[daip] Re: new problem with AIPS task

[Eric Greisen]

computer support writes:
 > >Date: Tue, 30 Oct 2001 08:24:13 -0700
 > >From: Eric Greisen <egreisen at cv3.cv.nrao.edu>
 > >To: computer support <help at astro.ucla.edu>, daip at primate.aoc.NRAO.EDU
 > >
 > >computer support writes:
 > > > One user now reports that the task PBCOR suddenly no longer works correctly
 > > > (although I can't think of anything that has changed in our AIPS installation
 > > > that would affect that).  Basically, it says that it completed successfully,
 > > > but the data it writes are identical to the input, no matter what parameters
 > > > are chosen.  This has been seen on different user areas by different users on
 > > > different data sets.
 > > > 
 > > > Do you have any idea what might be causing that?  (Another subtle compilation
 > > > optimization problem maybe?  We did change our compilers since it last worked.)
 > > > The PBCOR patch (<ftp://aips.nrao.edu/pub/aips/31DEC00/patches.html#004>) was
 > > > applied, so it's not that.  Maybe you've seen that happen before...
 > > > 
 > >
 > >It does sound just like the bug the patch was designed to fix.  So are
 > >you sure?
 > 
 > Absolutely.
 > 
 > >What compiler are you using - and remind me what aips version and architecture?
 > 
 > 31DEC00 under SunOS 5.6 (AIPS platform "SUL").  I recently upgraded 
 > our compilers to: Sun WorkShop 6 update 2 FORTRAN 77 5.3 2001/05/15.
 > 
 > >Eric
 > 
 > In general, is there a need to recompile when the system libraries 
 > get changed?  (That's the only difference I can think of.)  Thanks.

Your message says "suddenly".  Programs do not change what they do
suddenly - they change when something changes, usually when they
themselves are recompiled.  I would not expect that a change of the
run time libraries would cause PBCOR to appear to run but give no
correction and not affect all other tasks.  Either such a change would
be more or less disastrous or it would be benign.

That leaves just 2 theories:

1. The user is wrong.  Either he/she ran the task wrongly or that the
   image needed no primary beam correction or that a correction was
   applied but it was not very visible.  The way to test the latter is
   to run COMB with OPCODE 'SUM and APARM=1,-1 to difference the input
   and output to PBCOR.

2. You are wrong about the patch in some way - the user is not using
   the patched version or ??  Does this user claim it worked after you
   patched it but not now even though it has not been compiled in the
   interim?

Eric Greisen