[fitsbits] reopening of Public Comment Period on the Green Bank convention

[William Pence]

Actually, NAXIS2 = 0 in the primary array of the UV visibility data files, so there is no primary data array. The real data is in the following binary table.  The weird value for NAXIS1 was apparently deliberately chosen to serve as a signal to the processing software that this file contains visibility data.  

> On Jun 2, 2016, at 9:45 AM, Arnold Rots <arots at cfa.harvard.edu> wrote:
> 
> Let me make sure I get this right:
> I suspect you mean that the second axis is one pixel long.
> If it were zero pixels long, there would be no pixels in the image at all.
> I note that this mechanism (single-pixel axes) has been used extensively
> in radio astronomy to specify the polarization and frequency of 2-D images.
> 
> Cheers,
> 
>   - Arnold
> 
> -------------------------------------------------------------------------------------------------------------
> Arnold H. Rots                                          Chandra X-ray Science Center
> Smithsonian Astrophysical Observatory                   tel:  +1 617 496 7701
> 60 Garden Street, MS 67                                      fax:  +1 617 495 7356
> Cambridge, MA 02138                                         arots at cfa.harvard.edu
> USA                                                   http://hea-www.harvard.edu/~arots/
> --------------------------------------------------------------------------------------------------------------
> 
> 
>> On Thu, Jun 2, 2016 at 1:04 AM, William Pence <William.Pence at nasa.gov> wrote:
>>> On 6/1/2016 11:14 AM, Demitri Muna wrote:
>>> 
>>> There is a significant difference between documenting use of the FITS format and incorporating it into the standard. As an example, I recently came across UV data, which immediately crashed my FITS viewer, and then this in the AIPS File Format Memo:
>>>> In the UV-tables form, the visibility data are written as a FITS binary table, normally placed after the other table extensions. The primary HDU has an AIPS conventional form meant primarily to be so odd as to act as a reliable identifier. The primary HDU asserts that the primary data has two axes, the first of which has 777777701 values while the second has zero values. This is sufficient to tell all FITS readers that the primary data set is not a random groups data set and otherwise contains no data. 
>>>> 
>>> 
>>> This may be convention, but it's *lying*. I'm sure that there are untold numbers of FITS files of UV data that use this convention. Should it be part of the standard? Absolutely not.
>> It is legal for a FITS image to have one or more zero length axes as well as other non-zero length axes, so these UV visibility data do conform to the requirements of the FITS standard and are not lying.  One could say that these 2D images really are 777777701 pixels wide, but because they are 0 pixel high they are hard to see.  :-)
>>   
>> -Bill
>> 
>> 
>> 
>> _______________________________________________
>> fitsbits mailing list
>> fitsbits at listmgr.nrao.edu
>> https://listmgr.nrao.edu/mailman/listinfo/fitsbits
> 
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://listmgr.nrao.edu/pipermail/fitsbits/attachments/20160602/4e434df6/attachment-0001.html>