[daip] AVSPC

[Eric Greisen]

Crystal Brogan wrote:
> Hi,
> 
> A CASA user recently reported difficulty importing a UV-fits file created 
> by AIPS FITTP in CASA -- it would die with an error message saying that 
> the number of channels is zero. The problem has been traced back to the 
> use of AVSPC in AIPS, which had been used to collapse a pseudo-continuum 
> dataset to a single channel. The file created by AVSPC purported to have 
> "0" channels instead of "1" in the "pixels" column of the spectral axis. 
> Changing the number of pixels to "1" before exporting with FITTP fixed the 
> problem.
> 
> Thought you might want to know about this anomalous behavior of AVSPC.
>

I took a look at AVSPC and indeed, if a users asked for averaging mode 
SUBS with CHANNEL set larger than the number of input channels, AVSPC 
would happily write a data set out with no channels.  But, this data set 
would consist of random parameters only, no data.  Furthermore, there 
would be no way for PUTHEAD to change the number of points on the freq 
axis.  PUTHEAD is blocked from that since it affects the I/O, file size 
etc and so is not simply a header change.

I have blocked this error hole in the task even though I do not 
understand many of the aspects of your e-mail.

Thanks

Eric Greisen