Re: String representations of numeric values

From: Thomas McGlynn <tam-at-lheapop.gsfc.nasa.gov>
Date: Mon, 24 Apr 2006 12:37:57 -0400


Hi Doug,

I certainly thought that is what we intended but I don't think that's what the document says.

The SIA document is quite clear that all inputs are in decimal degrees. However it doesn't say a thing about outputs. All it does is to specify that the type of the column with the POS_EQ_RA/DEC_MAIN UCDs shall be of type double. I believed this meant that the type was a scalar, but the discussion that have come up on sexagesimal coordinates and in particular the USNO Cone search services, indicates that others don't in practice agree even with that.

I don't think there is any support in the standard for the statement that the units of the coordinate centers are degrees. Only the units of the Image_Scales are specified (of the required fields).

The language for the coordinate columns in the Cone search is almost the same as for the SIA.

   Cone search spec:

     Exactly one FIELD must have ucd="POS_EQ_RA_MAIN", with type double,
     representing the J2000 right-ascension of the source.

   SIA Spec:
     Exactly one field MUST have ucd="POS_EQ_RA_MAIN", with datatype="double",
     representing the ICRS right-ascension of the center of the image.

I would suggest that in both standards we need language more like:

    The table must contain one and only one field with the UCD "POS_EQ_RA_MAIN". This field     must be a scalar, with no arraysize specified, and with datatype="double" and unit="degrees".     The value of this field shall be the ICRS right ascension of the center of the     image expressed in decimal degrees.

        Tom

P.S., <Nit>Is 'right ascension' hyphenated? I didn't think so but it is a number of times in the SIA document even when used as a noun. Or were there so many uses an an adjective that the hyphen became engrained in one's fingers.</Nit>

Doug Tody wrote:
> Hi All -
>
> Our intention with SIA was to restrict coordinates to decimal degrees -
> this is specified in a number of places, although perhaps not clearly
> enough.
>
> If there is an explicit data model, as for SIA/SSA etc., then the
> data model should specify the allowable units and representations.
> This should agree with but supercede whatever the container (VOTable,
> FITS, etc.) defines.
>
> Cone search is a slightly different case as there is no catalog data
> model, and data tends to carry over directly from whatever catalog
> is being queried. Nonetheless it would be desirable for cone search
> to require that coordinates be converted to some well defined units
> and representation within the VOTable. Of course, astronomers may
> still want to see sexagesimal notation for display, but that can be
> done by allowing the service to define a suggested format for display
> To support this might require a small addition to the VOTable spec.
> (Anita - I would claim this satisfies the requirement you mentioned).
>
> - Doug
>
>
>
> On Fri, 21 Apr 2006, Thomas McGlynn wrote:
>

>> Mark Taylor wrote:
>>
>>> On Thu, 20 Apr 2006, Thomas McGlynn wrote:
>>>
>>>
>>>> At it's clear that the USNO table is incorrect in the new version
>>>> (since it specifies that the column is a double).
>>>
>>>
>>>
>>> I've just retrieved a table from the USNO-B conesearch using the URL:
>>> http://www.nofs.navy.mil/cgi-bin/vo_cone.cgi?CAT=USNO-B1&RA=50.1&DEC=-37.5&SR=0.16 
>>>
>>> It has RA and DEC FIELD components with the attributes
>>>
>>>     datatype="double" arraysize="3x1"
>>>
>>> which, while you might argue it's eccentric, is not incorrect, this is
>>> exactly how you specify an array-valued column according to both 
>>> versions 1.0 and 1.1 of the standard.
>>>
>>
>> Well I guess my reading of the Cone Search spec was that the type was
>> a double, but I suppose that doesn't explicitly preclude it being an
>> array -- though I think it should be forbidden and that that was
>> the original intent.
>>
>> The SIA protocol also doesn't explicitly forbid arrays but given this
>> discussion I think it should.
>>
>> The language is squishy here....
>>  "Exactly one field MUST have ucd="POS_EQ_RA_MAIN", with 
>> datatype="double", representing the ICRS right-ascension of the center 
>> of the image."
>>
>> So by a liberal interpration if I had a position with RA=123.456, any 
>> of the following
>> are valid responses (in addition to the sexagesimal issues).
>>
>>      123.456       (Yea!)
>>      1 2 3. 4 5 6
>>      6 5 4. 3 2 1 (no reason to make it easy for the reader!).
>>      1  2 3 -1 4  5  6
>>      100 23 0.45 .006 .000 1 -1  (and any other numbers that sum to 0)
>>
>> They all "represent" the ICRS coordinate in the sense that I can 
>> easily define the
>> mapping between these values and the coordinate, but if the standard 
>> is to be useful
>> it must be more restrictive and it had never crossed my mind that 
>> providing coordinates
>> in something other than a scalar giving decimal degrees was legal...
>>
>> Sigh...  There are more things in VOTable and SIAP then are dreamt of 
>> in my philosophy.
>>
>>     Tom
>>

>
>
Received on 2006-04-24Z18:38:53