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

[Bob Garwood]

DATE-OBS was standard practice in 1989.  The single dish FITS convention 
could use to be update to modern practice (including modern WCS syntax).

TDIMx indicates the shape of the spectra.  Included as a column to 
accommodate the case where the number of channels varies slightly from 
row to row.  TFORMx indicates the total number of elements in the 
column.  I would advise anyone to never make TFORMx a column - as it's 
use as a keyword is required and the Green Bank convention is unclear 
what is to be done if the same name appears as both a keyword and a column.

The Green Bank convention is really simply the last sentence of that pdf:

> This concept of expanding a keyword into a table column (or conversely,
> collapsing a column of identical values into a single header keyword) 
> is now
> generally known as the Green Bank keyword convention.

All of the rest is merely use case and historical.  The draft changes to 
the standard seem to be both a statement of the general Green Bank 
convention (the final paragraph of the new 8.2.1, which as noted here 
should use some other example besides DATE-OBS) and a specific 
discussion as to how one might make use of the Green Bank convention 
when storing multiple images within a table.  But WCS paper I already 
covers that (and it also mentions the Green Bank convention) so I would 
think that any changes to the FITS standard related to the image 
coordinate axes as done in 8.2.1 would be more correct to talk about WCS 
paper I (I haven't looked through the standard in detail in some time, 
so I'm not sure if that's already been done elsewhere) rather than the 
Green Bank convention and if the Green Bank convention is to be included 
by name that it be limited to just the simple statement above perhaps 
with a simple example (what the DATE-OBS value is trying to do, but 
obviously with something more appropriate than DATE-OBS).

Bob


On 06/22/2015 12:03 PM, Peter Weilbacher wrote:
> Dear all,
>
> On Fri, 19 Jun 2015, Lucio Chiappetti wrote:
>
>> The proposed draft text is available at
>> http://sax.iasf-milano.inaf.it/~lucio/FITS/Conventions/greenbank-upd2.pdf
> I have to say I find way of storing images pretty confusing. What's the
> advantage of storing them this way instead of creating separate FITS
> extensions? Is it just of historical interest? How do I know which of
> the table columns contains that actual data of the image? (Ah, I see
> that Bob has answered these questions in the meantime.)
>
> The text as suggested for the standard also promotes use of a dash in
> TTYPEn keywords ("DATE-OBS"), which according to 7.3.2 is "not
> recommended". Additionally, it mentions "FITS binary table" at the start
> of the added new section 8.2.1 but in the last paragraph applies it to
> both ascii and binary tables.
> n
>> Supporting material is provided in the FITS Convention Registry at the
>> http://fits.gsfc.nasa.gov/registry/greenbank.html
> I don't think the material there helps the case. That old (?) PDF
> spec says that TDIMn would be use for the table arrays but the example
> FILE shown actually uses TFORMn. I also put the demo FITS file there
> through fitsverify which says:n
> [...]
> =================== HDU 2: BINARY Table ====================
> *** Warning: Column #3: Name "DATE-OBS" contains character '-' other than
>               letters, digits, and "_".
> *** Warning: Keyword #13, CTYPE4: index 4 is not in range 1-2 (NAXIS).
> [...]
> =================== HDU 3: BINARY Table ====================
> *** Warning: The HDU 3 and 2 have identical type/name/version
> *** Warning: Column #3: Name "DATE-OBS" contains character '-' other than
>               letters, digits, and "_".
> *** Warning: Keyword #13, CTYPE4: index 4 is not in range 1-2 (NAXIS).
>
> To me this says that this convention is mostly of historical interest
> and should not be part of the main FITS standard.
>
>> The Public Comment Period starts today 19 June 2015 and will last formally
>> for 3 weeks until July 9 included.
> As others have noticed, this is far from ideal.
> Looking at specifications for a FITS convention requires a different
> level of attention than now for inclusion in the standard.
>
> Best regards,
>     Peter.