone with experience > there? I'd love to hear any stories about COBOL or old enterprise mainframe > applications you've worked with. We're probably going to be hiring soon too > if anyone would be interested in a similar gig. :) > Early in the new century, my old financials shop was looking to replace two overlapping business critical applications, one Mainframe COBOL and one VMS COBOL, with something new. (We'd already replaced the PL/1 application running on Stratus.) Eventually* instead of paying a vendor to upgrade their Unix/Linux C++ app with Java UI to handle the needed features, the vendor for the IBM M/F COBOL app added the features needed to retire the VMS app. (I didn't directly touch the Mainframe, but dealt with the problems of transferring LRECL EBCDIC files to CRLF ASCII Unix/Linux hosts and vice versa, as well as App/OS/HW interface/capacity issues on Unix/Linux platforms. Much hilarity with file transfers.) *Eventually = I think they finally finished?? -- Bill Ricker bill.n1...@gmail.com bric...@theperlshop.com https://www.linkedin.com/in/n1vux ___ gnhlug-discuss mailing list gnhlug-discuss@mail.gnhlug.org http://mail.gnhlug.org/mailman/listinfo/gnhlug-discuss/
Re: COBOL on HPUX
On 2020-01-06 22:44, R. Anthony Lomartire wrote: So I recently landed a job working in COBOL on HP-UX. It's been a trip! Oh, man. You just had to go there. Why, yes, as a matter of fact, I *do* have a COBOL on HP-UX story. I was working at a startup c. 2002, and we wanted to use the PointMan ERP system on our HP-UX hosts. (Linux wasn't yet an option for PointMan.) So I, a relative HP-UX neophyte, ordered COBOL for some thousands of dollars. I got, in a FedEx envelope: one (1) sheet of paper with one (1) serial number and a phone number to order more stuff. Period. I mean, silly me. For a couple of thousand bucks, I'd expected install media, release notes, some accompanying documentation. *SOMETHING* So I call the phone number and am like, "What in the world do I *do* with this??" They transfer me to another number. Which transfers me to another number. Which transfers me to another number. Who gives me a number they promise will be able to help. It's only after I hang up that I realize it's the first number I'd called -- the one on the piece of paper. At this point, I begin to doubt my sanity. Oh -- and did I mention the ERP system, itself, cost something north of $150K, and I had the CFO breathing down my neck to get it installed, like, yesterday? I finally find some poor woman who's at least, like, *heard* of COBOL. And she gets me to people who are willing to help me -- if I pay the $750 (? -- I think that's right) maintenance fee. So I do. And get connected with a very helpful engineer who explains the software is on the install media that *came with the system*; I just needed the serial number to activate it. "Except, oh, yeah, YOUR version of the install media has a bug, and COBOL won't install. I need to mail you a file." "So, you mean, even if I knew HP-UX super-duper well, I *STILL* wouldn't have been able to install it?" "Yeah, that about sums it up." Again: release notes. Errata. An fscking URL. ANYTHING. I wrote our HP rep a letter the likes of which I generally try not to write. He called me up and asked what he could do to make it right. I said that was impossible, but implored him not to screw over other customers. That's a top-five most-frustrating-thing ever. I sincerely hope that things have changed in the intervening time. -Ken This stuff is from before my time but it's been really interesting to learn. Have any of you folks worked with this stuff? We're looking to migrate away eventually, maybe anyone with experience there? I'd love to hear any stories about COBOL or old enterprise mainframe applications you've worked with. We're probably going to be hiring soon too if anyone would be interested in a similar gig. :) ___ gnhlug-discuss mailing list gnhlug-discuss@mail.gnhlug.org http://mail.gnhlug.org/mailman/listinfo/gnhlug-discuss/___ gnhlug-discuss mailing list gnhlug-discuss@mail.gnhlug.org http://mail.gnhlug.org/mailman/listinfo/gnhlug-discuss/
COBOL on HPUX
So I recently landed a job working in COBOL on HP-UX. It's been a trip! This stuff is from before my time but it's been really interesting to learn. Have any of you folks worked with this stuff? We're looking to migrate away eventually, maybe anyone with experience there? I'd love to hear any stories about COBOL or old enterprise mainframe applications you've worked with. We're probably going to be hiring soon too if anyone would be interested in a similar gig. :) ___ gnhlug-discuss mailing list gnhlug-discuss@mail.gnhlug.org http://mail.gnhlug.org/mailman/listinfo/gnhlug-discuss/
Re: Runaway log...
On 2020-01-06 21:43, Joshua Judson Rosen wrote: > On 1/6/20 8:45 PM, Ken D'Ambrosio wrote: > > Buffered in journald, maybe? GNU bless you, good sir. Did the trick -- and a good thing, as it was still happily spamming away. Thanks! -Ken ___ gnhlug-discuss mailing list gnhlug-discuss@mail.gnhlug.org http://mail.gnhlug.org/mailman/listinfo/gnhlug-discuss/
Re: Runaway log...
On 1/6/20 8:45 PM, Ken D'Ambrosio wrote: > OK, guys. CentOS 7.1. I've got an OpenStack process that wigged out > and was logging like crazy to /var/log/messages. So I killed it. FORTY > FIVE MINUTES AGO. And still, log lines that must've been buffered... > somewhere, are flying into the messages file. Gigabytes of them, e.g., > > Jan 6 20:42:56 sca1-drstack01 neutron-server[27127]: Exception > RuntimeError: 'maxiException mum RuntimeErrorr: e'cmuaxrismuim roencu > rdsieonp tdehp the xecxcddede wdhi lew cahlillien gc aal lPiyntgh > oan Poybtjheocnt 'o in bject' > ignored > > Now, 27127 is dead, gone, not in the process table. Not a zombie, not > nothing. I restarted the syslog... and the logging stopped for a few > seconds, and then restarted. How in blazes do I find what's buffering > the logs, and how do I flush it?! Buffered in journald, maybe? ___ gnhlug-discuss mailing list gnhlug-discuss@mail.gnhlug.org http://mail.gnhlug.org/mailman/listinfo/gnhlug-discuss/
Runaway log...
OK, guys. CentOS 7.1. I've got an OpenStack process that wigged out and was logging like crazy to /var/log/messages. So I killed it. FORTY FIVE MINUTES AGO. And still, log lines that must've been buffered... somewhere, are flying into the messages file. Gigabytes of them, e.g., Jan 6 20:42:56 sca1-drstack01 neutron-server[27127]: Exception RuntimeError: 'maxiException mum RuntimeErrorr: e'cmuaxrismuim roencu rdsieonp tdehp the xecxcddede wdhi lew cahlillien gc aal lPiyntgh oan Poybtjheocnt 'o in bject'> ignored Now, 27127 is dead, gone, not in the process table. Not a zombie, not nothing. I restarted the syslog... and the logging stopped for a few seconds, and then restarted. How in blazes do I find what's buffering the logs, and how do I flush it?! I've run into this once before and did *something*, but damned if I can remember what. All ears; my disk space is finite. (I've already truncated the file twice.) Thanks, -Ken ___ gnhlug-discuss mailing list gnhlug-discuss@mail.gnhlug.org http://mail.gnhlug.org/mailman/listinfo/gnhlug-discuss/