[bananas] Patches for 15APR99

Patrick P. Murphy pmurphy at nrao.edu
Wed Jul 28 10:03:50 EDT 1999


Here are the current patches for 15APR99 AIPS.

   Task     Description of problem                             Date
   -------------------------------------------------------------------
   TVCPS    (and others); file seq# not stored/retrieved... 1999-05-14
        						   (1999-06-18)
   APGPS    (and TECOR) delay calibration modes             1999-05-14
   CLIP     (and RESEQ, DTCHK) compressed data problems     1999-05-25
   REGRD    crashed under Solaris, others                   1999-07-06
   UVCOP    errors selecting channels                       1999-07-21
   IMAGR    in OVERLAP 2 mode selects wrong fields          1999-07-22
   FITAB    followed by FITTP makes unreadable files        1999-07-27

Complete information on these, including the modified source code and
instructions for rebuilding the affected tasks, can be obtained via:

	 http://www.cv.nrao.edu/aips/15APR99/patches.html

Descriptions of the individual problems for the patches are shown below,
without the details of files and steps to fix the problems.  Refer to the
web page for complete instructions. 

NOTE!  The TVCPS patch was updated on June 18; the patch on our servers
       during the period May 14 through June 18 was INCORRECT and if you
       used it, please re-apply the correct patch.  Sorry...

  ------------------------------------------------------------------------

Task:         TVCPS and other tasks that use the image catalogue

Summary:      The file sequence number is not stored in nor retrieved
              from the image catalog on little-endian (Intel byte-
              order) computers. This causes TVCPS (and perhaps other
              tasks and verbs that use the image catalogue) to fail.
              You do not need to apply this patch on big-endian
              computers. 
                If you are not sure whether you should apply this
              patch find out which version of ZDCHI2 is used and check
              the value assigned to BYTFLP: if this is zero then you
              do not need this patch.

              The file was incorrect in the period 1999-05-14 to
              1999-06-18.

Documentor:   Chris Flatters    Eric Greisen
Date:         1999-05-14        1999-06-18

  ------------------------------------------------------------------------

Task:         TECOR and APGPS.

Summary:      TECOR and APGPS store the dispersive delay with the wrong
              sign.

Documentor:   Chris Flatters
Date:         1999-05-14

  ------------------------------------------------------------------------

Task:         CLIP, DTCHK, RESEQ

Summary:      CLIP assumed that the Stokes axis was the second axis
              with compressed data and hence could get the addressing
              of the UV data wrong, e.g. for the VLBA Correlator
              output.  DTCHK did not fix the address pointers to
              address VIS(3,*) and so worked only for compressed data.
              RESEQ did not multiply the pointers by 3 for expanded
              (previously compressed) data and so failed on compressed
              data.

Documentor:   Eric W. Greisen
Date:         1999-05-25

  ------------------------------------------------------------------------

Task:         REGRD

Summary:      REGRD would fail unpredictably on Solaris (and
              possibly other platforms too) due to a buffer being 2x too
              small.  

Documentor:   Patrick P. Murphy
Date:	      1999-07-06

  ------------------------------------------------------------------------

Task:         UVCOP

Summary:      UVCOP used the increments for frequency and IF suitable
              to the output data set to compute a channel-selection
              mask into the input data set.  If the frequency axis
              precedes the IF axis, data selection errors occured if
              both BIF > 1 and BCHAN/ECHAN selected fewer than all of
              the spectral channels.  The error appeared in 15OCT98 on
              August 24. 

Documentor:   Eric W. Greisen
Date:	      1999-07-21

  ------------------------------------------------------------------------

Task:         IMAGR

Summary:      The selection of fields in OVERLAP 2 mode was made
              improper by incorrect maxima information.

Documentor:   Eric W. Greisen
Date:         1999-07-22

  ------------------------------------------------------------------------

Task:         FITAB followed by FITTP

Summary:      Until this patch, UV data sets written out with FITAB,
              read back in with UVLOD or FITLD>, processed, and then
              written out with FITTP could contain AIPS history
              records which will define the data set falsely to the
              standard reading tasks in AIPS.  The data-writing tasks
              have been changed to avoid writing out pre-existing AIPS
              history and some of the data reading tasks have been
              modified to allow the error to be circumvented in data 
              sets written since FITAB was released.

Documentor:   Eric W. Greisen
Date:         1999-07-27

  ------------------------------------------------------------------------

-- 
  \|/  ____  \|/   Pat Murphy; email http://orangutan.cv.nrao.edu/kippure.html
   @~ / oO \ ~@                  Division Head, NRAO/Charlottesville Computing
   /_( \__/ )_\       Plan, PGP, Pages: http://orangutan.cv.nrao.edu/plan.html
      \_U__/                               "The truth points to itself" - Kosh




More information about the Bananas mailing list