[daip] [!13508]: AIPS - CLCOR EOPS produces NANs in CL table using 31DEC18 and 31DEC19 on OS X 10.14

[Eskil Varenius]

Eskil Varenius updated #13508
-----------------------------

CLCOR EOPS produces NANs in CL table using 31DEC18 and 31DEC19 on OS X 10.14
----------------------------------------------------------------------------

           Ticket ID: 13508
                 URL: https://help.nrao.edu/staff/index.php?/Tickets/Ticket/View/13508
                Name: Eskil Varenius
       Email address: varenius at chalmers.se
             Creator: User
          Department: AIPS Data Reduction
       Staff (Owner): -- Unassigned --
                Type: Issue
              Status: Open
            Priority: Default
                 SLA: NRAO E2E
      Template group: Default
             Created: 07 February 2019 03:45 PM
             Updated: 07 February 2019 03:45 PM
           Reply due: 11 February 2019 03:45 PM (4d 0h 0m)
      Resolution due: 03 November 2021 12:00 AM (999d 8h 15m)

Dear support,I recently formatted my Mac and installed the latest OS X 10.14.3. I installed the binary version of AIPS 31DEC19 following instructions. Everything appeared to work fine, until I tried running CLCOR on archival VLBA data (BP129 Cband) with these arguments
OPCODE = "EOPS"CLCORPRM = 1, 0INFILE = PWD:usno_finals.erpCLINVER = 2CLOUTVER = 3
The task gives no error message, the output ends with
CLCOR2: ZTXOP2: using translated file name =CLCOR2: ZTXOP2: /Users/eskil/TEST/usno_finals.erpCLCOR2: UT1-UTC(sec), POLX, POLY(milliasec) picked up from the USNO fileCLCOR2:     1 2453743.5      0.333946    49.430   382.040CLCOR2:     2 2453744.5      0.333483    49.400   381.820CLCOR2:     3 2453745.5      0.333279    49.540   381.660CLCOR2:     4 2453746.5      0.333297    49.160   381.580CLCOR2:     5 2453747.5      0.333489    48.660   381.270CLCOR2:   8684 Records modifiedCLCOR2: Appears to have ended successfullyCLCOR2: localhos 31DEC19 TST: Cpu=      1.0  Real=      1  IO=        13
But the resulting CL table has NAN values in it. Subsequent tasks applying this CL table to data find zero unflagged points, and hence data analysis cannot continue. I also tested installing 31DEC18 binaries on my Intel Mac, and had the same issues again. I ran the same task with the same data on an older 31DEC16 AIPS version on an Ubuntu Linux machine without any issues, so it should not be related to my input data or parameters.I suspect my issue is related to the issue currently described on http://www.aips.nrao.edu/index.shtml under "Warning about modern Linux operating systems". I suspect perhaps rebuilding the OS X binaries with gcc and gfortran, as appears to work for the LINUX version of this issue, could work. I have tried compiling AIPS from source myself, but without success so far. If possible to provide non-Intel-compiler binaries for OS X, I would be very interested in testing them on my machine. Are there any plans to provide such binaries for the MACINT binary AIPS installation?Best regards,Eskil

------------------------------------------------------
Staff CP:  https://help.nrao.edu/staff
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://listmgr.nrao.edu/pipermail/daip/attachments/20190207/87ded97a/attachment.html>