[fitsbits] CLASS to BINTABLE

[Kuiper, Thomas B]

Greetings from Mauna Kea.

I've used the version of CLASS installed on the CSO VAX to convert CLASS data files to FITS BINTABLE format. I was led to believe that this conforms to the SDFITS format, but the files generated do not.  The command I used is

LAS\FITS   WRITE  newfile.fits  /MODE  INDEX  /BITS  -32

In the help for this command it says: 
"The USER MUST make sure that all index spectra are consistent (same source, same line, same frequency setup, ...)"  
Can someone explain this?  The SDFITS standard clearly allows all these variables to be in table columns.  One should be able to select the rows one needs, as the ASAP selector does.

I found that for a session in which I used two back ends with different numbers of channels, the FITS writer did put MAXIS1 in a column.  Alas, ASAP chokes on that old convention MAXIS convention.   It wants TDIM.

The file was also pretty confused about, for example, channel frequency spacing, which was in the main extension header and could only apply to one backend.

After generating a FITS file for one backend, there's no contradiction but there is not enough information that I can find to figure out, for example, the Vlsr of the source.  The data for the first axis, CRVAL1 etc. are all in the extension header but the RESTFREQ and IMAGFREQ are in the table.  Since CRVAL1 was 0.0 instead of the OBSFREQ, it's not possible to use the axis1 frequency and the rest frequency to calculate Doppler velocity.

This message is probably as incoherent as the file I generated seems to be.  Does anyone have any suggestions?

Best regards

Tom

p.s. I've got a little Python script in which I'm trying to take the FITS file from CSO and make ASAP happy with it.  For anyone who is interested it documents pretty well the discrepancies between the file generated with the above command and a file ASAP accepts.

Tom