From: <r.dodson_at_email.protected>
Date: Thu, 8 Jun 2006 17:03:04 +0200

All

>I would love to be able to attach Tsys and gain tables to the data. It
>would be easy with the software correlator. If we had a robust scheme to
>produce good calibration files in some format, then we could work on
>including them with the correlated output. This will not be a part of our
>contracted obligations, however.
>
>In fact, I would make this suggestion. If someone can determine a scheme
>to automatically collate the Tsys files in their native formats, from each
>antenna, on the ATNF ftp site, the functionallity of logs2keyin could be
>incorporated into a pre-correlation step at Swinburne. This would remove
>the problems of multiple versions of logs2keyin floating around and would
>consolidate the process in one place.

Seeing I am looking at LBA data at the moment I will make some
comments on logs2keyin.pl

We all agree it is a bit of a dogs breakfast, with code drawn in from
me, Steve and
Jim, and maybe others, into a big loop which works -- but is not pretty.

I kept it working while I was at ISAS, because we (I) needed an
pipeline as there were so many datasets to reduce. I wrote another
script that would collect everything from everywhere. With this we
would/could collect any LBA/EVN tsys/logfiles, generate the required
antabs, read the tape - or disk - datafiles, load, sort, index,
amplitude correct on date, correct on autocorrelation, and fringe fit.
I could never get a decent fringe fit solution filter as the VSOP
solutions were just too difficult for this and hand editing was
required.

Therefore it was not and could not be an end-to-end pipeline. Also
this was for post-correlation data, not pre-correlation.

It is of course a huge improvement on nothing.

It worked on LBA data, as that is just a subset of the APT/VSOP
network. Now it will not work as there is a free for all on where the
files are and the format of these. This would be a `minor' problem to
overcome.

Therefore the scriipt would fufill most of the requirements that Steve
lays out. The work would be small to make it work

 However I have doubts.

 I agree it needs to be maintained in Australia.
 I don't think that logs2keyin should be carried forwards, but replaced

      Richard

PS the Mopra format has changed, which is why the script complains
bitterly and puts zeros as the last column. This is not the same error
as the misformating of the date, but extra columns of data .. when?
why? shall I extend logs2keyin to include these changes? It is more
like the tied array now.

New
! UT time Az El Ta1 Tb1 Ta2 Tb2
Old
! UT time Az El Ta Tb Tab

If so can we agree on a default source for the script? I have a
version 1.2.12 -- but 1.2.11 would be fine for general use. 1.2.12 is
to be found on http://www.vsop.isas.jaxa.jp/survey/publications/logs2keyin.1.2.12.pl
but a better home is needed.

The version I inherited I numbered 1.0. The development was as so:

# j?l Version 1.0 From Jim Lovell in Nov
# rgd Version 1.1 Added pcfs version & made it y2k matching
# rgd Version 1.2 Added Parkes & "Getopt::Long" (which is only in perl 5)
# 1.2.1 add the decimate option (for mopra mainly)
# 1.2.2 Change the parkes Tsys extraction
# 1.2.3 add an pre-historic PCFS version (for geodosy)
# 1.2.4 bug in decimate, tdy not returned in call,
# so not decimated. Not used anymore, floor(td) instead
# Also Put in a no pgplot option
# 1.2.5 Made Mopra varible NIFs
# 1.2.6 Made a get_antab_tsys
# 1.2.7 Made an average
# 1.2.8 Only the first Tsys record is used to set the NIF's
# if tsys_pks_old
# 1.2.9 Swapped the Tid formats around -- I don't have enough
# examples to be sure. Tied Array Log Added
# 1.2.10 small alterations to improve the ofcolumns description
# & YET another MkIII format (HH) (/tsys/[1,2][u,l],)
# & another (NT) (/tsys/[1-e][u,l],)
# 1.2.11 added pipe & nogain option (for use with nominal_tsys)
# 1.2.12 patch for a short term error in the mop logs

     Richard
Received on 2006-06-09 01:03:46