[daip] Re: FITLD and JIVE data

[Eric Greisen]

Huib Jan van Langevelde writes:

 > As you know, we are using IDI fits to export large datasets from the
 > EVN MkIV dataprocessor at JIVE. This is necessary as they are often too
 > large to fit a single medium. We run into a minor, but annoying problem
 > that I hope you can fix without too much effort.
 > 
 > The AIPS uv-entry header keyword 'Receiver' is filled from the FITS
 > keyword INSTRUME in a UVfits file. There is also a RECEIVER keyword
 > that we also set explictly to JIVE. However, both of these are ignored
 > and the header value is left at the default 'VLBA' for IDI fits, which
 > is politically incorrect around here ;-) I looked at the code and
 > indeed the CATH( KHINS ) seems to be left at its default 'VLBA', never
 > mind what is in the FITS files. Would it be possible to make that
 > follow the FITS keyword RECEIVER or INSTRUME?

     It turns out that most of the main header is ignored even when
parsed.  The header is initialized much later.  I added code to trap
the INSTRUME keyword value and get it into the right place.

 > 
 > A related question is about the sort order. We explicitly set it to TB,
 > as it comes out sorted. This is indeed copied to AIPS, unless you read
 > multiple (time-sequenced) FITS files, than it seems to be set to **.
 > I guess, as you cannot be certain about the order of the FITS files,
 > this is the desired behaviour? There is no way around this? Sorting is
 > quite a big step for these datasets.

The sort order is set to TB in IDI and then changed to ** only when an
out of order sample is found.  MSORT should not be a big deal now that
I have fixed it (very recently).

Eric Greisen