DAL sessions in Victoria

[Doug Tody]

Ultimately we will need ADQL for fully-expressive general queries, however
soemthing like a set of time intervals is more readily handled by a
specialized range-list syntax (i.e., a parameter) than by a parsed
language.  This is needed for both the spectral and time axes and is
essentially the same problem for both, although we probably want to
maximize compatibility with ISO 8601 for the time axis.  The problem
with the spatial axis is similar but in the general case requires
that a set of regions be specified, e.g., using STC regions probably
specified via a document uploaded by the client.  Software to query
based on such a region specification is already being developed.
Query by time regions is something which has been done for years for
high energy data, which consists of time-tagged events.   - Doug


On Wed, 26 Apr 2006, Guy Rixon wrote:

> On Wed, 26 Apr 2006, Rob Seaman wrote:
>
>> ...
>> The complications for temporal searching are several.  First, you
>> need to be able to search in a barycentric (or arbitrary) time
>> system.  Second, you need to be able to search through some sort of
>> evenly gridded comb filter in cadence - something like "return all
>> images/spectra whose mid-exposure occurred near an even quarter-
>> hour".  Third, you have to be able to specify "near" - "within a
>> window of plus or minus one minute".  Fourth, you may need to be
>> sensitive to some very creative observing modes and data structures.
>> SONG also experimented with a multiple ramped exposure mode for
>> acquiring a time sequence of spectra in a single exposure.  Makes me
>> wonder further about searching for off-band data from dual exposure
>> nod-and-shuffle observations from instruments such as MARS on Kitt
>> Peak.  The second field is shifted both spatially and temporally
>> (interleaved as with any chopping or beam switching mode).
>
> Thinks me, that sort of think needs a full query language on the scale of
> ADQL, not a little language like those inherent in the S**Ps. It's not the
> sort of thing that you're going to express in an HTTP-get URI.
>
> Unless...
>
>> ISO 8601 provides (as a central design feature) a simple way to
>> specify cadences on a wide range of timescales, i.e., so many seconds
>> or tens of seconds, minutes, hours, days, (even weeks, I believe),
>> months, years, centuries, etc.
>
> Does this imply that a single ISO8601 argument can handle the selection for
> your use-case above?
>
> Guy Rixon 				        gtr at ast.cam.ac.uk
> Institute of Astronomy   	                Tel: +44-1223-337542
> Madingley Road, Cambridge, UK, CB3 0HA		Fax: +44-1223-337523
>