[daip] SPLAT bug in AVGTIM
Olaf Wucknitz
wucknitz at astro.uni-bonn.de
Sat Nov 3 14:56:48 EDT 2007
Hello,
there seems to be a small but significant bug in the routine AVGTIM in
SPLAT.FOR.
The following line copies the time-averaged visibilities to the output
buffer immediately before UVDISK ('WRIT',...) is called:
CALL RCOPY (3*NCORR, BUFF1, BUFF2(BIND+NRPRIN))
This is line 2317 in the latest 31DEC07 version.
As far as I understand it, the line should read:
CALL RCOPY (NCORR, BUFF1, BUFF2(BIND+NRPRIN))
NCORR already counts all three components of the complex data. For
compressed data, this branch of the code is not active.
In my case the incorrect 3*NCORR had the effect that once we were close to
the end of the buffer, the RCOPY wrote over the allocated edge of BUFF2,
which then corrupted the RBUFF array. Because the search for repeating
baselines (within one SOLINT) then failed, error messages of the following
kind were produced:
UPC201> SPLAT1: AVGTIM: TOO MANY BASELINES FOR BUFFER AT 9 - 11
UPC201> SPLAT1: SPLCOP: ERROR 1 WRITING OUTPUT FILE
In some cases (when not all baselines were present or with APARM(5)=1), no
errors were produced, but the output file had corrupted baselines 0-0 in
it, which caused various problems at later stages.
Cheers,
Olaf
--
Olaf Wucknitz
Argelander-Institute for Astronomy wucknitz at astro.uni-bonn.de
University of Bonn, Germany http://www.astro.uni-bonn.de/~wucknitz
More information about the Daip
mailing list