Hi Francois
What I was proposing was to use the BNF I was describing in the email
before, which I believe is consistent with what Alex was suggesting,
plus the specific syntax of the REGION following the specification of
the footprint service. Therefore my suggestion would be like this:
<predicate> ::= <comparison_predicate>| <between_predicate> | <in_predicate> | <like_predicate> | <null_predicate> | <exists_predicate> | <region_predicate>
<region_predicate> ::=
REGION<period><region_verbs>
<left_paren>
<region_specification>
[ <comma> <column_reference> [ { <comma> <column_reference> }... ] ]
<right_paren>
<region_verbs> ::= INTERSECTS | CONTAINS ...
<region_specification>
::= !! 'Footprint Service'-like syntax, e.g. RECT LATLON 37 -109.55 41 -102.05
Francois Ochsenbein wrote:
> Hi,
>
> Sorry, but the Region construct on the quoted page is a formal
> definition on how a REGION is written -- it clarifies the vocabulary
> and the syntax, but it is not defined as a function neither it specifies
> how it's written and combined with other predicates in a query.
> Or did I overlook ?
>
> For circular regions, Alex's examples use a function
> fDistanceLonLat(lon1, lat1, lon2, lat2)
> which returns the spherical distance between 2 points which
> can be table columns or actual numbers. If the 2 sets of positions
> are in the same frame, there is no need to specify the actual frame --
> just the four numbers are necessary and sufficient for this purpose.
> This function is therefore the only thing which is required to define /
> compute / select on the basis of any combination of circular regions.
> Making the selection efficient is completely different question.
>
> See you soon for lively discussions....
> --francois
>
>
>> Hi there,
>>
>> Alex suggested us to have a look at the NVO Footprint Service to check
>> whether it can be of use within the ADQL. As he is traveling today and
>> might not be able to start the discussion, I will comment it here,
>> pointing to the relevant pages.
>> In our opinion all that we need to specify a region, not the full
>> predicate construct, is already there. From the main pages of the
>> Footprint Service (announced last November,
>> http://www.voservices.net/footprint/), you can access a link to a very
>> simple and useful grammar definition that we could use:
>> http://www.pha.jhu.edu/~gyuri/htm/HtmPrimer/regiongrammar.html
>> This looks to us as a clean and reasonable approach. So please disregard
>> my previous suggestion of contacting Arnold Rots for clarification on an
>> eventual STC-like string serialization. Fortunately, everything is
>> already there :-)
>> Please tell us what you think.
>> Inaki
>>
>>
>
> ================================================================================
> Francois Ochsenbein ------ Observatoire Astronomique de Strasbourg
> 11, rue de l'Universite F-67000 STRASBOURG Phone: +33-(0)390 24 24 29
> Email: francois-at-astro.u-strasbg.fr (France) Fax: +33-(0)390 24 24 17
> ================================================================================
>