[daip] AIPS oddity (fwd)

[Eric Greisen]

Shoshana Rosenthal wrote:
> Hi Eric
> 
> Just got the following message from Mark Reid. rglinux9 is a 64bit
> computer and LNX64 is installed on it.
> 
> Could you help please.
> 
> Shoshana
> 
> 
> ---------- Forwarded message ----------
> Date: Mon, 15 Jun 2009 12:28:05 -0400
> From: Mark Reid <reid at cfa.harvard.edu>
> To: Shoshana Rosenthal <shoshana at cfa.harvard.edu>
> Subject: AIPS oddity
> 
> Shoshana,
> 
>     I'm running AIPS on RGLINUX9 and hit a snag running task SNFLG (see
> below).    This task usually doesn't fail.  Any ideas what is happening?
> 
> (FYI, I've been analyzing a similar data set on RGLINUX7 and had no
> problem with this step.)
> 
> 
>  >version 'tst'
>  >inp
> AIPS 1: SNFLG:  writes flagging suggestions based on phases in SN files
> AIPS 1: Adverbs     Values                 Comments
> AIPS 1: ----------------------------------------------------------------
> AIPS 1: INNAME     '2006OCT09-C'           UV data (name).
> AIPS 1: INCLASS    'B2NCON'                UV data (class).
> AIPS 1: INSEQ         2                    UV data (seq. #).  0 => high
> AIPS 1: INDISK        2                    Disk unit #.  0 => any
> AIPS 1: INEXT      'SN'                    Input'SN' or 'CL' table
> AIPS 1: INVERS        4                    Input table file version no.
> AIPS 1: SOURCES    *all ' '                Source list
> AIPS 1: QUAL         -1                    Source qualifier -1=>all
> AIPS 1: TIMERANG   *all 0                  Time range to consider
> AIPS 1: STOKES     ' '                     Stokes type to consider
> AIPS 1: SELBAND      -1                    Bandwidth to include (kHz)
> AIPS 1: SELFREQ      -1                    Frequency to include (MHz)
> AIPS 1: FREQID       -1                    Freq. ID to include, -1=>all
> AIPS 1: BIF           0                    First IF to consider, 0=>1.
> AIPS 1: EIF           0                    Last IF to consider
> AIPS 1: DOSTOKES      1                    > 0 -> flag all Stokes if
> AIPS 1:                                    one is bad
> AIPS 1: DOIFS         1                    > 0 -> flag all IFs if one
> AIPS 1:                                    is bad
> AIPS 1: SUBARRAY      0                    Subarray to consider: 0 ->1
> AIPS 1: ANTENNAS   *all 0                  Antennas to consider 0=>all
> AIPS 1: CUTOFF        0                    Table weight below which data
> AIPS 1:                                    will not be considered, 0 =>
> AIPS 1:                                    consider all.
> AIPS 1: FLAGVER       0                    FG table to be copied and
> AIPS 1:                                    updated.  -1 -> none.
> AIPS 1: DPARM        60          10        (1) Max.allowed phs jump(deg)
> AIPS 1:            *rest 0                             over
> AIPS 1:                                    (2) time interval (sec).
> AIPS 1:                                    Usually a bit < than solint.
>  >
>  >go
> SNFLG1: Task SNFLG  (release of 31DEC09) begins
> SNFLG1: ZABORS: signal 11 received
> SNFLG1: ABORT!
> AIPS 1: Resumes
> 
> 
>  >version 'new'
>  >go
> AIPS 1: Found in Version=NEW
> SNFLG1: error while loading shared libraries: libsvml.so: wrong ELF
> class: ELFCLASS64
> 
> 
The MNJ tonight should fix this.  The abort was due to an array being 
exceeded by 1 word - a word that was innocent in the old 32-bit load 
modules but the new 64-bit load module had an important parameter there.

The failure on NEW is due to the installation being of 32-bit load 
modules available to this 64-bit machine.

Eric Greisen