[daip] CVEL

[Eric Greisen]

Lynn D. Matthews writes:
 > I did not get these messages when I first ran CVEL on these data back in
 > January (I even saved a copy of the message file!).  I only began
 > experimenting again last week after seeing the messages appear during
 > processing of a different epoch of these data.
 
In January, the code did not test for missing or other "bad" columns
but just blithely assumed that all was well and in the usual order
(although little forces that order in FITLD IDI files).  I added tests
for curiosities and then fixed routines - CALINI and BPINI were among
the first - to account for overlooked matters like # columns depending
on number polarizations.

 > Yesterday in making a subset of data to send to Amy, I saw similar
 > messages produced by UVCOP.

   The details may matter on this too.  Which subroutines said what?

 > 
 > Another thing I cannot figure out is that if I run on my own Linux machine
 > I get the CALINI errors only, but on the public workstation (after having
 > used FITTP/FITLD to transfer the data) I get both the CALINI and BPINI
 > errors.

    This is telling - CVEL called CALINI with 4 arguments set to the
same variable.  On a READ this causes that variable to change values 4
times and to decide whether to print the messages based on the 4th
value which I bet was not 1 but 2.  That same error is in no other
task and will be gone from CVEL tomorrow.

    I wonder if the public workstation depends on a machine that
compiles its own code and if that machine could have failed to
recompile the subroutines when I fixed them and so be linking now with
the old versions that did not check the number of polarizations.  That
sort of problem has occurred at sites that compile their own code.

 > 
 > Running PRTAB on the CL and BP tables as you suggested, it thinks NO_POL=1
 > and NO_IF=2 (which is correct). There are 11 columns in the BP table, 22
 > in the CL table.  I've placed the full outprints of these tables in ftp
 > along with a sample of data: cfa-ftp.harvard.edu, cd outgoing/lmatthew.

There should be 12 columns in the BP table I think (8 + 4*numpol).
The 22 is right, but the CVEL error probably explains the CALINI
messages on your machine.

Eric Greisen