[daip] FITLD aborts on large FITS-IDI file

[Eric Greisen]

Walter Brisken wrote:
> At the VLBA test meeting yesteday I mentioned a failure mode of FITLD when 
> attempting to load large FITS-IDI files.  The size threshhold for failure 
> may depend on the value of DOUVCOMP.
> 
> for DOUVCOMP=0, a FITS-IDI file of size 2894800320 was successfully 
> loaded, but one of size 3649544640 fails.  For DOUVCOMP=1, no failure was 
> seen on this file, the largest one I have handy.
> 
> The large input file that failed for DOUVCOMP=0 is: 
> /home/swc/difx/projects/bp147/BP147.0.FITS
> 
> There are other large FITS-IDI files of varying size in the same 
> directory.
> 
> The failure results in the following message:
> 
> FITLD1: Create COMP0-0     .UVDATA.   1 (UV)  on disk  1  cno    3
> FITLD1: Sky frequencies (MHz) before re-ordering occurs:
> FITLD1: Incoming FREQID #   1
> FITLD1: 22230.94
> FITLD1: CL table interval set at  1.00 minutes
> FITLD1:            !!!!!!!!!!!
> FITLD1: Digital corrections being performed
> FITLD1: Total power corrected even if no zero-padding
> FITLD1:            !!!!!!!!!!!
> FITLD1: Current file has   441517 visibilities
> FITLD1: File ref. freq:    22230.94 MHz
> FITLD1: ZEXPND: INVALID RECORD REQUEST = ******
> FITLD1: FILE EXPANSION FAILED - QUITTING
> FITLD1: ZMI2: REQUEST FOR BYTES 1 THRU = 394240 BEYOND EOF = 364544
> FITLD1: ZMIO: OPER=WRIT LUN=16 BLKNO=           1 8-BIT-BYTES=394240
> FITLD1: ZERROR: IN ZMI2 ERRNO = 22 (Invalid argument)
> FITLD1: ZWAIT: ZWAI2 RETURNS ERROR 3 FOR LUN = 16 BUFFER = 1
> FITLD1: ZERROR: IN ZWAI2  ERRNO = 22 (Invalid argument)
> FITLD1: GET3DT: ERROR   3 FLUSHING UV DATA FILE
> FITLD1: UT2UV: ERROR   3 TRANSLATING UV BINARY TABLE
> FITLD1: Image=MULTI     (UV)         Filename=COMP0-0     .UVDATA.   1
> FITLD1: Telescope=VLBA               Receiver=VLBA
> FITLD1: Observer=BP147               User #= 2179
> FITLD1: Observ. date=22-JUN-2008     Map date=18-JUL-2008
> FITLD1: # visibilities     10000     Sort order  TB
> FITLD1: Rand axes: UU-L-SIN  VV-L-SIN  WW-L-SIN  TIME1  BASELINE
> FITLD1:            SOURCE  FREQSEL  INTTIM  GATEID  CORR-ID
> FITLD1: ----------------------------------------------------------------
> FITLD1: Type    Pixels   Coord value     at Pixel     Coord incr   Rotat
> FITLD1: COMPLEX      3   1.0000000E+00       1.00  1.0000000E+00    0.00
> FITLD1: STOKES       2  -1.0000000E+00       1.00 -1.0000000E+00    0.00
> FITLD1: FREQ       512   2.2230940E+10       1.00  1.5625000E+04    0.00
> FITLD1: IF           1   1.0000000E+00       1.00  1.0000000E+00    0.00
> FITLD1: RA           1    00 00 00.000       1.00       0.000000    0.00
> FITLD1: DEC          1    00 00 00.000       1.00       0.000000    0.00
> FITLD1: ----------------------------------------------------------------
> 
> Please let me know if more information is needed.

The error messages alone were enough.  More Phil code contained a 
computation of how much expansion was needed in AIPS bytes (which are 2 
bytes in length).  I made the computation go in double precision before 
conversion to suitable (kilo real byte) units.

Will be fixed with next MNJ

Eric Greisen